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         - check_profile: Update remote profile data
  20         */
  21
  22         if (($argc > 2) && ($argv[1] == "dirsearch")) {
  23                 $search = urldecode($argv[2]);
  24                 $mode = 1;
  25         } elseif (($argc == 2) && ($argv[1] == "checkcontact")) {
  26                 $mode = 2;
  27         } elseif (($argc == 2) && ($argv[1] == "suggestions")) {
  28                 $mode = 3;
  29         } elseif (($argc == 3) && ($argv[1] == "server")) {
  30                 $mode = 4;
  31         } elseif (($argc == 2) && ($argv[1] == "update_server")) {
  32                 $mode = 5;
  33         } elseif (($argc == 3) && ($argv[1] == "update_server_directory")) {
  34                 $mode = 6;
  35         } elseif (($argc > 5) && ($argv[1] == "poco_load")) {
  36                 $mode = 7;
  37         } elseif (($argc == 3) && ($argv[1] == "check_profile")) {
  38                 $mode = 8;
  39         } elseif ($argc == 1) {
  40                 $search = "";
  41                 $mode = 0;
  42         } else {
  43                 die("Unknown or missing parameter ".$argv[1]."\n");
  44         }
  45
  46         logger('start '.$search);
  47
  48         if ($mode == 8) {
  49                 $profile_url = base64_decode($argv[2]);
  50                 if ($profile_url != "") {
  51                         poco_last_updated($profile_url, true);
  52                 }
  53         } elseif ($mode == 7) {
  54                 if ($argc == 6) {
  55                         $url = base64_decode($argv[5]);
  56                 } else {
  57                         $url = '';
  58                 }
  59                 poco_load_worker(intval($argv[2]), intval($argv[3]), intval($argv[4]), $url);
  60         } elseif ($mode == 6) {
  61                 poco_discover_single_server(intval($argv[2]));
  62         } elseif ($mode == 5) {
  63                 update_server();
  64         } elseif ($mode == 4) {
  65                 $server_url = base64_decode($argv[2]);
  66                 if ($server_url == "") {
  67                         return;
  68                 }
  69                 $server_url = filter_var($server_url, FILTER_SANITIZE_URL);
  70                 if (substr(normalise_link($server_url), 0, 7) != "http://") {
  71                         return;
  72                 }
  73                 $result = "Checking server ".$server_url." - ";
  74                 $ret = poco_check_server($server_url);
  75                 if ($ret) {
  76                         $result .= "success";
  77                 } else {
  78                         $result .= "failed";
  79                 }
  80                 logger($result, LOGGER_DEBUG);
  81         } elseif ($mode == 3) {
  82                 update_suggestions();
  83         } elseif (($mode == 2) AND get_config('system','poco_completion')) {
  84                 discover_users();
  85         } elseif (($mode == 1) AND ($search != "") and get_config('system','poco_local_search')) {
  86                 discover_directory($search);
  87                 gs_search_user($search);
  88         } elseif (($mode == 0) AND ($search == "") and (get_config('system','poco_discovery') > 0)) {
  89                 // Query Friendica and Hubzilla servers for their users
  90                 poco_discover();
  91
  92                 // Query GNU Social servers for their users ("statistics" addon has to be enabled on the GS server)
  93                 if (!get_config('system','ostatus_disabled'))
  94                         gs_discover();
  95         }
  96
  97         logger('end '.$search);
  98
  99         return;
 100 }
 101
 102 /**
 103  * @brief Updates the first 250 servers
 104  *
 105  */
 106 function update_server() {
 107         $r = q("SELECT `url`, `created`, `last_failure`, `last_contact` FROM `gserver` ORDER BY rand()");
 108
 109         if (!dbm::is_result($r)) {
 110                 return;
 111         }
 112
 113         $updated = 0;
 114
 115         foreach ($r AS $server) {
 116                 if (!poco_do_update($server["created"], "", $server["last_failure"], $server["last_contact"])) {
 117                         continue;
 118                 }
 119                 logger('Update server status for server '.$server["url"], LOGGER_DEBUG);
 120
 121                 proc_run(PRIORITY_LOW, "include/discover_poco.php", "server", base64_encode($server["url"]));
 122
 123                 if (++$updated > 250) {
 124                         return;
 125                 }
 126         }
 127 }
 128
 129 function discover_users() {
 130         logger("Discover users", LOGGER_DEBUG);
 131
 132         $starttime = time();
 133
 134         $users = q("SELECT `url`, `created`, `updated`, `last_failure`, `last_contact`, `server_url`, `network` FROM `gcontact`
 135                         WHERE `last_contact` < UTC_TIMESTAMP - INTERVAL 1 MONTH AND
 136                                 `last_failure` < UTC_TIMESTAMP - INTERVAL 1 MONTH AND
 137                                 `network` IN ('%s', '%s', '%s', '%s', '') ORDER BY rand()",
 138                         dbesc(NETWORK_DFRN), dbesc(NETWORK_DIASPORA),
 139                         dbesc(NETWORK_OSTATUS), dbesc(NETWORK_FEED));
 140
 141         if (!$users) {
 142                 return;
 143         }
 144         $checked = 0;
 145
 146         foreach ($users AS $user) {
 147
 148                 $urlparts = parse_url($user["url"]);
 149                 if (!isset($urlparts["scheme"])) {
 150                         q("UPDATE `gcontact` SET `network` = '%s' WHERE `nurl` = '%s'",
 151                                 dbesc(NETWORK_PHANTOM), dbesc(normalise_link($user["url"])));
 152                         continue;
 153                  }
 154
 155                 if (in_array($urlparts["host"], array("www.facebook.com", "facebook.com", "twitter.com",
