3 * @file src/Network/Probe.php
5 namespace Friendica\Network;
8 * @file src/Network/Probe.php
9 * @brief Functions for probing URL
13 use Friendica\Core\Cache;
14 use Friendica\Core\Config;
15 use Friendica\Core\System;
16 use Friendica\Database\DBA;
17 use Friendica\Model\Profile;
18 use Friendica\Protocol\Email;
19 use Friendica\Protocol\Feed;
20 use Friendica\Util\Crypto;
21 use Friendica\Util\DateTimeFormat;
22 use Friendica\Util\Network;
23 use Friendica\Util\XML;
26 require_once 'include/dba.php';
29 * @brief This class contain functions for probing URL
34 private static $baseurl;
37 * @brief Rearrange the array so that it always has the same order
39 * @param array $data Unordered data
41 * @return array Ordered data
43 private static function rearrangeData($data)
45 $fields = ["name", "nick", "guid", "url", "addr", "alias",
46 "photo", "community", "keywords", "location", "about",
47 "batch", "notify", "poll", "request", "confirm", "poco",
48 "priority", "network", "pubkey", "baseurl"];
51 foreach ($fields as $field) {
52 if (isset($data[$field])) {
53 $newdata[$field] = $data[$field];
55 $newdata[$field] = "";
59 // We don't use the "priority" field anymore and replace it with a dummy.
60 $newdata["priority"] = 0;
66 * @brief Check if the hostname belongs to the own server
68 * @param string $host The hostname that is to be checked
70 * @return bool Does the testes hostname belongs to the own server?
72 private static function ownHost($host)
74 $own_host = get_app()->get_hostname();
76 $parts = parse_url($host);
78 if (!isset($parts['scheme'])) {
79 $parts = parse_url('http://'.$host);
82 if (!isset($parts['host'])) {
85 return $parts['host'] == $own_host;
89 * @brief Probes for webfinger path via "host-meta"
91 * We have to check if the servers in the future still will offer this.
92 * It seems as if it was dropped from the standard.
94 * @param string $host The host part of an url
96 * @return array with template and type of the webfinger template for JSON or XML
98 private static function hostMeta($host)
100 // Reset the static variable
103 $ssl_url = "https://".$host."/.well-known/host-meta";
104 $url = "http://".$host."/.well-known/host-meta";
106 $xrd_timeout = Config::get('system', 'xrd_timeout', 20);
109 logger("Probing for ".$host, LOGGER_DEBUG);
112 $ret = Network::curl($ssl_url, false, $redirects, ['timeout' => $xrd_timeout, 'accept_content' => 'application/xrd+xml']);
113 if ($ret['success']) {
115 $xrd = XML::parseString($xml, false);
116 $host_url = 'https://'.$host;
119 if (!is_object($xrd)) {
120 $ret = Network::curl($url, false, $redirects, ['timeout' => $xrd_timeout, 'accept_content' => 'application/xrd+xml']);
121 if (!empty($ret["errno"]) && ($ret['errno'] == CURLE_OPERATION_TIMEDOUT)) {
122 logger("Probing timeout for ".$url, LOGGER_DEBUG);
126 $xrd = XML::parseString($xml, false);
127 $host_url = 'http://'.$host;
129 if (!is_object($xrd)) {
130 logger("No xrd object found for ".$host, LOGGER_DEBUG);
134 $links = XML::elementToArray($xrd);
135 if (!isset($links["xrd"]["link"])) {
136 logger("No xrd data found for ".$host, LOGGER_DEBUG);
141 // The following webfinger path is defined in RFC 7033 https://tools.ietf.org/html/rfc7033
142 // Problem is that Hubzilla currently doesn't provide all data in the JSON webfinger
143 // compared to the XML webfinger. So this is commented out by now.
