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         $maxsysload = intval(get_config('system','maxloadavg'));
  29         if($maxsysload < 1)
  30                 $maxsysload = 50;
  31         if(function_exists('sys_getloadavg')) {
  32                 $load = sys_getloadavg();
  33                 if(intval($load[0]) > $maxsysload) {
  34                         logger('system: load ' . $load[0] . ' too high. discover_poco deferred to next scheduled run.');
  35                         return;
  36                 }
  37         }
  38
  39         if(($argc > 2) && ($argv[1] == "dirsearch")) {
  40                 $search = urldecode($argv[2]);
  41                 $mode = 1;
  42         } elseif(($argc == 2) && ($argv[1] == "checkcontact")) {
  43                 $mode = 2;
  44         } elseif(($argc == 2) && ($argv[1] == "suggestions")) {
  45                 $mode = 3;
  46         } elseif ($argc == 1) {
  47                 $search = "";
  48                 $mode = 0;
  49         } else
  50                 die("Unknown or missing parameter ".$argv[1]."\n");
  51
  52         $lockpath = get_lockpath();
  53         if ($lockpath != '') {
  54                 $pidfile = new pidfile($lockpath, 'discover_poco'.$mode.urlencode($search));
  55                 if($pidfile->is_already_running()) {
  56                         logger("discover_poco: Already running");
  57                         if ($pidfile->running_time() > 19*60) {
  58                                 $pidfile->kill();
  59                                 logger("discover_poco: killed stale process");
  60                                 // Calling a new instance
  61                                 if ($mode == 0)
  62                                         proc_run('php','include/discover_poco.php');
  63                         }
  64                         exit;
  65                 }
  66         }
  67
  68         $a->set_baseurl(get_config('system','url'));
  69
  70         load_hooks();
  71
  72         logger('start '.$search);
  73
  74         if ($mode==3)
  75                 update_suggestions();
  76         elseif (($mode == 2) AND get_config('system','poco_completion'))
  77                 discover_users();
  78         elseif (($mode == 1) AND ($search != "") and get_config('system','poco_local_search'))
  79                 discover_directory($search);
  80         elseif (($mode == 0) AND ($search == "") and (get_config('system','poco_discovery') > 0))
  81                 poco_discover();
  82
  83         logger('end '.$search);
  84
  85         return;
  86 }
  87
  88 function discover_users() {
  89         logger("Discover users", LOGGER_DEBUG);
  90
  91         $users = q("SELECT `url`, `created`, `updated`, `last_failure`, `last_contact`, `server_url` FROM `gcontact`
  92                         WHERE `last_contact` < UTC_TIMESTAMP - INTERVAL 1 MONTH AND
  93                                 `last_failure` < UTC_TIMESTAMP - INTERVAL 1 MONTH AND
  94                                 `network` IN ('%s', '%s', '%s', '%s', '') ORDER BY rand()",
  95                         dbesc(NETWORK_DFRN), dbesc(NETWORK_DIASPORA),
  96                         dbesc(NETWORK_OSTATUS), dbesc(NETWORK_FEED));
  97
  98         if (!$users)
  99                 return;
 100
 101         $checked = 0;
 102
 103         foreach ($users AS $user) {
 104
 105                 $urlparts = parse_url($user["url"]);
 106                 if (!isset($urlparts["scheme"])) {
 107                         q("UPDATE `gcontact` SET `network` = '%s' WHERE `nurl` = '%s'",
 108                                 dbesc(NETWORK_PHANTOM), dbesc(normalise_link($user["url"])));
 109                         continue;
 110                  }
 111
 112                 if (in_array($urlparts["host"], array("www.facebook.com", "facebook.com", "twitter.com",
 113                                                         "identi.ca", "alpha.app.net"))) {
 114                         $networks = array("www.facebook.com" => NETWORK_FACEBOOK,
 115                                         "facebook.com" => NETWORK_FACEBOOK,
 116                                         "twitter.com" => NETWORK_TWITTER,
 117                                         "identi.ca" => NETWORK_PUMPIO,