 156                                                         "identi.ca", "alpha.app.net"))) {
 157                         $networks = array("www.facebook.com" => NETWORK_FACEBOOK,
 158                                         "facebook.com" => NETWORK_FACEBOOK,
 159                                         "twitter.com" => NETWORK_TWITTER,
 160                                         "identi.ca" => NETWORK_PUMPIO,
 161                                         "alpha.app.net" => NETWORK_APPNET);
 162
 163                         q("UPDATE `gcontact` SET `network` = '%s' WHERE `nurl` = '%s'",
 164                                 dbesc($networks[$urlparts["host"]]), dbesc(normalise_link($user["url"])));
 165                         continue;
 166                 }
 167
 168                 if ($user["server_url"] != "") {
 169                         $server_url = $user["server_url"];
 170                 } else {
 171                         $server_url = poco_detect_server($user["url"]);
 172                 }
 173                 if ((($server_url == "") AND ($user["network"] == NETWORK_FEED)) OR poco_check_server($server_url, $user["network"])) {
 174                         logger('Check profile '.$user["url"]);
 175                         proc_run(PRIORITY_LOW, "include/discover_poco.php", "check_profile", base64_encode($user["url"]));
 176
 177                         if (++$checked > 100) {
 178                                 return;
 179                         }
 180                 } else {
 181                         q("UPDATE `gcontact` SET `last_failure` = '%s' WHERE `nurl` = '%s'",
 182                                 dbesc(datetime_convert()), dbesc(normalise_link($user["url"])));
 183                 }
 184
 185                 // Quit the loop after 3 minutes
 186                 if (time() > ($starttime + 180)) {
 187                         return;
 188                 }
 189         }
 190 }
 191
 192 function discover_directory($search) {
 193
 194         $data = Cache::get("dirsearch:".$search);
 195         if (!is_null($data)) {
 196                 // Only search for the same item every 24 hours
 197                 if (time() < $data + (60 * 60 * 24)) {
 198                         logger("Already searched for ".$search." in the last 24 hours", LOGGER_DEBUG);
 199                         return;
 200                 }
 201         }
 202
 203         $x = fetch_url(get_server()."/lsearch?p=1&n=500&search=".urlencode($search));
 204         $j = json_decode($x);
 205
 206         if (count($j->results)) {
 207                 foreach($j->results as $jj) {
 208                         // Check if the contact already exists
 209                         $exists = q("SELECT `id`, `last_contact`, `last_failure`, `updated` FROM `gcontact` WHERE `nurl` = '%s'", normalise_link($jj->url));
 210                         if ($exists) {
 211                                 logger("Profile ".$jj->url." already exists (".$search.")", LOGGER_DEBUG);
 212
 213                                 if (($exists[0]["last_contact"] < $exists[0]["last_failure"]) AND
 214                                         ($exists[0]["updated"] < $exists[0]["last_failure"])) {
 215                                         continue;
 216                                 }
 217                                 // Update the contact
 218                                 poco_last_updated($jj->url);
 219                                 continue;
 220                         }
 221
 222                         $server_url = poco_detect_server($jj->url);
 223                         if ($server_url != '') {
 224                                 if (!poco_check_server($server_url)) {
 225                                         logger("Friendica server ".$server_url." doesn't answer.", LOGGER_DEBUG);
 226                                         continue;
 227                                 }
 228                                 logger("Friendica server ".$server_url." seems to be okay.", LOGGER_DEBUG);
 229                         }
 230
 231                         $data = probe_url($jj->url);
 232                         if ($data["network"] == NETWORK_DFRN) {
 233                                 logger("Profile ".$jj->url." is reachable (".$search.")", LOGGER_DEBUG);
 234                                 logger("Add profile ".$jj->url." to local directory (".$search.")", LOGGER_DEBUG);
 235                                 poco_check($data["url"], $data["name"], $data["network"], $data["photo"], "", "", "", $jj->tags, $data["addr"], "", 0);
 236                         } else {
 237                                 logger("Profile ".$jj->url." is not responding or no Friendica contact - but network ".$data["network"], LOGGER_DEBUG);
 238                         }
 239                 }
 240         }
 241         Cache::set("dirsearch:".$search, time(), CACHE_DAY);
 242 }
 243
 244 /**
 245  * @brief Search for GNU Social user with gstools.org
 246  *
 247  * @param str $search User name
 248  */
 249 function gs_search_user($search) {
 250
 251         // Currently disabled, since the service isn't available anymore.
 252         // It is not removed since I hope that there will be a successor.
 253         return false;
 254
 255         $a = get_app();
 256
 257         $url = "http://gstools.org/api/users_search/".urlencode($search);
 258
 259         $result = z_fetch_url($url);
 260         if (!$result["success"]) {
 261                 return false;
 262         }
 263         $contacts = json_decode($result["body"]);
 264
 265         if ($contacts->status == 'ERROR') {
 266                 return false;
 267         }
 268         foreach($contacts->data AS $user) {
 269                 $contact = probe_url($user->site_address."/".$user->name);
 270                 if ($contact["network"] != NETWORK_PHANTOM) {
 271                         $contact["about"] = $user->description;
 272                         update_gcontact($contact);
 273                 }
 274         }
 275 }