144 // $lrdd = array("application/jrd+json" => $host_url.'/.well-known/webfinger?resource={uri}');
146 foreach ($links["xrd"]["link"] as $value => $link) {
147 if (!empty($link["@attributes"])) {
148 $attributes = $link["@attributes"];
149 } elseif ($value == "@attributes") {
155 if (($attributes["rel"] == "lrdd") && !empty($attributes["template"])) {
156 $type = (empty($attributes["type"]) ? '' : $attributes["type"]);
158 $lrdd[$type] = $attributes["template"];
162 self::$baseurl = "http://".$host;
164 logger("Probing successful for ".$host, LOGGER_DEBUG);
170 * @brief Perform Webfinger lookup and return DFRN data
172 * Given an email style address, perform webfinger lookup and
173 * return the resulting DFRN profile URL, or if no DFRN profile URL
174 * is located, returns an OStatus subscription template (prefixed
175 * with the string 'stat:' to identify it as on OStatus template).
176 * If this isn't an email style address just return $webbie.
177 * Return an empty string if email-style addresses but webfinger fails,
178 * or if the resultant personal XRD doesn't contain a supported
179 * subscription/friend-request attribute.
181 * amended 7/9/2011 to return an hcard which could save potentially loading
182 * a lengthy content page to scrape dfrn attributes
184 * @param string $webbie Address that should be probed
185 * @param string $hcard_url Link to the hcard - is returned by reference
187 * @return string profile link
189 public static function webfingerDfrn($webbie, &$hcard_url)
193 $links = self::lrdd($webbie);
194 logger('webfingerDfrn: '.$webbie.':'.print_r($links, true), LOGGER_DATA);
196 foreach ($links as $link) {
197 if ($link['@attributes']['rel'] === NAMESPACE_DFRN) {
198 $profile_link = $link['@attributes']['href'];
200 if (($link['@attributes']['rel'] === NAMESPACE_OSTATUSSUB) && ($profile_link == "")) {
201 $profile_link = 'stat:'.$link['@attributes']['template'];
203 if ($link['@attributes']['rel'] === 'http://microformats.org/profile/hcard') {
204 $hcard_url = $link['@attributes']['href'];
208 return $profile_link;
212 * @brief Check an URI for LRDD data
214 * this is a replacement for the "lrdd" function.
215 * It isn't used in this class and has some redundancies in the code.
216 * When time comes we can check the existing calls for "lrdd" if we can rework them.
218 * @param string $uri Address that should be probed
220 * @return array uri data
222 public static function lrdd($uri)
224 $lrdd = self::hostMeta($uri);
227 if (is_bool($lrdd)) {
232 $parts = @parse_url($uri);
237 $host = $parts["host"];
238 if (!empty($parts["port"])) {
239 $host .= ':'.$parts["port"];
242 $path_parts = explode("/", trim($parts["path"], "/"));
244 $nick = array_pop($path_parts);
247 $lrdd = self::hostMeta($host);
248 $host .= "/".array_shift($path_parts);
249 } while (!$lrdd && (sizeof($path_parts) > 0));
253 logger("No lrdd data found for ".$uri, LOGGER_DEBUG);
257 foreach ($lrdd as $type => $template) {
262 $path = str_replace('{uri}', urlencode($uri), $template);
263 $webfinger = self::webfinger($path, $type);
265 if (!$webfinger && (strstr($uri, "@"))) {
266 $path = str_replace('{uri}', urlencode("acct:".$uri), $template);
267 $webfinger = self::webfinger($path, $type);
270 // Special treatment for Mastodon
271 // Problem is that Mastodon uses an URL format like http://domain.tld/@nick
272 // But the webfinger for this format fails.
273 if (!$webfinger && !empty($nick)) {
274 // Mastodon uses a "@" as prefix for usernames in their url format
275 $nick = ltrim($nick, '@');
277 $addr = $nick."@".$host;
279 $path = str_replace('{uri}', urlencode("acct:".$addr), $template);
280 $webfinger = self::webfinger($path, $type);
284 if (!is_array($webfinger["links"])) {
285 logger("No webfinger links found for ".$uri, LOGGER_DEBUG);
291 foreach ($webfinger["links"] as $link) {
292 $data[] = ["@attributes" => $link];
295 if (is_array($webfinger["aliases"])) {
296 foreach ($webfinger["aliases"] as $alias) {
297 $data[] = ["@attributes" =>
307 * @brief Fetch information (protocol endpoints and user information) about a given uri
309 * @param string $uri Address that should be probed
310 * @param string $network Test for this specific network
311 * @param integer $uid User ID for the probe (only used for mails)
312 * @param boolean $cache Use cached values?
