]> git.mxchange.org Git - friendica.git/blobdiff - src/Network/Probe.php
Use webfinger in apcontact fetch
[friendica.git] / src / Network / Probe.php
index 1e6d8406a12697b58bc181708705ede0d4310fbe..db4ef517d54a7270b6423e50a50539cd2496601e 100644 (file)
@@ -30,6 +30,7 @@ use Friendica\Core\System;
 use Friendica\Database\DBA;
 use Friendica\DI;
 use Friendica\Model\Contact;
+use Friendica\Model\GServer;
 use Friendica\Model\Profile;
 use Friendica\Protocol\ActivityNamespace;
 use Friendica\Protocol\ActivityPub;
@@ -84,16 +85,18 @@ class Probe
        {
                $fields = ["name", "nick", "guid", "url", "addr", "alias", "photo", "account-type",
                                "community", "keywords", "location", "about", "hide",
-                               "batch", "notify", "poll", "request", "confirm", "poco",
+                               "batch", "notify", "poll", "request", "confirm", "subscribe", "poco",
                                "following", "followers", "inbox", "outbox", "sharedinbox",
-                               "priority", "network", "pubkey", "baseurl"];
+                               "priority", "network", "pubkey", "baseurl", "gsid"];
 
                $newdata = [];
                foreach ($fields as $field) {
                        if (isset($data[$field])) {
                                $newdata[$field] = $data[$field];
-                       } else {
+                       } elseif ($field != "gsid") {
                                $newdata[$field] = "";
+                       } else {
+                               $newdata[$field] = null;
                        }
                }
 
@@ -169,7 +172,7 @@ class Probe
                } elseif ($curlResult->isTimeout()) {
                        Logger::info('Probing timeout', ['url' => $ssl_url], Logger::DEBUG);
                        self::$istimeout = true;
-                       return false;
+                       return [];
                }
 
                if (!is_object($xrd) && !empty($url)) {
@@ -178,10 +181,10 @@ class Probe
                        if ($curlResult->isTimeout()) {
                                Logger::info('Probing timeout', ['url' => $url], Logger::DEBUG);
                                self::$istimeout = true;
-                               return false;
+                               return [];
                        } elseif ($connection_error && $ssl_connection_error) {
                                self::$istimeout = true;
-                               return false;
+                               return [];
                        }
 
                        $xml = $curlResult->getBody();
@@ -267,37 +270,15 @@ class Probe
                return $profile_link;
        }
 
-       /**
-        * Get the link for the remote follow page for a given profile link
-        *
-        * @param sting $profile
-        * @return string Remote follow page link
-        */
-       public static function getRemoteFollowLink(string $profile)
-       {
-               $follow_link = '';
-
-               $links = self::lrdd($profile);
-
-               if (!empty($links) && is_array($links)) {
-                       foreach ($links as $link) {
-                               if ($link['@attributes']['rel'] === ActivityNamespace::OSTATUSSUB) {
-                                       $follow_link = $link['@attributes']['template'];
-                               }
-                       }
-               }
-               return $follow_link;
-       }
-
        /**
         * Check an URI for LRDD data
         *
-        * @param string $uri Address that should be probed
+        * @param string $uri     Address that should be probed
         *
         * @return array uri data
         * @throws HTTPException\InternalServerErrorException
         */
-       public static function lrdd($uri)
+       public static function lrdd(string $uri)
        {
                $lrdd = self::hostMeta($uri);
                $webfinger = null;
@@ -359,9 +340,9 @@ class Probe
                        }
                }
 
-               if (!is_array($webfinger["links"])) {
+               if (empty($webfinger["links"])) {
                        Logger::log("No webfinger links found for ".$uri, Logger::DEBUG);
-                       return false;
+                       return [];
                }
 
                $data = [];
@@ -370,7 +351,7 @@ class Probe
                        $data[] = ["@attributes" => $link];
                }
 
-               if (is_array($webfinger["aliases"])) {
+               if (!empty($webfinger["aliases"]) && is_array($webfinger["aliases"])) {
                        foreach ($webfinger["aliases"] as $alias) {
                                $data[] = ["@attributes" =>
                                                        ["rel" => "alias",
@@ -433,9 +414,7 @@ class Probe
                        $data['url'] = $uri;
                }
 
-               if (!empty($data['photo']) && !empty($data['baseurl'])) {
-                       $data['baseurl'] = Network::getUrlMatch(Strings::normaliseLink($data['baseurl']), Strings::normaliseLink($data['photo']));
-               } elseif (empty($data['photo'])) {
+               if (empty($data['photo'])) {
                        $data['photo'] = DI::baseUrl() . '/images/person-300.jpg';
                }
 
@@ -461,6 +440,10 @@ class Probe
                        $data['baseurl'] = self::$baseurl;
                }
 
