]> git.mxchange.org Git - friendica.git/commitdiff
Merge remote-tracking branch 'upstream/develop' into 1702-detect-server
authorMichael <heluecht@pirati.ca>
Sun, 12 Mar 2017 20:49:48 +0000 (20:49 +0000)
committerMichael <heluecht@pirati.ca>
Sun, 12 Mar 2017 20:49:48 +0000 (20:49 +0000)
1  2 
include/socgraph.php

diff --combined include/socgraph.php
index 2d27c863b2113c72840dc22bc36ca7f8456dfeae,01d2cff019f34cbf0e859b19f1284d4b5c96a6be..dee3bfc198ff14c48c0dd2d0f816e427b6a83c95
@@@ -309,18 -309,7 +309,18 @@@ function poco_check($profile_url, $name
  
        logger("profile-check generation: ".$generation." Network: ".$network." URL: ".$profile_url." name: ".$name." avatar: ".$profile_photo, LOGGER_DEBUG);
  
 -      poco_check_server($server_url, $network);
 +      // We check the server url to be sure that it is a real one
 +      $server_url2 = poco_detect_server($profile_url);
 +
 +      // We are no sure that it is a correct URL. So we use it in the future
 +      if ($server_url2 != "") {
 +              $server_url = $server_url2;
 +      }
 +
 +      // The server URL doesn't seem to be valid, so we don't store it.
 +      if (!poco_check_server($server_url, $network)) {
 +              $server_url = "";
 +      }
  
        $gcontact = array("url" => $profile_url,
                        "addr" => $addr,
@@@ -412,47 -401,13 +412,47 @@@ function poco_detect_server($profile) 
  
        // Mastodon
        if ($server_url == "") {
 -              $red = preg_replace("=(https?://)(.*)/users/(.*)=ism", "$1$2", $profile);
 -              if ($red != $profile) {
 -                      $server_url = $red;
 +              $mastodon = preg_replace("=(https?://)(.*)/users/(.*)=ism", "$1$2", $profile);
 +              if ($mastodon != $profile) {
 +                      $server_url = $mastodon;
                        $network = NETWORK_OSTATUS;
                }
        }
  
 +      // Numeric OStatus variant
 +      if ($server_url == "") {
 +              $ostatus = preg_replace("=(https?://)(.*)/user/(.*)=ism", "$1$2", $profile);
 +              if ($ostatus != $profile) {
 +                      $server_url = $ostatus;
 +                      $network = NETWORK_OSTATUS;
 +              }
 +      }
 +
 +      // Wild guess
 +      if ($server_url == "") {
 +              $base = preg_replace("=(https?://)(.*?)/(.*)=ism", "$1$2", $profile);
 +              if (base != $profile) {
 +                      $server_url = $base;
 +                      $network = NETWORK_PHANTOM;
 +              }
 +      }
 +
 +      if ($server_url == "") {
 +              return "";
 +      }
 +
 +      $r = q("SELECT `id` FROM `gserver` WHERE `nurl` = '%s' AND `last_contact` > `last_failure`",
 +              dbesc(normalise_link($server_url)));
 +      if (dbm::is_result($r)) {
 +              return $server_url;
 +      }
 +
 +      // Fetch the host-meta to check if this really is a server
 +      $serverret = z_fetch_url($server_url."/.well-known/host-meta");
 +      if (!$serverret["success"]) {
 +              return "";
 +      }
 +
        return $server_url;
  }
  
@@@ -469,12 -424,10 +469,12 @@@ function poco_last_updated($profile, $f
                q("UPDATE `gcontact` SET `created` = '%s' WHERE `nurl` = '%s'",
                        dbesc(datetime_convert()), dbesc(normalise_link($profile)));
  
 -      if ($gcontacts[0]["server_url"] != "")
 +      if ($gcontacts[0]["server_url"] != "") {
                $server_url = $gcontacts[0]["server_url"];
 -      else
 +      }
 +      if (($server_url == '') OR ($gcontacts[0]["server_url"] == $gcontacts[0]["nurl"])) {
                $server_url = poco_detect_server($profile);
 +      }
  
        if (!in_array($gcontacts[0]["network"], array(NETWORK_DFRN, NETWORK_DIASPORA, NETWORK_FEED, NETWORK_OSTATUS, ""))) {
                logger("Profile ".$profile.": Network type ".$gcontacts[0]["network"]." can't be checked", LOGGER_DEBUG);
@@@ -727,106 -680,6 +727,106 @@@ function poco_to_boolean($val) 
        return ($val);
  }
  