314 * @return array uri data
316 public static function uri($uri, $network = "", $uid = -1, $cache = true)
319 $result = Cache::get("Probe::uri:".$network.":".$uri);
320 if (!is_null($result)) {
329 $data = self::detect($uri, $network, $uid);
331 if (!isset($data["url"])) {
335 if (x($data, "photo")) {
336 $data["baseurl"] = Network::getUrlMatch(normalise_link(defaults($data, "baseurl", "")), normalise_link($data["photo"]));
338 $data["photo"] = System::baseUrl().'/images/person-175.jpg';
341 if (empty($data["name"])) {
342 if (!empty($data["nick"])) {
343 $data["name"] = $data["nick"];
346 if (!x($data, "name")) {
347 $data["name"] = $data["url"];
351 if (empty($data["nick"])) {
352 $data["nick"] = strtolower($data["name"]);
354 if (strpos($data['nick'], ' ')) {
355 $data['nick'] = trim(substr($data['nick'], 0, strpos($data['nick'], ' ')));
359 if (!empty(self::$baseurl)) {
360 $data["baseurl"] = self::$baseurl;
363 if (empty($data["network"])) {
364 $data["network"] = NETWORK_PHANTOM;
367 $data = self::rearrangeData($data);
369 // Only store into the cache if the value seems to be valid
370 if (!in_array($data['network'], [NETWORK_PHANTOM, NETWORK_MAIL])) {
371 Cache::set("Probe::uri:".$network.":".$uri, $data, CACHE_DAY);
373 /// @todo temporary fix - we need a real contact update function that updates only changing fields
374 /// The biggest problem is the avatar picture that could have a reduced image size.
375 /// It should only be updated if the existing picture isn't existing anymore.
376 /// We only update the contact when it is no probing for a specific network.
377 if (($data['network'] != NETWORK_FEED)
385 $fields = ['name' => $data['name'],
386 'nick' => $data['nick'],
387 'url' => $data['url'],
388 'addr' => $data['addr'],
389 'photo' => $data['photo'],
390 'keywords' => $data['keywords'],
391 'location' => $data['location'],
392 'about' => $data['about'],
393 'notify' => $data['notify'],
394 'network' => $data['network'],
395 'server_url' => $data['baseurl']];
397 // This doesn't cover the case when a community isn't a community anymore
398 if (!empty($data['community']) && $data['community']) {
399 $fields['community'] = $data['community'];
400 $fields['contact-type'] = ACCOUNT_TYPE_COMMUNITY;
405 foreach ($fields as $key => $val) {
407 unset($fields[$key]);
409 $fieldnames[] = $key;
413 $fields['updated'] = DateTimeFormat::utcNow();
415 $condition = ['nurl' => normalise_link($data["url"])];
417 $old_fields = DBA::selectFirst('gcontact', $fieldnames, $condition);
419 // When the gcontact doesn't exist, the value "true" will trigger an insert.
420 // In difference to the public contacts we want to have every contact
421 // in the world in our global contacts.
425 // These values have to be set only on insert
426 $fields['photo'] = $data['photo'];
427 $fields['created'] = DateTimeFormat::utcNow();
430 DBA::update('gcontact', $fields, $condition, $old_fields);
432 $fields = ['name' => $data['name'],
433 'nick' => $data['nick'],
434 'url' => $data['url'],
435 'addr' => $data['addr'],
436 'alias' => $data['alias'],
437 'keywords' => $data['keywords'],
438 'location' => $data['location'],
439 'about' => $data['about'],
440 'batch' => $data['batch'],
441 'notify' => $data['notify'],
442 'poll' => $data['poll'],
443 'request' => $data['request'],
444 'confirm' => $data['confirm'],
445 'poco' => $data['poco'],
446 'network' => $data['network'],
447 'pubkey' => $data['pubkey'],
448 'priority' => $data['priority'],
450 'rel' => CONTACT_IS_SHARING];
454 foreach ($fields as $key => $val) {
456 unset($fields[$key]);
458 $fieldnames[] = $key;
462 $condition = ['nurl' => normalise_link($data["url"]), 'self' => false, 'uid' => 0];
464 // "$old_fields" will return a "false" when the contact doesn't exist.