+               if (!empty($data['baseurl']) && empty($data['gsid'])) {
+                       $data['gsid'] = GServer::getID($data['baseurl']);
+               }
+
                if (empty($data['network'])) {
                        $data['network'] = Protocol::PHANTOM;
                }
@@ -592,6 +575,28 @@ class Probe
                return $webfinger;
        }
 
+       /**
+        * Fetch the "subscribe" and add it to the result
+        *
+        * @param array $result
+        * @param array $webfinger
+        * @return array result
+        */
+       private static function getSubscribeLink(array $result, array $webfinger)
+       {
+               if (empty($webfinger['links'])) {
+                       return $result;
+               }
+
+               foreach ($webfinger['links'] as $link) {
+                       if ($link['rel'] === ActivityNamespace::OSTATUSSUB) {
+                               $result['subscribe'] = $link['template'];
+                       }
+               }
+
+               return $result;
+       }
+
        /**
         * Fetch information (protocol endpoints and user information) about a given uri
         *
@@ -671,7 +676,7 @@ class Probe
                        $addr = $uri;
                } else {
                        Logger::log("Uri ".$uri." was not detectable", Logger::DEBUG);
-                       return false;
+                       return [];
                }
 
                $webfinger = false;
@@ -713,9 +718,9 @@ class Probe
                        return self::feed($uri);
                }
 
-               $result = false;
+               $result = [];
 
-               Logger::log("Probing ".$uri, Logger::DEBUG);
+               Logger::info("Probing", ['uri' => $uri]);
 
                if (in_array($network, ["", Protocol::DFRN])) {
                        $result = self::dfrn($webfinger);
@@ -746,6 +751,8 @@ class Probe
                        }
                }
 
+               $result = self::getSubscribeLink($result, $webfinger);
+
                if (empty($result["network"])) {
                        $result["network"] = Protocol::PHANTOM;
                }
@@ -762,6 +769,7 @@ class Probe
                                $result["baseurl"] = substr($result["url"], 0, $pos).$host;
                        }
                }
+
                return $result;
        }
 
@@ -915,22 +923,22 @@ class Probe
         * @return array webfinger data
         * @throws HTTPException\InternalServerErrorException
         */
-       private static function webfinger($url, $type)
+       public static function webfinger($url, $type)
        {
                $xrd_timeout = DI::config()->get('system', 'xrd_timeout', 20);
 
                $curlResult = Network::curl($url, false, ['timeout' => $xrd_timeout, 'accept_content' => $type]);
                if ($curlResult->isTimeout()) {
                        self::$istimeout = true;
-                       return false;
+                       return [];
                }
                $data = $curlResult->getBody();
 
                $webfinger = json_decode($data, true);
-               if (is_array($webfinger)) {
+               if (!empty($webfinger)) {
                        if (!isset($webfinger["links"])) {
                                Logger::log("No json webfinger links for ".$url, Logger::DEBUG);
-                               return false;
+                               return [];
                        }
                        return $webfinger;
                }
@@ -939,13 +947,13 @@ class Probe
                $xrd = XML::parseString($data, true);
                if (!is_object($xrd)) {
                        Logger::log("No webfinger data retrievable for ".$url, Logger::DEBUG);
-                       return false;
+                       return [];
                }
 
                $xrd_arr = XML::elementToArray($xrd);
                if (!isset($xrd_arr["xrd"]["link"])) {
                        Logger::log("No XML webfinger links for ".$url, Logger::DEBUG);
-                       return false;
+                       return [];
                }
 
                $webfinger = [];
@@ -991,18 +999,18 @@ class Probe
                $curlResult = Network::curl($noscrape_url);
                if ($curlResult->isTimeout()) {
                        self::$istimeout = true;
-                       return false;
+                       return [];
                }
                $content = $curlResult->getBody();
                if (!$content) {
                        Logger::log("Empty body for ".$noscrape_url, Logger::DEBUG);
-                       return false;
+                       return [];
                }
 
                $json = json_decode($content, true);
                if (!is_array($json)) {
                        Logger::log("No json data for ".$noscrape_url, Logger::DEBUG);
-                       return false;
+                       return [];
                }
 
                if (!empty($json["fn"])) {
@@ -1212,7 +1220,7 @@ class Probe
                }
 
                if (!isset($data["network"]) || ($hcard_url == "")) {
-                       return false;
+                       return [];
                }
 
                // Fetch data via noscrape - this is faster
@@ -1249,23 +1257,23 @@ class Probe
                $curlResult = Network::curl($hcard_url);
                if ($curlResult->isTimeout()) {
                        self::$istimeout = true;
-                       return false;
+                       return [];
                }
                $content = $curlResult->getBody();
                if (!$content) {
-                       return false;
+                       return [];
                }
 