 +/**
 + * @brief Detect server type (Hubzilla or Friendica) via the poco data
 + *
 + * @param object $data POCO data
 + * @return array Server data
 + */
 +function poco_detect_poco_data($data) {
 +      $server = false;
 +
 +      if (!isset($data->entry)) {
 +              return false;
 +      }
 +
 +      if (count($data->entry) == 0) {
 +              return false;
 +      }
 +
 +      if (!isset($data->entry[0]->urls)) {
 +              return false;
 +      }
 +
 +      if (count($data->entry[0]->urls) == 0) {
 +              return false;
 +      }
 +
 +      foreach ($data->entry[0]->urls AS $url) {
 +              if ($url->type == 'zot') {
 +                      $server = array();
 +                      $server["platform"] = 'Hubzilla';
 +                      $server["network"] = NETWORK_DIASPORA;
 +                      return $server;
 +              }
 +      }
 +      return false;
 +}
 +
 +/**
 + * @brief Detect server type (Hubzilla or Friendica) via the front page body
 + *
 + * @param string $body Front page of the server
 + * @return array Server data
 + */
 +function poco_detect_server_type($body) {
 +      $server = false;
 +
 +      $doc = new \DOMDocument();
 +      @$doc->loadHTML($body);
 +      $xpath = new \DomXPath($doc);
 +
 +      $list = $xpath->query("//meta[@name]");
 +
 +      foreach ($list as $node) {
 +              $attr = array();
 +              if ($node->attributes->length) {
 +                      foreach ($node->attributes as $attribute) {
 +                              $attr[$attribute->name] = $attribute->value;
 +                      }
 +              }
 +              if ($attr['name'] == 'generator') {
 +                      $version_part = explode(" ", $attr['content']);
 +                      if (count($version_part) == 2) {
 +                              if (in_array($version_part[0], array("Friendika", "Friendica"))) {
 +                                      $server = array();
 +                                      $server["platform"] = $version_part[0];
 +                                      $server["version"] = $version_part[1];
 +                                      $server["network"] = NETWORK_DFRN;
 +                              }
 +                      }
 +              }
 +      }
 +
 +      if (!$server) {
 +              $list = $xpath->query("//meta[@property]");
 +
 +              foreach ($list as $node) {
 +                      $attr = array();
 +                      if ($node->attributes->length) {
 +                              foreach ($node->attributes as $attribute) {
 +                                      $attr[$attribute->name] = $attribute->value;
 +                              }
 +                      }
 +                      if ($attr['property'] == 'generator') {
 +                              if (in_array($attr['content'], array("hubzilla", "BlaBlaNet"))) {
 +                                      $server = array();
 +                                      $server["platform"] = $attr['content'];
 +                                      $server["version"] = "";
 +                                      $server["network"] = NETWORK_DIASPORA;
 +                              }
 +                      }
 +              }
 +      }
 +
 +      if (!$server) {
 +              return false;
 +      }
 +
 +      $server["site_name"] = $xpath->evaluate($element."//head/title/text()", $context)->item(0)->nodeValue;
 +      return $server;
 +}
 +
  function poco_check_server($server_url, $network = "", $force = false) {
  
        // Unify the server address
                return false;
  
        $servers = q("SELECT * FROM `gserver` WHERE `nurl` = '%s'", dbesc(normalise_link($server_url)));
-       if ($servers) {
+       if (dbm::is_result($servers)) {
  
                if ($servers[0]["created"] == "0000-00-00 00:00:00")
                        q("UPDATE `gserver` SET `created` = '%s' WHERE `nurl` = '%s'",
        logger("Server ".$server_url." is outdated or unknown. Start discovery. Force: ".$force." Created: ".$servers[0]["created"]." Failure: ".$last_failure." Contact: ".$last_contact, LOGGER_DEBUG);
  
        $failure = false;
 +      $possible_failure = false;
        $orig_last_failure = $last_failure;
 +      $orig_last_contact = $last_contact;
  
