include/discover_poco.php

   1 <?php
   2
   3 use \Friendica\Core\Config;
   4
   5 require_once("boot.php");
   6 require_once("include/socgraph.php");
   7
   8 function discover_poco_run(&$argv, &$argc){
   9         global $a, $db;
  10
  11         if(is_null($a)) {
  12                 $a = new App;
  13         }
  14
  15         if(is_null($db)) {
  16             @include(".htconfig.php");
  17         require_once("include/dba.php");
  18             $db = new dba($db_host, $db_user, $db_pass, $db_data);
  19         unset($db_host, $db_user, $db_pass, $db_data);
  20         };
  21
  22         require_once('include/session.php');
  23         require_once('include/datetime.php');
  24
  25         Config::load();
  26
  27         // Don't check this stuff if the function is called by the poller
  28         if (App::callstack() != "poller_run")
  29                 if ($a->maxload_reached())
  30                         return;
  31
  32         /*
  33         This function can be called in these ways:
  34         - dirsearch <search pattern>: Searches for "search pattern" in the directory. "search pattern" is url encoded.
  35         - checkcontact: Updates gcontact entries
  36         - suggestions: Discover other servers for their contacts.
  37         - server <poco url>: Searches for the poco server list. "poco url" is base64 encoded.
  38         - update_server: Frequently check the first 250 servers for vitality.
  39         */
  40
  41         if(($argc > 2) && ($argv[1] == "dirsearch")) {
  42                 $search = urldecode($argv[2]);
  43                 $mode = 1;
  44         } elseif(($argc == 2) && ($argv[1] == "checkcontact")) {
  45                 $mode = 2;
  46         } elseif(($argc == 2) && ($argv[1] == "suggestions")) {
  47                 $mode = 3;
  48         } elseif(($argc == 3) && ($argv[1] == "server")) {
  49                 $mode = 4;
  50         } elseif(($argc == 2) && ($argv[1] == "update_server")) {
  51                 $mode = 5;
  52         } elseif ($argc == 1) {
  53                 $search = "";
  54                 $mode = 0;
  55         } else {
  56                 die("Unknown or missing parameter ".$argv[1]."\n");
  57         }
  58
  59         // Don't check this stuff if the function is called by the poller
  60         if (App::callstack() != "poller_run")
  61                 if (App::is_already_running('discover_poco'.$mode.urlencode($search), 'include/discover_poco.php', 1140))
  62                         return;
  63
  64         $a->set_baseurl(get_config('system','url'));
  65
  66         load_hooks();
  67
  68         logger('start '.$search);
  69
  70         if ($mode == 5) {
  71                 update_server();
  72         } elseif ($mode == 4) {
  73                 $server_url = base64_decode($argv[2]);
  74                 if ($server_url == "") {
  75                         return;
  76                 }
  77                 $server_url = filter_var($server_url, FILTER_SANITIZE_URL);
  78                 if (substr(normalise_link($server_url), 0, 7) != "http://") {
  79                         return;
  80                 }
  81                 $result = "Checking server ".$server_url." - ";
  82                 $ret = poco_check_server($server_url);
  83                 if ($ret) {
  84                         $result .= "success";
  85                 } else {
  86                         $result .= "failed";
  87                 }
  88                 logger($result, LOGGER_DEBUG);
  89         } elseif ($mode == 3) {
  90                 update_suggestions();
  91         } elseif (($mode == 2) AND get_config('system','poco_completion')) {
  92                 discover_users();
  93         } elseif (($mode == 1) AND ($search != "") and get_config('system','poco_local_search')) {
  94                 discover_directory($search);
  95                 gs_search_user($search);
  96         } elseif (($mode == 0) AND ($search == "") and (get_config('system','poco_discovery') > 0)) {
  97                 // Query Friendica and Hubzilla servers for their users
  98                 poco_discover();
  99
 100                 // Query GNU Social servers for their users ("statistics" addon has to be enabled on the GS server)
 101                 if (!get_config('system','ostatus_disabled'))
 102                         gs_discover();
 103         }
 104
 105         logger('end '.$search);
 106
 107         return;
 108 }
 109
 110 /**
 111  * @brief Updates the first 250 servers
 112  *
 113  */
 114 function update_server() {
 115         $r = q("SELECT `url`, `created`, `last_failure`, `last_contact` FROM `gserver` ORDER BY rand()");
 116
 117         if (!dbm::is_result($r)) {
 118                 return;
 119         }
 120
 121         $updated = 0;
 122
 123         foreach ($r AS $server) {
 124                 if (!poco_do_update($server["created"], "", $server["last_failure"], $server["last_contact"])) {
 125                         continue;
 126                 }
 127                 logger('Update server status for server '.$server["url"], LOGGER_DEBUG);
 128
 129                 proc_run(PRIORITY_LOW, "include/discover_poco.php", "server", base64_encode($server["url"]));
 130
 131                 if (++$updated > 250) {
 132                         return;
 133                 }
 134         }
 135 }
 136
 137 function discover_users() {
 138         logger("Discover users", LOGGER_DEBUG);
 139
 140         $users = q("SELECT `url`, `created`, `updated`, `last_failure`, `last_contact`, `server_url` FROM `gcontact`
 141                         WHERE `last_contact` < UTC_TIMESTAMP - INTERVAL 1 MONTH AND
 142                                 `last_failure` < UTC_TIMESTAMP - INTERVAL 1 MONTH AND
 143                                 `network` IN ('%s', '%s', '%s', '%s', '') ORDER BY rand()",
 144                         dbesc(NETWORK_DFRN), dbesc(NETWORK_DIASPORA),
 145                         dbesc(NETWORK_OSTATUS), dbesc(NETWORK_FEED));
 146
 147         if (!$users)
 148                 return;
 149
 150         $checked = 0;
 151
 152         foreach ($users AS $user) {
 153
 154                 $urlparts = parse_url($user["url"]);
 155                 if (!isset($urlparts["scheme"])) {