                $doc = new DOMDocument();
                if (!@$doc->loadHTML($content)) {
-                       return false;
+                       return [];
                }
 
                $xpath = new DomXPath($doc);
 
                $vcards = $xpath->query("//div[contains(concat(' ', @class, ' '), ' vcard ')]");
                if (!is_object($vcards)) {
-                       return false;
+                       return [];
                }
 
                if (!isset($data["baseurl"])) {
@@ -1403,7 +1411,7 @@ class Probe
                }
 
                if (empty($data["url"]) || empty($hcard_url)) {
-                       return false;
+                       return [];
                }
 
                if (!empty($webfinger["aliases"]) && is_array($webfinger["aliases"])) {
@@ -1424,7 +1432,7 @@ class Probe
                $data = self::pollHcard($hcard_url, $data);
 
                if (!$data) {
-                       return false;
+                       return [];
                }
 
                if (!empty($data["url"])
@@ -1444,7 +1452,7 @@ class Probe
                        $data["notify"] = $data["baseurl"] . "/receive/users/" . $data["guid"];
                        $data["batch"]  = $data["baseurl"] . "/receive/public";
                } else {
-                       return false;
+                       return [];
                }
 
                return $data;
@@ -1477,7 +1485,7 @@ class Probe
                        $data["addr"] = str_replace('acct:', '', $webfinger["subject"]);
                }
 
-               if (is_array($webfinger["links"])) {
+               if (!empty($webfinger["links"])) {
                        // The array is reversed to take into account the order of preference for same-rel links
                        // See: https://tools.ietf.org/html/rfc7033#section-4.4.4
                        foreach (array_reverse($webfinger["links"]) as $link) {
@@ -1503,7 +1511,7 @@ class Probe
                                                $curlResult = Network::curl($pubkey);
                                                if ($curlResult->isTimeout()) {
                                                        self::$istimeout = true;
-                                                       return false;
+                                                       return $short ? false : [];
                                                }
                                                $pubkey = $curlResult->getBody();
                                        }
@@ -1525,7 +1533,7 @@ class Probe
                ) {
                        $data["network"] = Protocol::OSTATUS;
                } else {
-                       return false;
+                       return $short ? false : [];
                }
 
                if ($short) {
@@ -1536,12 +1544,12 @@ class Probe
                $curlResult = Network::curl($data["poll"]);
                if ($curlResult->isTimeout()) {
                        self::$istimeout = true;
-                       return false;
+                       return [];
                }
                $feed = $curlResult->getBody();
                $feed_data = Feed::import($feed);
                if (!$feed_data) {
-                       return false;
+                       return [];
                }
 
                if (!empty($feed_data["header"]["author-name"])) {
@@ -1588,12 +1596,12 @@ class Probe
        {
                $curlResult = Network::curl($profile_link);
                if (!$curlResult->isSuccess()) {
-                       return false;
+                       return [];
                }
 
                $doc = new DOMDocument();
                if (!@$doc->loadHTML($curlResult->getBody())) {
-                       return false;
+                       return [];
                }
 
                $xpath = new DomXPath($doc);
@@ -1674,13 +1682,13 @@ class Probe
 
                        $data["network"] = Protocol::PUMPIO;
                } else {
-                       return false;
+                       return [];
                }
 
                $profile_data = self::pumpioProfileData($data["url"]);
 
                if (!$profile_data) {
-                       return false;
+                       return [];
                }
 
                $data = array_merge($data, $profile_data);
@@ -1756,50 +1764,87 @@ class Probe
        }
 
        /**
-        * Check page for feed link
+        * Checks HTML page for RSS feed link
         *
-        * @param string $url Page link
-        *
-        * @return string feed link
+        * @param string $url  Page link
+        * @param string $body Page body string
+        * @return string|false Feed link or false if body was invalid HTML document
         */
-       private static function getFeedLink($url)
+       public static function getFeedLink(string $url, string $body)
        {
-               $curlResult = Network::curl($url);
-               if (!$curlResult->isSuccess()) {
-                       return false;
-               }
-
                $doc = new DOMDocument();
-               if (!@$doc->loadHTML($curlResult->getBody())) {
+               if (!@$doc->loadHTML($body)) {
                        return false;
                }
 
-               $xpath = new DomXPath($doc);
+               $xpath = new DOMXPath($doc);
 
-               //$feeds = $xpath->query("/html/head/link[@type='application/rss+xml']");
-               $feeds = $xpath->query("/html/head/link[@type='application/rss+xml' and @rel='alternate']");
-               if (!is_object($feeds)) {
-                       return false;
-               }
+               $feedUrl = $xpath->evaluate('string(/html/head/link[@type="application/rss+xml" and @rel="alternate"]/@href)');
 