        // Check if the page is accessible via SSL.
+       $orig_server_url = $server_url;
        $server_url = str_replace("http://", "https://", $server_url);
-       $serverret = z_fetch_url($server_url."/.well-known/host-meta");
+       // We set the timeout to 20 seconds since this operation should be done in no time if the server was vital
+       $serverret = z_fetch_url($server_url."/.well-known/host-meta", false, $redirects, array('timeout' => 20));
+       // Quit if there is a timeout.
+       // But we want to make sure to only quit if we are mostly sure that this server url fits.
+       if (dbm::is_result($servers) AND ($orig_server_url == $server_url) AND
+               ($serverret['errno'] == CURLE_OPERATION_TIMEDOUT)) {
+               logger("Connection to server ".$server_url." timed out.", LOGGER_DEBUG);
+               return false;
+       }
  
        // Maybe the page is unencrypted only?
        $xmlobj = @simplexml_load_string($serverret["body"],'SimpleXMLElement',0, "http://docs.oasis-open.org/ns/xri/xrd-1.0");
        if (!$serverret["success"] OR ($serverret["body"] == "") OR (@sizeof($xmlobj) == 0) OR !is_object($xmlobj)) {
                $server_url = str_replace("https://", "http://", $server_url);
-               $serverret = z_fetch_url($server_url."/.well-known/host-meta");
+               // We set the timeout to 20 seconds since this operation should be done in no time if the server was vital
+               $serverret = z_fetch_url($server_url."/.well-known/host-meta", false, $redirects, array('timeout' => 20));
+               // Quit if there is a timeout
+               if ($serverret['errno'] == CURLE_OPERATION_TIMEDOUT) {
+                       logger("Connection to server ".$server_url." timed out.", LOGGER_DEBUG);
+                       return false;
+               }
  
                $xmlobj = @simplexml_load_string($serverret["body"],'SimpleXMLElement',0, "http://docs.oasis-open.org/ns/xri/xrd-1.0");
        }
  
        if (!$serverret["success"] OR ($serverret["body"] == "") OR (sizeof($xmlobj) == 0) OR !is_object($xmlobj)) {
                // Workaround for bad configured servers (known nginx problem)
-               if ($serverret["debug"]["http_code"] != "403") {
+               if (!in_array($serverret["debug"]["http_code"], array("403", "404"))) {
                        $last_failure = datetime_convert();
                        $failure = true;
                }
 +              $possible_failure = true;
        } elseif ($network == NETWORK_DIASPORA)
                $last_contact = datetime_convert();
  
 +      // If the server has no possible failure we reset the cached data
 +      if (!$possible_failure) {
 +              $version = "";
 +              $platform = "";
 +              $site_name = "";
 +              $info = "";
 +              $register_policy = -1;
 +      }
 +
 +      // Look for poco
        if (!$failure) {
 -              // Test for Diaspora
 +              $serverret = z_fetch_url($server_url."/poco");
 +              if ($serverret["success"]) {
 +                      $data = json_decode($serverret["body"]);
 +                      if (isset($data->totalResults)) {
 +                              $poco = $server_url."/poco";
 +                              $last_contact = datetime_convert();
 +
 +                              $server = poco_detect_poco_data($data);
 +                              if ($server) {
 +                                      $platform = $server['platform'];
 +                                      $network = $server['network'];
 +                                      $version = '';
 +                                      $site_name = '';
 +                              }
 +                      }
 +              }
 +      }
 +
 +      if (!$failure) {
 +              // Test for Diaspora, Hubzilla, Mastodon or older Friendica servers
                $serverret = z_fetch_url($server_url);
  
 -              if (!$serverret["success"] OR ($serverret["body"] == ""))
 +              if (!$serverret["success"] OR ($serverret["body"] == "")) {
 +                      $last_failure = datetime_convert();
                        $failure = true;
 -              else {
 +              } else {
 +                      $server = poco_detect_server_type($serverret["body"]);
 +                      if ($server) {
 +                              $platform = $server['platform'];
 +                              $network = $server['network'];
 +                              $version = $server['version'];
 +                              $site_name = $server['site_name'];
 +                              $last_contact = datetime_convert();
 +                      }
 +
                        $lines = explode("\n",$serverret["header"]);
 -                      if(count($lines))
 +                      if(count($lines)) {
                                foreach($lines as $line) {
                                        $line = trim($line);
                                        if(stristr($line,'X-Diaspora-Version:')) {
                                                $network = NETWORK_DIASPORA;
                                                $versionparts = explode("-", $version);
                                                $version = $versionparts[0];
 +                                              $last_contact = datetime_convert();
                                        }
  
