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                 $server_url = poco_detect_server($user["url"]);
 169                 $force_update = false;
 170
 171                 if ($user["server_url"] != "") {
 172
 173                         $force_update = (normalise_link($user["server_url"]) != normalise_link($server_url));
 174
 175                         $server_url = $user["server_url"];
 176                 }
 177
 178                 if ((($server_url == "") AND ($user["network"] == NETWORK_FEED)) OR $force_update OR poco_check_server($server_url, $user["network"])) {
 179                         logger('Check profile '.$user["url"]);
 180                         proc_run(PRIORITY_LOW, "include/discover_poco.php", "check_profile", base64_encode($user["url"]));
 181
 182                         if (++$checked > 100) {
 183                                 return;
 184                         }
 185                 } else {
 186                         q("UPDATE `gcontact` SET `last_failure` = '%s' WHERE `nurl` = '%s'",
 187                                 dbesc(datetime_convert()), dbesc(normalise_link($user["url"])));
 188                 }
 189
 190                 // Quit the loop after 3 minutes
 191                 if (time() > ($starttime + 180)) {
 192                         return;
 193                 }
 194         }
 195 }
 196
 197 function discover_directory($search) {
 198
 199         $data = Cache::get("dirsearch:".$search);
 200         if (!is_null($data)) {
 201                 // Only search for the same item every 24 hours
 202                 if (time() < $data + (60 * 60 * 24)) {
 203                         logger("Already searched for ".$search." in the last 24 hours", LOGGER_DEBUG);
 204                         return;
 205                 }
 206         }
 207
 208         $x = fetch_url(get_server()."/lsearch?p=1&n=500&search=".urlencode($search));
 209         $j = json_decode($x);
 210
 211         if (count($j->results)) {
 212                 foreach ($j->results as $jj) {
 213                         // Check if the contact already exists
 214                         $exists = q("SELECT `id`, `last_contact`, `last_failure`, `updated` FROM `gcontact` WHERE `nurl` = '%s'", normalise_link($jj->url));
 215                         if (dbm::is_result($exists)) {
 216                                 logger("Profile ".$jj->url." already exists (".$search.")", LOGGER_DEBUG);
 217
 218                                 if (($exists[0]["last_contact"] < $exists[0]["last_failure"]) AND
 219                                         ($exists[0]["updated"] < $exists[0]["last_failure"])) {
 220                                         continue;
 221                                 }
 222                                 // Update the contact
 223                                 poco_last_updated($jj->url);
 224                                 continue;
 225                         }
 226
 227                         $server_url = poco_detect_server($jj->url);
 228                         if ($server_url != '') {
 229                                 if (!poco_check_server($server_url)) {
 230                                         logger("Friendica server ".$server_url." doesn't answer.", LOGGER_DEBUG);
 231                                         continue;
 232                                 }
 233                                 logger("Friendica server ".$server_url." seems to be okay.", LOGGER_DEBUG);
 234                         }
 235
 236                         $data = probe_url($jj->url);
 237                         if ($data["network"] == NETWORK_DFRN) {
 238                                 logger("Profile ".$jj->url." is reachable (".$search.")", LOGGER_DEBUG);
 239                                 logger("Add profile ".$jj->url." to local directory (".$search.")", LOGGER_DEBUG);
 240
 241                                 if ($jj->tags != "") {
 242                                         $data["keywords"] = $jj->tags;
 243                                 }
 244
 245                                 $data["server_url"] = $data["baseurl"];
 246
 247                                 update_gcontact($data);
 248                         } else {
 249                                 logger("Profile ".$jj->url." is not responding or no Friendica contact - but network ".$data["network"], LOGGER_DEBUG);
 250                         }
 251                 }
 252         }
 253         Cache::set("dirsearch:".$search, time(), CACHE_DAY);
 254 }
 255
 256 /**
 257  * @brief Search for GNU Social user with gstools.org
 258  *
 259  * @param str $search User name
 260  */
 261 function gs_search_user($search) {
 262
 263         // Currently disabled, since the service isn't available anymore.
 264         // It is not removed since I hope that there will be a successor.
 265         return false;
 266
 267         $a = get_app();
 268
 269         $url = "http://gstools.org/api/users_search/".urlencode($search);
 270
 271         $result = z_fetch_url($url);
 272         if (!$result["success"]) {
 273                 return false;
 274         }
 275
 276         $contacts = json_decode($result["body"]);
 277
 278         if ($contacts->status == 'ERROR') {
 279                 return false;
 280         }
 281
 282         /// @TODO AS is considered as a notation for constants (as they usually being written all upper-case)
 283         /// @TODO find all those and convert to all lower-case which is a keyword then
 284         foreach ($contacts->data AS $user) {
 285                 $contact = probe_url($user->site_address."/".$user->name);
 286                 if ($contact["network"] != NETWORK_PHANTOM) {
 287                         $contact["about"] = $user->description;
 288                         update_gcontact($contact);
 289                 }
 290         }
 291 }