 156                         q("UPDATE `gcontact` SET `network` = '%s' WHERE `nurl` = '%s'",
 157                                 dbesc(NETWORK_PHANTOM), dbesc(normalise_link($user["url"])));
 158                         continue;
 159                  }
 160
 161                 if (in_array($urlparts["host"], array("www.facebook.com", "facebook.com", "twitter.com",
 162                                                         "identi.ca", "alpha.app.net"))) {
 163                         $networks = array("www.facebook.com" => NETWORK_FACEBOOK,
 164                                         "facebook.com" => NETWORK_FACEBOOK,
 165                                         "twitter.com" => NETWORK_TWITTER,
 166                                         "identi.ca" => NETWORK_PUMPIO,
 167                                         "alpha.app.net" => NETWORK_APPNET);
 168
 169                         q("UPDATE `gcontact` SET `network` = '%s' WHERE `nurl` = '%s'",
 170                                 dbesc($networks[$urlparts["host"]]), dbesc(normalise_link($user["url"])));
 171                         continue;
 172                 }
 173
 174                 if ($user["server_url"] != "")
 175                         $server_url = $user["server_url"];
 176                 else
 177                         $server_url = poco_detect_server($user["url"]);
 178
 179                 if (($server_url == "") OR poco_check_server($server_url, $gcontacts[0]["network"])) {
 180                         logger('Check user '.$user["url"]);
 181                         poco_last_updated($user["url"], true);
 182
 183                         if (++$checked > 100)
 184                                 return;
 185                 } else
 186                         q("UPDATE `gcontact` SET `last_failure` = '%s' WHERE `nurl` = '%s'",
 187                                 dbesc(datetime_convert()), dbesc(normalise_link($user["url"])));
 188         }
 189 }
 190
 191 function discover_directory($search) {
 192
 193         $data = Cache::get("dirsearch:".$search);
 194         if (!is_null($data)){
 195                 // Only search for the same item every 24 hours
 196                 if (time() < $data + (60 * 60 * 24)) {
 197                         logger("Already searched for ".$search." in the last 24 hours", LOGGER_DEBUG);
 198                         return;
 199                 }
 200         }
 201
 202         $x = fetch_url(get_server()."/lsearch?p=1&n=500&search=".urlencode($search));
 203         $j = json_decode($x);
 204
 205         if(count($j->results))
 206                 foreach($j->results as $jj) {
 207                         // Check if the contact already exists
 208                         $exists = q("SELECT `id`, `last_contact`, `last_failure`, `updated` FROM `gcontact` WHERE `nurl` = '%s'", normalise_link($jj->url));
 209                         if ($exists) {
 210                                 logger("Profile ".$jj->url." already exists (".$search.")", LOGGER_DEBUG);
 211
 212                                 if (($exists[0]["last_contact"] < $exists[0]["last_failure"]) AND
 213                                         ($exists[0]["updated"] < $exists[0]["last_failure"]))
 214                                         continue;
 215
 216                                 // Update the contact
 217                                 poco_last_updated($jj->url);
 218                                 continue;
 219                         }
 220
 221                         // Harcoded paths aren't so good. But in this case it is okay.
 222                         // First: We only will get Friendica contacts (which always are using this url schema)
 223                         // Second: There will be no further problems if we are doing a mistake
 224                         $server_url = preg_replace("=(https?://)(.*)/profile/(.*)=ism", "$1$2", $jj->url);
 225                         if ($server_url != $jj->url)
 226                                 if (!poco_check_server($server_url)) {
 227                                         logger("Friendica server ".$server_url." doesn't answer.", LOGGER_DEBUG);
 228                                         continue;
 229                                 }
 230                                         logger("Friendica server ".$server_url." seems to be okay.", LOGGER_DEBUG);
 231
 232                         logger("Check if profile ".$jj->url." is reachable (".$search.")", LOGGER_DEBUG);
 233                         $data = probe_url($jj->url);
 234                         if ($data["network"] == NETWORK_DFRN) {
 235                                 logger("Add profile ".$jj->url." to local directory (".$search.")", LOGGER_DEBUG);
 236                                 poco_check($data["url"], $data["name"], $data["network"], $data["photo"], "", "", "", $jj->tags, $data["addr"], "", 0);
 237                         }
 238                 }
 239         Cache::set("dirsearch:".$search, time(), CACHE_DAY);
 240 }
 241
 242 /**
 243  * @brief Search for GNU Social user with gstools.org
 244  *
 245  * @param str $search User name
 246  */
 247 function gs_search_user($search) {
 248
 249         // Currently disabled, since the service isn't available anymore.
 250         // It is not removed since I hope that there will be a successor.
 251         return false;
 252
 253         $a = get_app();
 254
 255         $url = "http://gstools.org/api/users_search/".urlencode($search);
 256
 257         $result = z_fetch_url($url);
 258         if (!$result["success"])
 259                 return false;
 260
 261         $contacts = json_decode($result["body"]);
 262
 263         if ($contacts->status == 'ERROR')
 264                 return false;
 265
 266         foreach($contacts->data AS $user) {
 267                 $contact = probe_url($user->site_address."/".$user->name);
 268                 if ($contact["network"] != NETWORK_PHANTOM) {
 269                         $contact["about"] = $user->description;
 270                         update_gcontact($contact);
 271                 }
 272         }
 273 }
 274
 275
 276 if (array_search(__file__,get_included_files())===0){
 277   discover_poco_run($_SERVER["argv"],$_SERVER["argc"]);
 278   killme();
 279 }