include/discover_poco.php

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