                                        if(stristr($line,'Server: Mastodon')) {
                                                $network = NETWORK_OSTATUS;
                                                // Mastodon doesn't reveal version numbers
                                                $version = "";
 +                                              $last_contact = datetime_convert();
                                        }
                                }
 +                      }
                }
        }
  
 -      if (!$failure) {
 +      if (!$failure AND ($poco == "")) {
                // Test for Statusnet
                // Will also return data for Friendica and GNU Social - but it will be overwritten later
                // The "not implemented" is a special treatment for really, really old Friendica versions
                if ($serverret["success"] AND ($serverret["body"] != '{"error":"not implemented"}') AND
                        ($serverret["body"] != '') AND (strlen($serverret["body"]) < 30)) {
                        $platform = "StatusNet";
 -                      $version = trim($serverret["body"], '"');
 +                      // Remove junk that some GNU Social servers return
 +                      $version = str_replace(chr(239).chr(187).chr(191), "", $serverret["body"]);
 +                      $version = trim($version, '"');
                        $network = NETWORK_OSTATUS;
 +                      $last_contact = datetime_convert();
                }
  
                // Test for GNU Social
                if ($serverret["success"] AND ($serverret["body"] != '{"error":"not implemented"}') AND
                        ($serverret["body"] != '') AND (strlen($serverret["body"]) < 30)) {
                        $platform = "GNU Social";
 -                      $version = trim($serverret["body"], '"');
 +                      // Remove junk that some GNU Social servers return
 +                      $version = str_replace(chr(239).chr(187).chr(191), "", $serverret["body"]);
 +                      $version = trim($version, '"');
                        $network = NETWORK_OSTATUS;
 +                      $last_contact = datetime_convert();
                }
 +      }
  
 +      if (!$failure) {
 +              // Test for Hubzilla, Redmatrix or Friendica
                $serverret = z_fetch_url($server_url."/api/statusnet/config.json");
                if ($serverret["success"]) {
                        $data = json_decode($serverret["body"]);
 -
                        if (isset($data->site->server)) {
                                $last_contact = datetime_convert();
  
 +                              if (isset($data->site->platform)) {
 +                                      $platform = $data->site->platform->PLATFORM_NAME;
 +                                      $version = $data->site->platform->STD_VERSION;
 +                                      $network = NETWORK_DIASPORA;
 +                              }
 +                              if (isset($data->site->BlaBlaNet)) {
 +                                      $platform = $data->site->BlaBlaNet->PLATFORM_NAME;
 +                                      $version = $data->site->BlaBlaNet->STD_VERSION;
 +                                      $network = NETWORK_DIASPORA;
 +                              }
                                if (isset($data->site->hubzilla)) {
                                        $platform = $data->site->hubzilla->PLATFORM_NAME;
                                        $version = $data->site->hubzilla->RED_VERSION;
                }
        }
  
 +
        // Query statistics.json. Optional package for Diaspora, Friendica and Redmatrix
        if (!$failure) {
                $serverret = z_fetch_url($server_url."/statistics.json");
                }
        }
  
 -      // Look for poco
 -      if (!$failure) {
 -              $serverret = z_fetch_url($server_url."/poco");
 -              if ($serverret["success"]) {
 -                      $data = json_decode($serverret["body"]);
 -                      if (isset($data->totalResults)) {
 -                              $poco = $server_url."/poco";
 -                              $last_contact = datetime_convert();
 -                      }
 -              }
 +      if ($possible_failure AND !$failure) {
 +              $last_failure = datetime_convert();
 +              $failure = true;
 +      }
 +
 +      if ($failure) {
 +              $last_contact = $orig_last_contact;
 +      } else {
 +              $last_failure = $orig_last_failure;
 +      }
 +
 +      if (($last_contact <= $last_failure) AND !$failure) {
 +              logger("Server ".$server_url." seems to be alive, but last contact wasn't set - could be a bug", LOGGER_DEBUG);
 +      } else if (($last_contact >= $last_failure) AND $failure) {
 +              logger("Server ".$server_url." seems to be dead, but last failure wasn't set - could be a bug", LOGGER_DEBUG);
        }
  