465 // This won't trigger an insert. This is intended, since we only need
466 // public contacts for everyone we store items from.
467 // We don't need to store every contact on the planet.
468 $old_fields = DBA::selectFirst('contact', $fieldnames, $condition);
470 $fields['name-date'] = DateTimeFormat::utcNow();
471 $fields['uri-date'] = DateTimeFormat::utcNow();
472 $fields['success_update'] = DateTimeFormat::utcNow();
474 DBA::update('contact', $fields, $condition, $old_fields);
482 * @brief Switch the scheme of an url between http and https
484 * @param string $url URL
486 * @return string switched URL
488 private static function switchScheme($url)
490 $parts = parse_url($url);
492 if (!isset($parts['scheme'])) {
496 if ($parts['scheme'] == 'http') {
497 $url = str_replace('http://', 'https://', $url);
498 } elseif ($parts['scheme'] == 'https') {
499 $url = str_replace('https://', 'http://', $url);
506 * @brief Checks if a profile url should be OStatus but only provides partial information
508 * @param array $webfinger Webfinger data
509 * @param string $lrdd Path template for webfinger request
510 * @param string $type type
512 * @return array fixed webfinger data
514 private static function fixOStatus($webfinger, $lrdd, $type)
516 if (empty($webfinger['links']) || empty($webfinger['subject'])) {
523 foreach ($webfinger['links'] as $link) {
524 if ($link['rel'] == NAMESPACE_OSTATUSSUB) {
527 if ($link['rel'] == 'magic-public-key') {
532 if (!$is_ostatus || $has_key) {
536 $url = self::switchScheme($webfinger['subject']);
537 $path = str_replace('{uri}', urlencode($url), $lrdd);
538 $webfinger2 = self::webfinger($path, $type);
540 // Is the new webfinger detectable as OStatus?
541 if (self::ostatus($webfinger2, true)) {
542 $webfinger = $webfinger2;
549 * @brief Fetch information (protocol endpoints and user information) about a given uri
551 * This function is only called by the "uri" function that adds caching and rearranging of data.
553 * @param string $uri Address that should be probed
554 * @param string $network Test for this specific network
555 * @param integer $uid User ID for the probe (only used for mails)
557 * @return array uri data
559 private static function detect($uri, $network, $uid)
561 $parts = parse_url($uri);
563 if (!empty($parts["scheme"]) && !empty($parts["host"]) && !empty($parts["path"])) {
564 $host = $parts["host"];
565 if (!empty($parts["port"])) {
566 $host .= ':'.$parts["port"];
569 if ($host == 'twitter.com') {
570 return ["network" => NETWORK_TWITTER];
572 $lrdd = self::hostMeta($host);
574 if (is_bool($lrdd)) {
578 $path_parts = explode("/", trim($parts["path"], "/"));
580 while (!$lrdd && (sizeof($path_parts) > 1)) {
581 $host .= "/".array_shift($path_parts);
582 $lrdd = self::hostMeta($host);
585 logger('No XRD data was found for '.$uri, LOGGER_DEBUG);
586 return self::feed($uri);
588 $nick = array_pop($path_parts);
590 // Mastodon uses a "@" as prefix for usernames in their url format
591 $nick = ltrim($nick, '@');
593 $addr = $nick."@".$host;
594 } elseif (strstr($uri, '@')) {
595 // If the URI starts with "mailto:" then jump directly to the mail detection
596 if (strpos($uri, 'mailto:') !== false) {
597 $uri = str_replace('mailto:', '', $uri);
598 return self::mail($uri, $uid);
601 if ($network == NETWORK_MAIL) {
602 return self::mail($uri, $uid);
604 // Remove "acct:" from the URI
605 $uri = str_replace('acct:', '', $uri);
607 $host = substr($uri, strpos($uri, '@') + 1);
608 $nick = substr($uri, 0, strpos($uri, '@'));
610 if (strpos($uri, '@twitter.com')) {
611 return ["network" => NETWORK_TWITTER];
613 $lrdd = self::hostMeta($host);
615 if (is_bool($lrdd)) {
620 logger('No XRD data was found for '.$uri, LOGGER_DEBUG);
621 return self::mail($uri, $uid);
625 logger("Uri ".$uri." was not detectable", LOGGER_DEBUG);
631 /// @todo Do we need the prefix "acct:" or "acct://"?