-               if ($feeds->length == 0) {
-                       return false;
+               $feedUrl = $feedUrl ? self::ensureAbsoluteLinkFromHTMLDoc($feedUrl, $url, $xpath) : '';
+
+               return $feedUrl;
+       }
+
+       /**
+        * Return an absolute URL in the context of a HTML document retrieved from the provided URL.
+        *
+        * Loosely based on RFC 1808
+        *
+        * @see https://tools.ietf.org/html/rfc1808
+        *
+        * @param string   $href  The potential relative href found in the HTML document
+        * @param string   $base  The HTML document URL
+        * @param DOMXPath $xpath The HTML document XPath
+        * @return string
+        */
+       private static function ensureAbsoluteLinkFromHTMLDoc(string $href, string $base, DOMXPath $xpath)
+       {
+               if (filter_var($href, FILTER_VALIDATE_URL)) {
+                       return $href;
                }
 
-               $feed_url = "";
+               $base = $xpath->evaluate('string(/html/head/base/@href)') ?: $base;
 
-               foreach ($feeds as $feed) {
-                       $attr = [];
-                       foreach ($feed->attributes as $attribute) {
-                               $attr[$attribute->name] = trim($attribute->value);
-                       }
+               $baseParts = parse_url($base);
+
+               // Naked domain case (scheme://basehost)
+               $path = $baseParts['path'] ?? '/';
+
+               // Remove the filename part of the path if it exists (/base/path/file)
+               $path = implode('/', array_slice(explode('/', $path), 0, -1));
 
-                       if (empty($feed_url) && !empty($attr['href'])) {
-                               $feed_url = $attr["href"];
+               $hrefParts = parse_url($href);
+
+               // Root path case (/path) including relative scheme case (//host/path)
+               if ($hrefParts['path'] && $hrefParts['path'][0] == '/') {
+                       $path = $hrefParts['path'];
+               } else {
+                       $path = $path . '/' . $hrefParts['path'];
+
+                       // Resolve arbitrary relative path
+                       // Lifted from https://www.php.net/manual/en/function.realpath.php#84012
+                       $parts = array_filter(explode('/', $path), 'strlen');
+                       $absolutes = array();
+                       foreach ($parts as $part) {
+                               if ('.' == $part) continue;
+                               if ('..' == $part) {
+                                       array_pop($absolutes);
+                               } else {
+                                       $absolutes[] = $part;
+                               }
                        }
+
+                       $path = '/' . implode('/', $absolutes);
                }
 
-               return $feed_url;
+               // Relative scheme case (//host/path)
+               $baseParts['host'] = $hrefParts['host'] ?? $baseParts['host'];
+               $baseParts['path'] = $path;
+               unset($baseParts['query']);
+               unset($baseParts['fragment']);
+
+               return Network::unparseURL($baseParts);
        }
 
        /**
@@ -1816,20 +1861,20 @@ class Probe
                $curlResult = Network::curl($url);
                if ($curlResult->isTimeout()) {
                        self::$istimeout = true;
-                       return false;
+                       return [];
                }
                $feed = $curlResult->getBody();
                $feed_data = Feed::import($feed);
 
                if (!$feed_data) {
                        if (!$probe) {
-                               return false;
+                               return [];
                        }
 
-                       $feed_url = self::getFeedLink($url);
+                       $feed_url = self::getFeedLink($url, $feed);
 
                        if (!$feed_url) {
-                               return false;
+                               return [];
                        }
 
                        return self::feed($feed_url, false);
@@ -1877,11 +1922,11 @@ class Probe
        private static function mail($uri, $uid)
        {
                if (!Network::isEmailDomainValid($uri)) {
-                       return false;
+                       return [];
                }
 
                if ($uid == 0) {
-                       return false;
+                       return [];
                }
 
                $user = DBA::selectFirst('user', ['prvkey'], ['uid' => $uid]);
@@ -1891,7 +1936,7 @@ class Probe
                $mailacct = DBA::selectFirst('mailacct', $fields, $condition);
 
                if (!DBA::isResult($user) || !DBA::isResult($mailacct)) {
-                       return false;
+                       return [];
                }
 
                $mailbox = Email::constructMailboxName($mailacct);
@@ -1899,14 +1944,14 @@ class Probe
                openssl_private_decrypt(hex2bin($mailacct['pass']), $password, $user['prvkey']);
                $mbox = Email::connect($mailbox, $mailacct['user'], $password);
                if (!$mbox) {
-                       return false;
+                       return [];
                }
 
                $msgs = Email::poll($mbox, $uri);
                Logger::log('searching '.$uri.', '.count($msgs).' messages found.', Logger::DEBUG);
 
                if (!count($msgs)) {
-                       return false;
+                       return [];
                }
 
                $phost = substr($uri, strpos($uri, '@') + 1);