]> git.mxchange.org Git - friendica.git/blobdiff - src/Network/Probe.php
Use DateTimeFormat::utc() instead of DBM::date()
[friendica.git] / src / Network / Probe.php
index f7c3b84fdd46803c35eb334a505366189595004f..71d36ea8ff57218a32a55d9e81be0c9ce4c3b109 100644 (file)
@@ -9,21 +9,20 @@ namespace Friendica\Network;
  * @brief Functions for probing URL
  */
 
-use Friendica\App;
-use Friendica\Core\System;
+use DOMDocument;
 use Friendica\Core\Cache;
 use Friendica\Core\Config;
+use Friendica\Core\System;
+use Friendica\Database\DBA;
 use Friendica\Database\DBM;
 use Friendica\Model\Profile;
 use Friendica\Protocol\Email;
 use Friendica\Protocol\Feed;
 use Friendica\Util\Crypto;
+use Friendica\Util\DateTimeFormat;
 use Friendica\Util\Network;
 use Friendica\Util\XML;
-
-use dba;
-use DOMXPath;
-use DOMDocument;
+use DomXPath;
 
 require_once 'include/dba.php';
 
@@ -90,6 +89,9 @@ class Probe
        /**
         * @brief Probes for webfinger path via "host-meta"
         *
+        * We have to check if the servers in the future still will offer this.
+        * It seems as if it was dropped from the standard.
+        *
         * @param string $host The host part of an url
         *
         * @return array with template and type of the webfinger template for JSON or XML
@@ -106,6 +108,7 @@ class Probe
                $redirects = 0;
 
                logger("Probing for ".$host, LOGGER_DEBUG);
+               $xrd = null;
 
                $ret = Network::curl($ssl_url, false, $redirects, ['timeout' => $xrd_timeout, 'accept_content' => 'application/xrd+xml']);
                if ($ret['success']) {
@@ -116,7 +119,7 @@ class Probe
 
                if (!is_object($xrd)) {
                        $ret = Network::curl($url, false, $redirects, ['timeout' => $xrd_timeout, 'accept_content' => 'application/xrd+xml']);
-                       if ($ret['errno'] == CURLE_OPERATION_TIMEDOUT) {
+                       if (!empty($ret["errno"]) && ($ret['errno'] == CURLE_OPERATION_TIMEDOUT)) {
                                logger("Probing timeout for ".$url, LOGGER_DEBUG);
                                return false;
                        }
@@ -331,7 +334,7 @@ class Probe
                }
 
                if (x($data, "photo")) {
-                       $data["baseurl"] = Network::getUrlMatch(normalise_link($data["baseurl"]), normalise_link($data["photo"]));
+                       $data["baseurl"] = Network::getUrlMatch(normalise_link(defaults($data, "baseurl", "")), normalise_link($data["photo"]));
                } else {
                        $data["photo"] = System::baseUrl().'/images/person-175.jpg';
                }
@@ -354,11 +357,11 @@ class Probe
                        }
                }
 
-               if (self::$baseurl != "") {
+               if (!empty(self::$baseurl)) {
                        $data["baseurl"] = self::$baseurl;
                }
 
-               if (!isset($data["network"])) {
+               if (empty($data["network"])) {
                        $data["network"] = NETWORK_PHANTOM;
                }
 
@@ -392,6 +395,12 @@ class Probe
                                                'network' => $data['network'],
                                                'server_url' => $data['baseurl']];
 
+                               // This doesn't cover the case when a community isn't a community anymore
+                               if (!empty($data['community']) && $data['community']) {
+                                       $fields['community'] = $data['community'];
+                                       $fields['contact-type'] = ACCOUNT_TYPE_COMMUNITY;
+                               }
+
                                $fieldnames = [];
 
                                foreach ($fields as $key => $val) {
@@ -402,13 +411,24 @@ class Probe
                                        }
                                }
 
-                               $fields['updated'] = DBM::date();
+                               $fields['updated'] = DateTimeFormat::utcNow();
 