 118                                         "alpha.app.net" => NETWORK_APPNET);
 119
 120                         q("UPDATE `gcontact` SET `network` = '%s' WHERE `nurl` = '%s'",
 121                                 dbesc($networks[$urlparts["host"]]), dbesc(normalise_link($user["url"])));
 122                         continue;
 123                 }
 124
 125                 if ($user["server_url"] != "")
 126                         $server_url = $user["server_url"];
 127                 else
 128                         $server_url = poco_detect_server($user["url"]);
 129
 130                 if (poco_check_server($server_url, $gcontacts[0]["network"])) {
 131                         logger('Check user '.$user["url"]);
 132                         poco_last_updated($user["url"], true);
 133
 134                         if (++$checked > 100)
 135                                 return;
 136                 } else
 137                         q("UPDATE `gcontact` SET `last_failure` = '%s' WHERE `nurl` = '%s'",
 138                                 dbesc(datetime_convert()), dbesc(normalise_link($user["url"])));
 139         }
 140 }
 141
 142 function discover_directory($search) {
 143
 144         $data = Cache::get("dirsearch:".$search);
 145         if (!is_null($data)){
 146                 // Only search for the same item every 24 hours
 147                 if (time() < $data + (60 * 60 * 24)) {
 148                         logger("Already searched for ".$search." in the last 24 hours", LOGGER_DEBUG);
 149                         return;
 150                 }
 151         }
 152
 153         $x = fetch_url(get_server()."/lsearch?p=1&n=500&search=".urlencode($search));
 154         $j = json_decode($x);
 155
 156         if(count($j->results))
 157                 foreach($j->results as $jj) {
 158                         // Check if the contact already exists
 159                         $exists = q("SELECT `id`, `last_contact`, `last_failure`, `updated` FROM `gcontact` WHERE `nurl` = '%s'", normalise_link($jj->url));
 160                         if ($exists) {
 161                                 logger("Profile ".$jj->url." already exists (".$search.")", LOGGER_DEBUG);
 162
 163                                 if (($exists[0]["last_contact"] < $exists[0]["last_failure"]) AND
 164                                         ($exists[0]["updated"] < $exists[0]["last_failure"]))
 165                                         continue;
 166
 167                                 // Update the contact
 168                                 poco_last_updated($jj->url);
 169                                 continue;
 170                         }
 171
 172                         // Harcoded paths aren't so good. But in this case it is okay.
 173                         // First: We only will get Friendica contacts (which always are using this url schema)
 174                         // Second: There will be no further problems if we are doing a mistake
 175                         $server_url = preg_replace("=(https?://)(.*)/profile/(.*)=ism", "$1$2", $jj->url);
 176                         if ($server_url != $jj->url)
 177                                 if (!poco_check_server($server_url)) {
 178                                         logger("Friendica server ".$server_url." doesn't answer.", LOGGER_DEBUG);
 179                                         continue;
 180                                 }
 181                                         logger("Friendica server ".$server_url." seems to be okay.", LOGGER_DEBUG);
 182
 183                         logger("Check if profile ".$jj->url." is reachable (".$search.")", LOGGER_DEBUG);
 184                         $data = probe_url($jj->url);
 185                         if ($data["network"] == NETWORK_DFRN) {
 186                                 logger("Add profile ".$jj->url." to local directory (".$search.")", LOGGER_DEBUG);
 187                                 poco_check($data["url"], $data["name"], $data["network"], $data["photo"], "", "", "", $jj->tags, $data["addr"], "", 0);
 188                         }
 189                 }
 190         Cache::set("dirsearch:".$search, time(), CACHE_DAY);
 191 }
 192
 193 if (array_search(__file__,get_included_files())===0){
 194   discover_poco_run($_SERVER["argv"],$_SERVER["argc"]);
 195   killme();
 196 }