include/discover_poco.php

   1 <?php
   2
   3 require_once("boot.php");
   4 require_once("include/socgraph.php");
   5
   6
   7 function discover_poco_run(&$argv, &$argc){
   8         global $a, $db;
   9
  10         if(is_null($a)) {
  11                 $a = new App;
  12         }
  13
  14         if(is_null($db)) {
  15             @include(".htconfig.php");
  16         require_once("include/dba.php");
  17             $db = new dba($db_host, $db_user, $db_pass, $db_data);
  18         unset($db_host, $db_user, $db_pass, $db_data);
  19         };
  20
  21         require_once('include/session.php');
  22         require_once('include/datetime.php');
  23         require_once('include/pidfile.php');
  24
  25         load_config('config');
  26         load_config('system');
  27
  28         // Don't check this stuff if the function is called by the poller
  29         if (App::callstack() != "poller_run") {
  30                 $maxsysload = intval(get_config('system','maxloadavg'));
  31                 if($maxsysload < 1)
  32                         $maxsysload = 50;
  33
  34                 $load = current_load();
  35                 if($load) {
  36                         if(intval($load) > $maxsysload) {
  37                                 logger('system: load '.$load.' too high. discover_poco deferred to next scheduled run.');
  38                                 return;
  39                         }
  40                 }
  41         }
  42
  43         if(($argc > 2) && ($argv[1] == "dirsearch")) {
  44                 $search = urldecode($argv[2]);
  45                 $mode = 1;
  46         } elseif(($argc == 2) && ($argv[1] == "checkcontact")) {
  47                 $mode = 2;
  48         } elseif(($argc == 2) && ($argv[1] == "suggestions")) {
  49                 $mode = 3;
  50         } elseif ($argc == 1) {
  51                 $search = "";
  52                 $mode = 0;
  53         } else
  54                 die("Unknown or missing parameter ".$argv[1]."\n");
  55
  56         // Don't check this stuff if the function is called by the poller
  57         if (App::callstack() != "poller_run") {
  58                 $lockpath = get_lockpath();
  59                 if ($lockpath != '') {
  60                         $pidfile = new pidfile($lockpath, 'discover_poco'.$mode.urlencode($search));
  61                         if($pidfile->is_already_running()) {
  62                                 logger("discover_poco: Already running");
  63                                 if ($pidfile->running_time() > 19*60) {
  64                                         $pidfile->kill();
  65                                         logger("discover_poco: killed stale process");
  66                                         // Calling a new instance
  67                                         if ($mode == 0)
  68                                                 proc_run('php','include/discover_poco.php');
  69                                 }
  70                                 exit;
  71                         }
  72                 }
  73         }
  74
  75         $a->set_baseurl(get_config('system','url'));
  76
  77         load_hooks();
  78
  79         logger('start '.$search);
  80
  81         if ($mode==3)
  82                 update_suggestions();
  83         elseif (($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 function discover_users() {
 103         logger("Discover users", LOGGER_DEBUG);
 104
 105         $users = q("SELECT `url`, `created`, `updated`, `last_failure`, `last_contact`, `server_url` FROM `gcontact`
 106                         WHERE `last_contact` < UTC_TIMESTAMP - INTERVAL 1 MONTH AND
 107                                 `last_failure` < UTC_TIMESTAMP - INTERVAL 1 MONTH AND
 108                                 `network` IN ('%s', '%s', '%s', '%s', '') ORDER BY rand()",
 109                         dbesc(NETWORK_DFRN), dbesc(NETWORK_DIASPORA),
 110                         dbesc(NETWORK_OSTATUS), dbesc(NETWORK_FEED));
 111
 112         if (!$users)
 113                 return;
 114
 115         $checked = 0;
 116
 117         foreach ($users AS $user) {
 118
 119                 $urlparts = parse_url($user["url"]);
 120                 if (!isset($urlparts["scheme"])) {
 121                         q("UPDATE `gcontact` SET `network` = '%s' WHERE `nurl` = '%s'",
 122                                 dbesc(NETWORK_PHANTOM), dbesc(normalise_link($user["url"])));
 123                         continue;
 124                  }
 125
 126                 if (in_array($urlparts["host"], array("www.facebook.com", "facebook.com", "twitter.com",
 127                                                         "identi.ca", "alpha.app.net"))) {
 128                         $networks = array("www.facebook.com" => NETWORK_FACEBOOK,
 129                                         "facebook.com" => NETWORK_FACEBOOK,
 130                                         "twitter.com" => NETWORK_TWITTER,
 131                                         "identi.ca" => NETWORK_PUMPIO,
 132                                         "alpha.app.net" => NETWORK_APPNET);
 133