        // Check again if the server exists
        $info = strip_tags($info);
        $platform = strip_tags($platform);
  
 -      if ($servers)
 +      if ($servers) {
                 q("UPDATE `gserver` SET `url` = '%s', `version` = '%s', `site_name` = '%s', `info` = '%s', `register_policy` = %d, `poco` = '%s', `noscrape` = '%s',
                        `network` = '%s', `platform` = '%s', `last_contact` = '%s', `last_failure` = '%s' WHERE `nurl` = '%s'",
                        dbesc($server_url),
                        dbesc($last_failure),
                        dbesc(normalise_link($server_url))
                );
 -      else
 +      } elseif (!$failure)
                q("INSERT INTO `gserver` (`url`, `nurl`, `version`, `site_name`, `info`, `register_policy`, `poco`, `noscrape`, `network`, `platform`, `created`, `last_contact`, `last_failure`)
                                        VALUES ('%s', '%s', '%s', '%s', '%s', %d, '%s', '%s', '%s', '%s', '%s', '%s', '%s')",
                                dbesc($server_url),
@@@ -1441,33 -1244,6 +1460,33 @@@ function update_suggestions() 
        }
  }
  
 +/**
 + * @brief Fetch server list from remote servers and adds them when they are new.
 + *
 + * @param string $poco URL to the POCO endpoint
 + */
 +function poco_fetch_serverlist($poco) {
 +      $serverret = z_fetch_url($poco."/@server");
 +      if (!$serverret["success"]) {
 +              return;
 +      }
 +      $serverlist = json_decode($serverret['body']);
 +
 +      if (!is_array($serverlist)) {
 +              return;
 +      }
 +
 +      foreach ($serverlist AS $server) {
 +              $server_url = str_replace("/index.php", "", $server->url);
 +
 +              $r = q("SELECT `nurl` FROM `gserver` WHERE `nurl` = '%s'", dbesc(normalise_link($server_url)));
 +              if (!dbm::is_result($r)) {
 +                      logger("Call server check for server ".$server_url, LOGGER_DEBUG);
 +                      proc_run(PRIORITY_LOW, "include/discover_poco.php", "server", base64_encode($server_url));
 +              }
 +      }
 +}
 +
  function poco_discover_federation() {
        $last = get_config('poco','last_federation_discovery');
  
        if ($serverdata) {
                $servers = json_decode($serverdata);
  
 -              foreach($servers->pods AS $server)
 -                      poco_check_server("https://".$server->host);
 +              foreach ($servers->pods AS $server) {
 +                      proc_run(PRIORITY_LOW, "include/discover_poco.php", "server", base64_encode("https://".$server->host));
 +              }
        }
  
        // Currently disabled, since the service isn't available anymore.
@@@ -1530,9 -1305,6 +1549,9 @@@ function poco_discover($complete = fals
                                continue;
                        }
  
 +                      // Discover new servers out there
 +                      poco_fetch_serverlist($server["poco"]);
 +
                        // Fetch all users from the other server
                        $url = $server["poco"]."/?fields=displayName,urls,photos,updated,network,aboutMe,currentLocation,tags,gender,contactType,generation";
  
@@@ -2107,20 -1879,4 +2126,20 @@@ function gs_discover() 
                q("UPDATE `gserver` SET `last_poco_query` = '%s' WHERE `nurl` = '%s'", dbesc(datetime_convert()), dbesc($server["nurl"]));
        }
  }
 +
 +/**
 + * @brief Returns a list of all known servers
 + * @return array List of server urls
 + */
 +function poco_serverlist() {
 +      $r = q("SELECT `url`, `site_name` AS `displayName`, `network`, `platform`, `version` FROM `gserver`
 +              WHERE `network` IN ('%s', '%s', '%s') AND `last_contact` > `last_failure`
 +              ORDER BY `last_contact`
 +              LIMIT 1000",
 +              dbesc(NETWORK_DFRN), dbesc(NETWORK_DIASPORA), dbesc(NETWORK_OSTATUS));
 +      if (!dbm::is_result($r)) {
 +              return false;
 +      }
 +      return $r;
 +}
  ?>