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         $a->start_process();
  22
  23         require_once('include/session.php');
  24         require_once('include/datetime.php');
  25
  26         load_config('config');
  27         load_config('system');
  28
  29         // Don't check this stuff if the function is called by the poller
  30         if (App::callstack() != "poller_run")
  31                 if (App::maxload_reached())
  32                         return;
  33
  34         if(($argc > 2) && ($argv[1] == "dirsearch")) {
  35                 $search = urldecode($argv[2]);
  36                 $mode = 1;
  37         } elseif(($argc == 2) && ($argv[1] == "checkcontact")) {
  38                 $mode = 2;
  39         } elseif(($argc == 2) && ($argv[1] == "suggestions")) {
  40                 $mode = 3;
  41         } elseif ($argc == 1) {
  42                 $search = "";
  43                 $mode = 0;
  44         } else
  45                 die("Unknown or missing parameter ".$argv[1]."\n");
  46
  47         // Don't check this stuff if the function is called by the poller
  48         if (App::callstack() != "poller_run")
  49                 if (App::is_already_running('discover_poco'.$mode.urlencode($search), 'include/discover_poco.php', 1140))
  50                         return;
  51
  52         $a->set_baseurl(get_config('system','url'));
  53
  54         load_hooks();
  55
  56         logger('start '.$search);
  57
  58         if ($mode==3)
  59                 update_suggestions();
  60         elseif (($mode == 2) AND get_config('system','poco_completion'))
  61                 discover_users();
  62         elseif (($mode == 1) AND ($search != "") and get_config('system','poco_local_search')) {
  63                 discover_directory($search);
  64                 gs_search_user($search);
  65         } elseif (($mode == 0) AND ($search == "") and (get_config('system','poco_discovery') > 0)) {
  66                 // Query Friendica and Hubzilla servers for their users
  67                 poco_discover();
  68
  69                 // Query GNU Social servers for their users ("statistics" addon has to be enabled on the GS server)
  70                 if (!get_config('system','ostatus_disabled'))
  71                         gs_discover();
  72         }
  73
  74         logger('end '.$search);
  75
  76         return;
  77 }
  78
  79 function discover_users() {
  80         logger("Discover users", LOGGER_DEBUG);
  81
  82         $users = q("SELECT `url`, `created`, `updated`, `last_failure`, `last_contact`, `server_url` FROM `gcontact`
  83                         WHERE `last_contact` < UTC_TIMESTAMP - INTERVAL 1 MONTH AND
  84                                 `last_failure` < UTC_TIMESTAMP - INTERVAL 1 MONTH AND
  85                                 `network` IN ('%s', '%s', '%s', '%s', '') ORDER BY rand()",
  86                         dbesc(NETWORK_DFRN), dbesc(NETWORK_DIASPORA),
  87                         dbesc(NETWORK_OSTATUS), dbesc(NETWORK_FEED));
  88
  89         if (!$users)
  90                 return;
  91
  92         $checked = 0;
  93
  94         foreach ($users AS $user) {
  95
  96                 $urlparts = parse_url($user["url"]);
  97                 if (!isset($urlparts["scheme"])) {
  98                         q("UPDATE `gcontact` SET `network` = '%s' WHERE `nurl` = '%s'",
  99                                 dbesc(NETWORK_PHANTOM), dbesc(normalise_link($user["url"])));
 100                         continue;
 101                  }
 102
 103                 if (in_array($urlparts["host"], array("www.facebook.com", "facebook.com", "twitter.com",
 104                                                         "identi.ca", "alpha.app.net"))) {
 105                         $networks = array("www.facebook.com" => NETWORK_FACEBOOK,
 106                                         "facebook.com" => NETWORK_FACEBOOK,
 107                                         "twitter.com" => NETWORK_TWITTER,
 108                                         "identi.ca" => NETWORK_PUMPIO,
 109                                         "alpha.app.net" => NETWORK_APPNET);
 110
 111                         q("UPDATE `gcontact` SET `network` = '%s' WHERE `nurl` = '%s'",
 112                                 dbesc($networks[$urlparts["host"]]), dbesc(normalise_link($user["url"])));
 113                         continue;
 114                 }
 115
 116                 if ($user["server_url"] != "")
 117                         $server_url = $user["server_url"];
 118                 else
 119                         $server_url = poco_detect_server($user["url"]);
 120