                                $condition = ['nurl' => normalise_link($data["url"])];
 
-                               $old_fields = dba::selectFirst('gcontact', $fieldnames, $condition);
+                               $old_fields = DBA::selectFirst('gcontact', $fieldnames, $condition);
+
+                               // When the gcontact doesn't exist, the value "true" will trigger an insert.
+                               // In difference to the public contacts we want to have every contact
+                               // in the world in our global contacts.
+                               if (!$old_fields) {
+                                       $old_fields = true;
+
+                                       // These values have to be set only on insert
+                                       $fields['photo'] = $data['photo'];
+                                       $fields['created'] = DateTimeFormat::utcNow();
+                               }
 
-                               dba::update('gcontact', $fields, $condition, $old_fields);
+                               DBA::update('gcontact', $fields, $condition, $old_fields);
 
                                $fields = ['name' => $data['name'],
                                                'nick' => $data['nick'],
@@ -425,7 +445,10 @@ class Probe
                                                'confirm' => $data['confirm'],
                                                'poco' => $data['poco'],
                                                'network' => $data['network'],
-                                               'success_update' => DBM::date()];
+                                               'pubkey' => $data['pubkey'],
+                                               'priority' => $data['priority'],
+                                               'writable' => true,
+                                               'rel' => CONTACT_IS_SHARING];
 
                                $fieldnames = [];
 
@@ -439,9 +462,17 @@ class Probe
 
                                $condition = ['nurl' => normalise_link($data["url"]), 'self' => false, 'uid' => 0];
 
-                               $old_fields = dba::selectFirst('contact', $fieldnames, $condition);
+                               // "$old_fields" will return a "false" when the contact doesn't exist.
+                               // This won't trigger an insert. This is intended, since we only need
+                               // public contacts for everyone we store items from.
+                               // We don't need to store every contact on the planet.
+                               $old_fields = DBA::selectFirst('contact', $fieldnames, $condition);
 
-                               dba::update('contact', $fields, $condition, $old_fields);
+                               $fields['name-date'] = DateTimeFormat::utcNow();
+                               $fields['uri-date'] = DateTimeFormat::utcNow();
+                               $fields['success_update'] = DateTimeFormat::utcNow();
+
+                               DBA::update('contact', $fields, $condition, $old_fields);
                        }
                }
 
@@ -649,7 +680,7 @@ class Probe
                        $result = self::ostatus($webfinger);
                }
                if ((!$result && ($network == "")) || ($network == NETWORK_PUMPIO)) {
-                       $result = self::pumpio($webfinger);
+                       $result = self::pumpio($webfinger, $addr);
                }
                if ((!$result && ($network == "")) || ($network == NETWORK_FEED)) {
                        $result = self::feed($uri);
@@ -665,6 +696,14 @@ class Probe
                        }
                }
 
+               if (empty($result["network"])) {
+                       $result["network"] = NETWORK_PHANTOM;
+               }
+
+               if (empty($result["url"])) {
+                       $result["url"] = $uri;
+               }
+
                logger($uri." is ".$result["network"], LOGGER_DEBUG);
 
                if (empty($result["baseurl"])) {
@@ -673,7 +712,6 @@ class Probe
                                $result["baseurl"] = substr($result["url"], 0, $pos).$host;
                        }
                }
-
                return $result;
        }
 
@@ -693,7 +731,7 @@ class Probe
                $redirects = 0;
 
                $ret = Network::curl($url, false, $redirects, ['timeout' => $xrd_timeout, 'accept_content' => $type]);
-               if ($ret['errno'] == CURLE_OPERATION_TIMEDOUT) {
+               if (!empty($ret["errno"]) && ($ret['errno'] == CURLE_OPERATION_TIMEDOUT)) {
                        return false;
                }
                $data = $ret['body'];
@@ -760,7 +798,7 @@ class Probe
        private static function pollNoscrape($noscrape_url, $data)
        {
                $ret = Network::curl($noscrape_url);
-               if ($ret['errno'] == CURLE_OPERATION_TIMEDOUT) {
+               if (!empty($ret["errno"]) && ($ret['errno'] == CURLE_OPERATION_TIMEDOUT)) {
                        return false;
                }
                $content = $ret['body'];
@@ -949,7 +987,7 @@ class Probe
                        }
                }
 
