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
  32         $load = current_load();
  33         if($load) {
  34                 if(intval($load) > $maxsysload) {
  35                         logger('system: load ' . $load . ' too high. discover_poco deferred to next scheduled run.');
  36                         return;
  37                 }
  38         }
  39
  40         if(($argc > 2) && ($argv[1] == "dirsearch")) {
  41                 $search = urldecode($argv[2]);
  42                 $mode = 1;
  43         } elseif(($argc == 2) && ($argv[1] == "checkcontact")) {
  44                 $mode = 2;
  45         } elseif(($argc == 2) && ($argv[1] == "suggestions")) {
  46                 $mode = 3;
  47         } elseif ($argc == 1) {
  48                 $search = "";
  49                 $mode = 0;
  50         } else
  51                 die("Unknown or missing parameter ".$argv[1]."\n");
  52
  53         $lockpath = get_lockpath();
  54         if ($lockpath != '') {
  55                 $pidfile = new pidfile($lockpath, 'discover_poco'.$mode.urlencode($search));
  56                 if($pidfile->is_already_running()) {
  57                         logger("discover_poco: Already running");
  58                         if ($pidfile->running_time() > 19*60) {
  59                                 $pidfile->kill();
  60                                 logger("discover_poco: killed stale process");
  61                                 // Calling a new instance
  62                                 if ($mode == 0)
  63                                         proc_run('php','include/discover_poco.php');
  64                         }
  65                         exit;
  66                 }
  67         }
  68
  69         $a->set_baseurl(get_config('system','url'));
  70
  71         load_hooks();
  72
  73         logger('start '.$search);
  74
  75         if ($mode==3)
  76                 update_suggestions();
  77         elseif (($mode == 2) AND get_config('system','poco_completion'))
  78                 discover_users();
  79         elseif (($mode == 1) AND ($search != "") and get_config('system','poco_local_search')) {
  80                 discover_directory($search);
  81                 gs_search_user($search);
  82         } elseif (($mode == 0) AND ($search == "") and (get_config('system','poco_discovery') > 0)) {
  83                 // Query Friendica and Hubzilla servers for their users
  84                 poco_discover();
  85
  86                 // Query GNU Social servers for their users ("statistics" addon has to be enabled on the GS server)
  87                 if (!get_config('system','ostatus_disabled'))
  88                         gs_discover();
  89         }
  90
  91         logger('end '.$search);
  92
  93         return;
  94 }
  95
  96 function discover_users() {
  97         logger("Discover users", LOGGER_DEBUG);
  98
  99         $users = q("SELECT `url`, `created`, `updated`, `last_failure`, `last_contact`, `server_url` FROM `gcontact`
 100                         WHERE `last_contact` < UTC_TIMESTAMP - INTERVAL 1 MONTH AND
 101                                 `last_failure` < UTC_TIMESTAMP - INTERVAL 1 MONTH AND
 102                                 `network` IN ('%s', '%s', '%s', '%s', '') ORDER BY rand()",
 103                         dbesc(NETWORK_DFRN), dbesc(NETWORK_DIASPORA),
 104                         dbesc(NETWORK_OSTATUS), dbesc(NETWORK_FEED));
 105
 106         if (!$users)
 107                 return;
 108
 109         $checked = 0;
 110
 111         foreach ($users AS $user) {
 112
 113                 $urlparts = parse_url($user["url"]);
 114                 if (!isset($urlparts["scheme"])) {
 115                         q("UPDATE `gcontact` SET `network` = '%s' WHERE `nurl` = '%s'",
 116                                 dbesc(NETWORK_PHANTOM), dbesc(normalise_link($user["url"])));
 117                         continue;
 118                  }
 119
 120                 if (in_array($urlparts["host"], array("www.facebook.com", "facebook.com", "twitter.com",
 121                                                         "identi.ca", "alpha.app.net"))) {
 122                         $networks = array("www.facebook.com" => NETWORK_FACEBOOK,
 123                                         "facebook.com" => NETWORK_FACEBOOK,
 124                                         "twitter.com" => NETWORK_TWITTER,
 125                                         "identi.ca" => NETWORK_PUMPIO,
 126                                         "alpha.app.net" => NETWORK_APPNET);
 127
 128                         q("UPDATE `gcontact` SET `network` = '%s' WHERE `nurl` = '%s'",
 129                                 dbesc($networks[$urlparts["host"]]), dbesc(normalise_link($user["url"])));
 130                         continue;
 131                 }
 132
 133                 if ($user["server_url"] != "")