633 foreach ($lrdd as $type => $template) {
638 // At first try it with the given uri
639 $path = str_replace('{uri}', urlencode($uri), $template);
640 $webfinger = self::webfinger($path, $type);
642 // Fix possible problems with GNU Social probing to wrong scheme
643 $webfinger = self::fixOStatus($webfinger, $template, $type);
645 // We cannot be sure that the detected address was correct, so we don't use the values
646 if ($webfinger && ($uri != $addr)) {
651 // Try webfinger with the address (user@domain.tld)
653 $path = str_replace('{uri}', urlencode($addr), $template);
654 $webfinger = self::webfinger($path, $type);
657 // Mastodon needs to have it with "acct:"
659 $path = str_replace('{uri}', urlencode("acct:".$addr), $template);
660 $webfinger = self::webfinger($path, $type);
665 return self::feed($uri);
670 logger("Probing ".$uri, LOGGER_DEBUG);
672 if (in_array($network, ["", NETWORK_DFRN])) {
673 $result = self::dfrn($webfinger);
675 if ((!$result && ($network == "")) || ($network == NETWORK_DIASPORA)) {
676 $result = self::diaspora($webfinger);
678 if ((!$result && ($network == "")) || ($network == NETWORK_OSTATUS)) {
679 $result = self::ostatus($webfinger);
681 if ((!$result && ($network == "")) || ($network == NETWORK_PUMPIO)) {
682 $result = self::pumpio($webfinger, $addr);
684 if ((!$result && ($network == "")) || ($network == NETWORK_FEED)) {
685 $result = self::feed($uri);
687 // We overwrite the detected nick with our try if the previois routines hadn't detected it.
688 // Additionally it is overwritten when the nickname doesn't make sense (contains spaces).
689 if ((empty($result["nick"]) || (strstr($result["nick"], " "))) && ($nick != "")) {
690 $result["nick"] = $nick;
693 if (empty($result["addr"]) && ($addr != "")) {
694 $result["addr"] = $addr;
698 if (empty($result["network"])) {
699 $result["network"] = NETWORK_PHANTOM;
702 if (empty($result["url"])) {
703 $result["url"] = $uri;
706 logger($uri." is ".$result["network"], LOGGER_DEBUG);
708 if (empty($result["baseurl"])) {
709 $pos = strpos($result["url"], $host);
711 $result["baseurl"] = substr($result["url"], 0, $pos).$host;
718 * @brief Perform a webfinger request.
720 * For details see RFC 7033: <https://tools.ietf.org/html/rfc7033>
722 * @param string $url Address that should be probed
723 * @param string $type type
725 * @return array webfinger data
727 private static function webfinger($url, $type)
729 $xrd_timeout = Config::get('system', 'xrd_timeout', 20);
732 $ret = Network::curl($url, false, $redirects, ['timeout' => $xrd_timeout, 'accept_content' => $type]);
733 if (!empty($ret["errno"]) && ($ret['errno'] == CURLE_OPERATION_TIMEDOUT)) {
736 $data = $ret['body'];
738 $webfinger = json_decode($data, true);
739 if (is_array($webfinger)) {
740 if (!isset($webfinger["links"])) {
741 logger("No json webfinger links for ".$url, LOGGER_DEBUG);
747 // If it is not JSON, maybe it is XML
748 $xrd = XML::parseString($data, false);
749 if (!is_object($xrd)) {
750 logger("No webfinger data retrievable for ".$url, LOGGER_DEBUG);
754 $xrd_arr = XML::elementToArray($xrd);
755 if (!isset($xrd_arr["xrd"]["link"])) {
756 logger("No XML webfinger links for ".$url, LOGGER_DEBUG);
762 if (!empty($xrd_arr["xrd"]["subject"])) {
763 $webfinger["subject"] = $xrd_arr["xrd"]["subject"];
766 if (!empty($xrd_arr["xrd"]["alias"])) {
767 $webfinger["aliases"] = $xrd_arr["xrd"]["alias"];
770 $webfinger["links"] = [];
772 foreach ($xrd_arr["xrd"]["link"] as $value => $data) {
773 if (!empty($data["@attributes"])) {
774 $attributes = $data["@attributes"];
775 } elseif ($value == "@attributes") {
781 $webfinger["links"][] = $attributes;
787 * @brief Poll the Friendica specific noscrape page.