-               if (is_array($webfinger["aliases"])) {
+               if (!empty($webfinger["aliases"]) && is_array($webfinger["aliases"])) {
                        foreach ($webfinger["aliases"] as $alias) {
                                if (normalise_link($alias) != normalise_link($data["url"]) && ! strstr($alias, "@")) {
                                        $data["alias"] = $alias;
@@ -959,6 +997,10 @@ class Probe
                        }
                }
 
+               if (!empty($webfinger["subject"]) && (substr($webfinger["subject"], 0, 5) == "acct:")) {
+                       $data["addr"] = substr($webfinger["subject"], 5);
+               }
+
                if (!isset($data["network"]) || ($hcard_url == "")) {
                        return false;
                }
@@ -994,7 +1036,7 @@ class Probe
        private static function pollHcard($hcard_url, $data, $dfrn = false)
        {
                $ret = Network::curl($hcard_url);
-               if ($ret['errno'] == CURLE_OPERATION_TIMEDOUT) {
+               if (!empty($ret["errno"]) && ($ret['errno'] == CURLE_OPERATION_TIMEDOUT)) {
                        return false;
                }
                $content = $ret['body'];
@@ -1014,13 +1056,17 @@ class Probe
                        return false;
                }
 
+               if (!isset($data["baseurl"])) {
+                       $data["baseurl"] = "";
+               }
+
                if ($vcards->length > 0) {
                        $vcard = $vcards->item(0);
 
                        // We have to discard the guid from the hcard in favour of the guid from lrdd
                        // Reason: Hubzilla doesn't use the value "uid" in the hcard like Diaspora does.
                        $search = $xpath->query("//*[contains(concat(' ', @class, ' '), ' uid ')]", $vcard); // */
-                       if (($search->length > 0) && ($data["guid"] == "")) {
+                       if (($search->length > 0) && empty($data["guid"])) {
                                $data["guid"] = $search->item(0)->nodeValue;
                        }
 
@@ -1054,21 +1100,23 @@ class Probe
                }
 
                $avatar = [];
-               $photos = $xpath->query("//*[contains(concat(' ', @class, ' '), ' photo ') or contains(concat(' ', @class, ' '), ' avatar ')]", $vcard); // */
-               foreach ($photos as $photo) {
-                       $attr = [];
-                       foreach ($photo->attributes as $attribute) {
-                               $attr[$attribute->name] = trim($attribute->value);
-                       }
+               if (!empty($vcard)) {
+                       $photos = $xpath->query("//*[contains(concat(' ', @class, ' '), ' photo ') or contains(concat(' ', @class, ' '), ' avatar ')]", $vcard); // */
+                       foreach ($photos as $photo) {
+                               $attr = [];
+                               foreach ($photo->attributes as $attribute) {
+                                       $attr[$attribute->name] = trim($attribute->value);
+                               }
 
-                       if (isset($attr["src"]) && isset($attr["width"])) {
-                               $avatar[$attr["width"]] = $attr["src"];
-                       }
+                               if (isset($attr["src"]) && isset($attr["width"])) {
+                                       $avatar[$attr["width"]] = $attr["src"];
+                               }
 
-                       // We don't have a width. So we just take everything that we got.
-                       // This is a Hubzilla workaround which doesn't send a width.
-                       if ((sizeof($avatar) == 0) && !empty($attr["src"])) {
-                               $avatar[] = $attr["src"];
+                               // We don't have a width. So we just take everything that we got.
+                               // This is a Hubzilla workaround which doesn't send a width.
+                               if ((sizeof($avatar) == 0) && !empty($attr["src"])) {
+                                       $avatar[] = $attr["src"];
+                               }
                        }
                }
 