 121                 if (($server_url == "") OR poco_check_server($server_url, $gcontacts[0]["network"])) {
 122                         logger('Check user '.$user["url"]);
 123                         poco_last_updated($user["url"], true);
 124
 125                         if (++$checked > 100)
 126                                 return;
 127                 } else
 128                         q("UPDATE `gcontact` SET `last_failure` = '%s' WHERE `nurl` = '%s'",
 129                                 dbesc(datetime_convert()), dbesc(normalise_link($user["url"])));
 130         }
 131 }
 132
 133 function discover_directory($search) {
 134
 135         $data = Cache::get("dirsearch:".$search);
 136         if (!is_null($data)){
 137                 // Only search for the same item every 24 hours
 138                 if (time() < $data + (60 * 60 * 24)) {
 139                         logger("Already searched for ".$search." in the last 24 hours", LOGGER_DEBUG);
 140                         return;
 141                 }
 142         }
 143
 144         $x = fetch_url(get_server()."/lsearch?p=1&n=500&search=".urlencode($search));
 145         $j = json_decode($x);
 146
 147         if(count($j->results))
 148                 foreach($j->results as $jj) {
 149                         // Check if the contact already exists
 150                         $exists = q("SELECT `id`, `last_contact`, `last_failure`, `updated` FROM `gcontact` WHERE `nurl` = '%s'", normalise_link($jj->url));
 151                         if ($exists) {
 152                                 logger("Profile ".$jj->url." already exists (".$search.")", LOGGER_DEBUG);
 153
 154                                 if (($exists[0]["last_contact"] < $exists[0]["last_failure"]) AND
 155                                         ($exists[0]["updated"] < $exists[0]["last_failure"]))
 156                                         continue;
 157
 158                                 // Update the contact
 159                                 poco_last_updated($jj->url);
 160                                 continue;
 161                         }
 162
 163                         // Harcoded paths aren't so good. But in this case it is okay.
 164                         // First: We only will get Friendica contacts (which always are using this url schema)
 165                         // Second: There will be no further problems if we are doing a mistake
 166                         $server_url = preg_replace("=(https?://)(.*)/profile/(.*)=ism", "$1$2", $jj->url);
 167                         if ($server_url != $jj->url)
 168                                 if (!poco_check_server($server_url)) {
 169                                         logger("Friendica server ".$server_url." doesn't answer.", LOGGER_DEBUG);
 170                                         continue;
 171                                 }
 172                                         logger("Friendica server ".$server_url." seems to be okay.", LOGGER_DEBUG);
 173
 174                         logger("Check if profile ".$jj->url." is reachable (".$search.")", LOGGER_DEBUG);
 175                         $data = probe_url($jj->url);
 176                         if ($data["network"] == NETWORK_DFRN) {
 177                                 logger("Add profile ".$jj->url." to local directory (".$search.")", LOGGER_DEBUG);
 178                                 poco_check($data["url"], $data["name"], $data["network"], $data["photo"], "", "", "", $jj->tags, $data["addr"], "", 0);
 179                         }
 180                 }
 181         Cache::set("dirsearch:".$search, time(), CACHE_DAY);
 182 }
 183
 184 /**
 185  * @brief Search for GNU Social user with gstools.org
 186  *
 187  * @param str $search User name
 188  */
 189 function gs_search_user($search) {
 190
 191         $a = get_app();
 192
 193         $url = "http://gstools.org/api/users_search/".urlencode($search);
 194
 195         $result = z_fetch_url($url);
 196         if (!$result["success"])
 197                 return false;
 198
 199         $contacts = json_decode($result["body"]);
 200
 201         if ($contacts->status == 'ERROR')
 202                 return false;
 203
 204         foreach($contacts->data AS $user) {
 205                 $contact = probe_url($user->site_address."/".$user->name);
 206                 if ($contact["network"] != NETWORK_PHANTOM) {
 207                         $contact["about"] = $user->description;
 208                         update_gcontact($contact);
 209                 }
 210         }
 211 }
 212
 213
 214 if (array_search(__file__,get_included_files())===0){
 215   discover_poco_run($_SERVER["argv"],$_SERVER["argc"]);
 216   killme();
 217 }