 134                         q("UPDATE `gcontact` SET `network` = '%s' WHERE `nurl` = '%s'",
 135                                 dbesc($networks[$urlparts["host"]]), dbesc(normalise_link($user["url"])));
 136                         continue;
 137                 }
 138
 139                 if ($user["server_url"] != "")
 140                         $server_url = $user["server_url"];
 141                 else
 142                         $server_url = poco_detect_server($user["url"]);
 143
 144                 if (($server_url == "") OR poco_check_server($server_url, $gcontacts[0]["network"])) {
 145                         logger('Check user '.$user["url"]);
 146                         poco_last_updated($user["url"], true);
 147
 148                         if (++$checked > 100)
 149                                 return;
 150                 } else
 151                         q("UPDATE `gcontact` SET `last_failure` = '%s' WHERE `nurl` = '%s'",
 152                                 dbesc(datetime_convert()), dbesc(normalise_link($user["url"])));
 153         }
 154 }
 155
 156 function discover_directory($search) {
 157
 158         $data = Cache::get("dirsearch:".$search);
 159         if (!is_null($data)){
 160                 // Only search for the same item every 24 hours
 161                 if (time() < $data + (60 * 60 * 24)) {
 162                         logger("Already searched for ".$search." in the last 24 hours", LOGGER_DEBUG);
 163                         return;
 164                 }
 165         }
 166
 167         $x = fetch_url(get_server()."/lsearch?p=1&n=500&search=".urlencode($search));
 168         $j = json_decode($x);
 169
 170         if(count($j->results))
 171                 foreach($j->results as $jj) {
 172                         // Check if the contact already exists
 173                         $exists = q("SELECT `id`, `last_contact`, `last_failure`, `updated` FROM `gcontact` WHERE `nurl` = '%s'", normalise_link($jj->url));
 174                         if ($exists) {
 175                                 logger("Profile ".$jj->url." already exists (".$search.")", LOGGER_DEBUG);
 176
 177                                 if (($exists[0]["last_contact"] < $exists[0]["last_failure"]) AND
 178                                         ($exists[0]["updated"] < $exists[0]["last_failure"]))
 179                                         continue;
 180
 181                                 // Update the contact
 182                                 poco_last_updated($jj->url);
 183                                 continue;
 184                         }
 185
 186                         // Harcoded paths aren't so good. But in this case it is okay.
 187                         // First: We only will get Friendica contacts (which always are using this url schema)
 188                         // Second: There will be no further problems if we are doing a mistake
 189                         $server_url = preg_replace("=(https?://)(.*)/profile/(.*)=ism", "$1$2", $jj->url);
 190                         if ($server_url != $jj->url)
 191                                 if (!poco_check_server($server_url)) {
 192                                         logger("Friendica server ".$server_url." doesn't answer.", LOGGER_DEBUG);
 193                                         continue;
 194                                 }
 195                                         logger("Friendica server ".$server_url." seems to be okay.", LOGGER_DEBUG);
 196
 197                         logger("Check if profile ".$jj->url." is reachable (".$search.")", LOGGER_DEBUG);
 198                         $data = probe_url($jj->url);
 199                         if ($data["network"] == NETWORK_DFRN) {
 200                                 logger("Add profile ".$jj->url." to local directory (".$search.")", LOGGER_DEBUG);
 201                                 poco_check($data["url"], $data["name"], $data["network"], $data["photo"], "", "", "", $jj->tags, $data["addr"], "", 0);
 202                         }
 203                 }
 204         Cache::set("dirsearch:".$search, time(), CACHE_DAY);
 205 }
 206
 207 /**
 208  * @brief Search for GNU Social user with gstools.org
 209  *
 210  * @param str $search User name
 211  */
 212 function gs_search_user($search) {
 213
 214         $a = get_app();
 215
 216         $url = "http://gstools.org/api/users_search/".urlencode($search);
 217
 218         $result = z_fetch_url($url);
 219         if (!$result["success"])
 220                 return false;
 221
 222         $contacts = json_decode($result["body"]);
 223
 224         if ($contacts->status == 'ERROR')
 225                 return false;
 226
 227         foreach($contacts->data AS $user) {
 228                 $contact = probe_url($user->site_address."/".$user->name);
 229                 if ($contact["network"] != NETWORK_PHANTOM) {
 230                         $contact["about"] = $user->description;
 231                         update_gcontact($contact);
 232                 }
 233         }
 234 }
 235
 236
 237 if (array_search(__file__,get_included_files())===0){
 238   discover_poco_run($_SERVER["argv"],$_SERVER["argc"]);
 239   killme();
 240 }