789 * "noscrape" is a faster alternative to fetch the data from the hcard.
790 * This functionality was originally created for the directory.
792 * @param string $noscrape_url Link to the noscrape page
793 * @param array $data The already fetched data
795 * @return array noscrape data
797 private static function pollNoscrape($noscrape_url, $data)
799 $ret = Network::curl($noscrape_url);
800 if (!empty($ret["errno"]) && ($ret['errno'] == CURLE_OPERATION_TIMEDOUT)) {
803 $content = $ret['body'];
805 logger("Empty body for ".$noscrape_url, LOGGER_DEBUG);
809 $json = json_decode($content, true);
810 if (!is_array($json)) {
811 logger("No json data for ".$noscrape_url, LOGGER_DEBUG);
815 if (!empty($json["fn"])) {
816 $data["name"] = $json["fn"];
819 if (!empty($json["addr"])) {
820 $data["addr"] = $json["addr"];
823 if (!empty($json["nick"])) {
824 $data["nick"] = $json["nick"];
827 if (!empty($json["guid"])) {
828 $data["guid"] = $json["guid"];
831 if (!empty($json["comm"])) {
832 $data["community"] = $json["comm"];
835 if (!empty($json["tags"])) {
836 $keywords = implode(" ", $json["tags"]);
837 if ($keywords != "") {
838 $data["keywords"] = $keywords;
842 $location = Profile::formatLocation($json);
844 $data["location"] = $location;
847 if (!empty($json["about"])) {
848 $data["about"] = $json["about"];
851 if (!empty($json["key"])) {
852 $data["pubkey"] = $json["key"];
855 if (!empty($json["photo"])) {
856 $data["photo"] = $json["photo"];
859 if (!empty($json["dfrn-request"])) {
860 $data["request"] = $json["dfrn-request"];
863 if (!empty($json["dfrn-confirm"])) {
864 $data["confirm"] = $json["dfrn-confirm"];
867 if (!empty($json["dfrn-notify"])) {
868 $data["notify"] = $json["dfrn-notify"];
871 if (!empty($json["dfrn-poll"])) {
872 $data["poll"] = $json["dfrn-poll"];
879 * @brief Check for valid DFRN data
881 * @param array $data DFRN data
883 * @return int Number of errors
885 public static function validDfrn($data)
888 if (!isset($data['key'])) {
891 if (!isset($data['dfrn-request'])) {
894 if (!isset($data['dfrn-confirm'])) {
897 if (!isset($data['dfrn-notify'])) {
900 if (!isset($data['dfrn-poll'])) {
907 * @brief Fetch data from a DFRN profile page and via "noscrape"
909 * @param string $profile_link Link to the profile page
911 * @return array profile data
913 public static function profile($profile_link)
917 logger("Check profile ".$profile_link, LOGGER_DEBUG);
919 // Fetch data via noscrape - this is faster
920 $noscrape_url = str_replace(["/hcard/", "/profile/"], "/noscrape/", $profile_link);
921 $data = self::pollNoscrape($noscrape_url, $data);
923 if (!isset($data["notify"])
924 || !isset($data["confirm"])
925 || !isset($data["request"])
926 || !isset($data["poll"])
927 || !isset($data["poco"])
928 || !isset($data["name"])
929 || !isset($data["photo"])
931 $data = self::pollHcard($profile_link, $data, true);
935 $prof_data["addr"] = $data["addr"];
936 $prof_data["nick"] = $data["nick"];
937 $prof_data["dfrn-request"] = $data["request"];
938 $prof_data["dfrn-confirm"] = $data["confirm"];
939 $prof_data["dfrn-notify"] = $data["notify"];