@@ -1093,7 +1141,7 @@ class Probe
                        }
 
                        // Older Friendica versions had used the "uid" field differently than newer versions
-                       if ($data["nick"] == $data["guid"]) {
+                       if (!empty($data["nick"]) && !empty($data["guid"]) && ($data["nick"] == $data["guid"])) {
                                unset($data["guid"]);
                        }
                }
@@ -1142,7 +1190,7 @@ class Probe
                        return false;
                }
 
-               if (is_array($webfinger["aliases"])) {
+               if (!empty($webfinger["aliases"]) && is_array($webfinger["aliases"])) {
                        foreach ($webfinger["aliases"] as $alias) {
                                if (normalise_link($alias) != normalise_link($data["url"]) && ! strstr($alias, "@")) {
                                        $data["alias"] = $alias;
@@ -1172,7 +1220,9 @@ class Probe
                        $data["network"] = NETWORK_DIASPORA;
 
                        // The Diaspora handle must always be lowercase
-                       $data["addr"] = strtolower($data["addr"]);
+                       if (!empty($data["addr"])) {
+                               $data["addr"] = strtolower($data["addr"]);
+                       }
 
                        // We have to overwrite the detected value for "notify" since Hubzilla doesn't send it
                        $data["notify"] = $data["baseurl"] . "/receive/users/" . $data["guid"];
@@ -1196,7 +1246,7 @@ class Probe
        {
                $data = [];
 
-               if (is_array($webfinger["aliases"])) {
+               if (!empty($webfinger["aliases"]) && is_array($webfinger["aliases"])) {
                        foreach ($webfinger["aliases"] as $alias) {
                                if (strstr($alias, "@") && !strstr(normalise_link($alias), "http://")) {
                                        $data["addr"] = str_replace('acct:', '', $alias);
@@ -1204,7 +1254,7 @@ class Probe
                        }
                }
 
-               if (is_string($webfinger["subject"]) && strstr($webfinger["subject"], "@")
+               if (!empty($webfinger["subject"]) && strstr($webfinger["subject"], "@")
                        && !strstr(normalise_link($webfinger["subject"]), "http://")
                ) {
                        $data["addr"] = str_replace('acct:', '', $webfinger["subject"]);
@@ -1233,7 +1283,7 @@ class Probe
                                                }
                                        } elseif (normalise_link($pubkey) == 'http://') {
                                                $ret = Network::curl($pubkey);
-                                               if ($ret['errno'] == CURLE_OPERATION_TIMEDOUT) {
+                                               if (!empty($ret["errno"]) && ($ret['errno'] == CURLE_OPERATION_TIMEDOUT)) {
                                                        return false;
                                                }
                                                $pubkey = $ret['body'];
@@ -1265,36 +1315,39 @@ class Probe
 
                // Fetch all additional data from the feed
                $ret = Network::curl($data["poll"]);
-               if ($ret['errno'] == CURLE_OPERATION_TIMEDOUT) {
+               if (!empty($ret["errno"]) && ($ret['errno'] == CURLE_OPERATION_TIMEDOUT)) {
                        return false;
                }
                $feed = $ret['body'];
+               $dummy1 = null;
+               $dummy2 = null;
+               $dummy2 = null;
                $feed_data = Feed::import($feed, $dummy1, $dummy2, $dummy3, true);
                if (!$feed_data) {
                        return false;
                }
 