 134                         $server_url = $user["server_url"];
 135                 else
 136                         $server_url = poco_detect_server($user["url"]);
 137
 138                 if (($server_url == "") OR poco_check_server($server_url, $gcontacts[0]["network"])) {
 139                         logger('Check user '.$user["url"]);
 140                         poco_last_updated($user["url"], true);
 141
 142                         if (++$checked > 100)
 143                                 return;
 144                 } else
 145                         q("UPDATE `gcontact` SET `last_failure` = '%s' WHERE `nurl` = '%s'",
 146                                 dbesc(datetime_convert()), dbesc(normalise_link($user["url"])));
 147         }
 148 }
 149
 150 function discover_directory($search) {
 151
 152         $data = Cache::get("dirsearch:".$search);
 153         if (!is_null($data)){
 154                 // Only search for the same item every 24 hours
 155                 if (time() < $data + (60 * 60 * 24)) {
 156                         logger("Already searched for ".$search." in the last 24 hours", LOGGER_DEBUG);
 157                         return;
 158                 }
 159         }
 160
 161         $x = fetch_url(get_server()."/lsearch?p=1&n=500&search=".urlencode($search));
 162         $j = json_decode($x);
 163
 164         if(count($j->results))
 165                 foreach($j->results as $jj) {
 166                         // Check if the contact already exists
 167                         $exists = q("SELECT `id`, `last_contact`, `last_failure`, `updated` FROM `gcontact` WHERE `nurl` = '%s'", normalise_link($jj->url));
 168                         if ($exists) {
 169                                 logger("Profile ".$jj->url." already exists (".$search.")", LOGGER_DEBUG);
 170
 171                                 if (($exists[0]["last_contact"] < $exists[0]["last_failure"]) AND
 172                                         ($exists[0]["updated"] < $exists[0]["last_failure"]))
 173                                         continue;
 174
 175                                 // Update the contact
 176                                 poco_last_updated($jj->url);
 177                                 continue;
 178                         }
 179
 180                         // Harcoded paths aren't so good. But in this case it is okay.
 181                         // First: We only will get Friendica contacts (which always are using this url schema)
 182                         // Second: There will be no further problems if we are doing a mistake
 183                         $server_url = preg_replace("=(https?://)(.*)/profile/(.*)=ism", "$1$2", $jj->url);
 184                         if ($server_url != $jj->url)
 185                                 if (!poco_check_server($server_url)) {
 186                                         logger("Friendica server ".$server_url." doesn't answer.", LOGGER_DEBUG);
 187                                         continue;
 188                                 }
 189                                         logger("Friendica server ".$server_url." seems to be okay.", LOGGER_DEBUG);
 190
 191                         logger("Check if profile ".$jj->url." is reachable (".$search.")", LOGGER_DEBUG);
 192                         $data = probe_url($jj->url);
 193                         if ($data["network"] == NETWORK_DFRN) {
 194                                 logger("Add profile ".$jj->url." to local directory (".$search.")", LOGGER_DEBUG);
 195                                 poco_check($data["url"], $data["name"], $data["network"], $data["photo"], "", "", "", $jj->tags, $data["addr"], "", 0);
 196                         }
 197                 }
 198         Cache::set("dirsearch:".$search, time(), CACHE_DAY);
 199 }
 200
 201 /**
 202  * @brief Search for GNU Social user with gstools.org
 203  *
 204  * @param str $search User name
 205  */
 206 function gs_search_user($search) {
 207
 208         $a = get_app();
 209
 210         $url = "http://gstools.org/api/users_search/".urlencode($search);
 211
 212         $result = z_fetch_url($url);
 213         if (!$result["success"])
 214                 return false;
 215
 216         $contacts = json_decode($result["body"]);
 217
 218         if ($contacts->status == 'ERROR')
 219                 return false;
 220
 221         foreach($contacts->data AS $user) {
 222                 $contact = probe_url($user->site_address."/".$user->name);
 223                 if ($contact["network"] != NETWORK_PHANTOM) {
 224                         $contact["about"] = $user->description;
 225                         update_gcontact($contact);
 226                 }
 227         }
 228 }
 229
 230
 231 if (array_search(__file__,get_included_files())===0){
 232   discover_poco_run($_SERVER["argv"],$_SERVER["argc"]);
 233   killme();
 234 }