-               if ($feed_data["header"]["author-name"] != "") {
+               if (!empty($feed_data["header"]["author-name"])) {
                        $data["name"] = $feed_data["header"]["author-name"];
                }
-               if ($feed_data["header"]["author-nick"] != "") {
+               if (!empty($feed_data["header"]["author-nick"])) {
                        $data["nick"] = $feed_data["header"]["author-nick"];
                }
-               if ($feed_data["header"]["author-avatar"] != "") {
+               if (!empty($feed_data["header"]["author-avatar"])) {
                        $data["photo"] = self::fixAvatar($feed_data["header"]["author-avatar"], $data["url"]);
                }
-               if ($feed_data["header"]["author-id"] != "") {
+               if (!empty($feed_data["header"]["author-id"])) {
                        $data["alias"] = $feed_data["header"]["author-id"];
                }
-               if ($feed_data["header"]["author-location"] != "") {
+               if (!empty($feed_data["header"]["author-location"])) {
                        $data["location"] = $feed_data["header"]["author-location"];
                }
-               if ($feed_data["header"]["author-about"] != "") {
+               if (!empty($feed_data["header"]["author-about"])) {
                        $data["about"] = $feed_data["header"]["author-about"];
                }
                // OStatus has serious issues when the the url doesn't fit (ssl vs. non ssl)
                // So we take the value that we just fetched, although the other one worked as well
-               if ($feed_data["header"]["author-link"] != "") {
+               if (!empty($feed_data["header"]["author-link"])) {
                        $data["url"] = $feed_data["header"]["author-link"];
                }
 
@@ -1325,14 +1378,33 @@ class Probe
 
                $data = [];
 
-               // This is ugly - but pump.io doesn't seem to know a better way for it
-               $data["name"] = trim($xpath->query("//h1[@class='media-header']")->item(0)->nodeValue);
-               $pos = strpos($data["name"], chr(10));
-               if ($pos) {
-                       $data["name"] = trim(substr($data["name"], 0, $pos));
+               $data["name"] = $xpath->query("//span[contains(@class, 'p-name')]")->item(0)->nodeValue;
+
+               if ($data["name"] == '') {
+                       // This is ugly - but pump.io doesn't seem to know a better way for it
+                       $data["name"] = trim($xpath->query("//h1[@class='media-header']")->item(0)->nodeValue);
+                       $pos = strpos($data["name"], chr(10));
+                       if ($pos) {
+                               $data["name"] = trim(substr($data["name"], 0, $pos));
+                       }
+               }
+
+               $data["location"] = XML::getFirstNodeValue($xpath, "//p[contains(@class, 'p-locality')]");
+
+               if ($data["location"] == '') {
+                       $data["location"] = XML::getFirstNodeValue($xpath, "//p[contains(@class, 'location')]");
+               }
+
+               $data["about"] = XML::getFirstNodeValue($xpath, "//p[contains(@class, 'p-note')]");
+
+               if ($data["about"] == '') {
+                       $data["about"] = XML::getFirstNodeValue($xpath, "//p[contains(@class, 'summary')]");
                }
 
-               $avatar = $xpath->query("//img[@class='img-rounded media-object']")->item(0);
+               $avatar = $xpath->query("//img[contains(@class, 'u-photo')]")->item(0);
+               if (!$avatar) {
+                       $avatar = $xpath->query("//img[@class='img-rounded media-object']")->item(0);
+               }
                if ($avatar) {
                        foreach ($avatar->attributes as $attribute) {
                                if ($attribute->name == "src") {
@@ -1341,9 +1413,6 @@ class Probe
                        }
                }
 
-               $data["location"] = $xpath->query("//p[@class='location']")->item(0)->nodeValue;
-               $data["about"] = $xpath->query("//p[@class='summary']")->item(0)->nodeValue;
-
                return $data;
        }
 
@@ -1354,7 +1423,7 @@ class Probe
         *
         * @return array pump.io data
         */
-       private static function pumpio($webfinger)
+       private static function pumpio($webfinger, $addr)
        {
                $data = [];
                foreach ($webfinger["links"] as $link) {
@@ -1392,6 +1461,13 @@ class Probe
 
                $data = array_merge($data, $profile_data);
 
+               if (($addr != '') && ($data['name'] != '')) {
+                       $name = trim(str_replace($addr, '', $data['name']));
+                       if ($name != '') {
+                               $data['name'] = $name;
+                       }
+               }
+
                return $data;
        }
 
@@ -1449,10 +1525,11 @@ class Probe
        private static function feed($url, $probe = true)
        {
                $ret = Network::curl($url);
-               if ($ret['errno'] == CURLE_OPERATION_TIMEDOUT) {
+               if (!empty($ret["errno"]) && ($ret['errno'] == CURLE_OPERATION_TIMEDOUT)) {
                        return false;
                }
                $feed = $ret['body'];
+               $dummy1 = $dummy2 = $dummy3 = null;
                $feed_data = Feed::import($feed, $dummy1, $dummy2, $dummy3, true);
 
                if (!$feed_data) {
@@ -1469,26 +1546,26 @@ class Probe
                        return self::feed($feed_url, false);
                }
 
-               if ($feed_data["header"]["author-name"] != "") {
+               if (!empty($feed_data["header"]["author-name"])) {
                        $data["name"] = $feed_data["header"]["author-name"];
                }
 
-               if ($feed_data["header"]["author-nick"] != "") {
+               if (!empty($feed_data["header"]["author-nick"])) {
                        $data["nick"] = $feed_data["header"]["author-nick"];
                }
 
-               if ($feed_data["header"]["author-avatar"] != "") {
+               if (!empty($feed_data["header"]["author-avatar"])) {
                        $data["photo"] = $feed_data["header"]["author-avatar"];
                }
 
-               if ($feed_data["header"]["author-id"] != "") {
+               if (!empty($feed_data["header"]["author-id"])) {
                        $data["alias"] = $feed_data["header"]["author-id"];
                }
 
                $data["url"] = $url;
                $data["poll"] = $url;
 
-               if ($feed_data["header"]["author-link"] != "") {
+               if (!empty($feed_data["header"]["author-link"])) {
                        $data["baseurl"] = $feed_data["header"]["author-link"];
                } else {
                        $data["baseurl"] = $data["url"];
@@ -1513,29 +1590,31 @@ class Probe
                        return false;
                }
 
-               if ($uid != 0) {
-                       $x = q("SELECT `prvkey` FROM `user` WHERE `uid` = %d LIMIT 1", intval($uid));
-
-                       $r = q("SELECT * FROM `mailacct` WHERE `uid` = %d AND `server` != '' LIMIT 1", intval($uid));
+               if ($uid == 0) {
+                       return false;
+               }
 
-                       if (DBM::is_result($x) && DBM::is_result($r)) {
-                               $mailbox = Email::constructMailboxName($r[0]);
-                               $password = '';
-                               openssl_private_decrypt(hex2bin($r[0]['pass']), $password, $x[0]['prvkey']);
-                               $mbox = Email::connect($mailbox, $r[0]['user'], $password);
-                               if (!mbox) {
-                                       return false;
-                               }
-                       }
+               $x = q("SELECT `prvkey` FROM `user` WHERE `uid` = %d LIMIT 1", intval($uid));
 
-                       $msgs = Email::poll($mbox, $uri);
-                       logger('searching '.$uri.', '.count($msgs).' messages found.', LOGGER_DEBUG);
+               $r = q("SELECT * FROM `mailacct` WHERE `uid` = %d AND `server` != '' LIMIT 1", intval($uid));
 
-                       if (!count($msgs)) {
+               if (DBM::is_result($x) && DBM::is_result($r)) {
+                       $mailbox = Email::constructMailboxName($r[0]);
+                       $password = '';
+                       openssl_private_decrypt(hex2bin($r[0]['pass']), $password, $x[0]['prvkey']);
+                       $mbox = Email::connect($mailbox, $r[0]['user'], $password);
+                       if (!$mbox) {
                                return false;
                        }
                }
 
+               $msgs = Email::poll($mbox, $uri);
+               logger('searching '.$uri.', '.count($msgs).' messages found.', LOGGER_DEBUG);
+
+               if (!count($msgs)) {
+                       return false;
+               }
+
                $phost = substr($uri, strpos($uri, '@') + 1);
 
                $data = [];