940 $prof_data["dfrn-poll"] = $data["poll"];
941 $prof_data["dfrn-poco"] = $data["poco"];
942 $prof_data["photo"] = $data["photo"];
943 $prof_data["fn"] = $data["name"];
944 $prof_data["key"] = $data["pubkey"];
946 logger("Result for profile ".$profile_link.": ".print_r($prof_data, true), LOGGER_DEBUG);
952 * @brief Check for DFRN contact
954 * @param array $webfinger Webfinger data
956 * @return array DFRN data
958 private static function dfrn($webfinger)
962 foreach ($webfinger["links"] as $link) {
963 if (($link["rel"] == NAMESPACE_DFRN) && ($link["href"] != "")) {
964 $data["network"] = NETWORK_DFRN;
965 } elseif (($link["rel"] == NAMESPACE_FEED) && ($link["href"] != "")) {
966 $data["poll"] = $link["href"];
967 } elseif (($link["rel"] == "http://webfinger.net/rel/profile-page") && ($link["type"] == "text/html") && ($link["href"] != "")) {
968 $data["url"] = $link["href"];
969 } elseif (($link["rel"] == "http://microformats.org/profile/hcard") && ($link["href"] != "")) {
970 $hcard_url = $link["href"];
971 } elseif (($link["rel"] == NAMESPACE_POCO) && ($link["href"] != "")) {
972 $data["poco"] = $link["href"];
973 } elseif (($link["rel"] == "http://webfinger.net/rel/avatar") && ($link["href"] != "")) {
974 $data["photo"] = $link["href"];
975 } elseif (($link["rel"] == "http://joindiaspora.com/seed_location") && ($link["href"] != "")) {
976 $data["baseurl"] = trim($link["href"], '/');
977 } elseif (($link["rel"] == "http://joindiaspora.com/guid") && ($link["href"] != "")) {
978 $data["guid"] = $link["href"];
979 } elseif (($link["rel"] == "diaspora-public-key") && ($link["href"] != "")) {
980 $data["pubkey"] = base64_decode($link["href"]);
982 //if (strstr($data["pubkey"], 'RSA ') || ($link["type"] == "RSA"))
983 if (strstr($data["pubkey"], 'RSA ')) {
984 $data["pubkey"] = Crypto::rsaToPem($data["pubkey"]);
989 if (!empty($webfinger["aliases"]) && is_array($webfinger["aliases"])) {
990 foreach ($webfinger["aliases"] as $alias) {
991 if (normalise_link($alias) != normalise_link($data["url"]) && ! strstr($alias, "@")) {
992 $data["alias"] = $alias;
993 } elseif (substr($alias, 0, 5) == 'acct:') {
994 $data["addr"] = substr($alias, 5);
999 if (!empty($webfinger["subject"]) && (substr($webfinger["subject"], 0, 5) == "acct:")) {
1000 $data["addr"] = substr($webfinger["subject"], 5);
1003 if (!isset($data["network"]) || ($hcard_url == "")) {
1007 // Fetch data via noscrape - this is faster
1008 $noscrape_url = str_replace("/hcard/", "/noscrape/", $hcard_url);
1009 $data = self::pollNoscrape($noscrape_url, $data);
1011 if (isset($data["notify"])
1012 && isset($data["confirm"])
1013 && isset($data["request"])
1014 && isset($data["poll"])
1015 && isset($data["name"])
1016 && isset($data["photo"])
1021 $data = self::pollHcard($hcard_url, $data, true);
1027 * @brief Poll the hcard page (Diaspora and Friendica specific)
1029 * @param string $hcard_url Link to the hcard page
1030 * @param array $data The already fetched data
1031 * @param boolean $dfrn Poll DFRN specific data
1033 * @return array hcard data
1035 private static function pollHcard($hcard_url, $data, $dfrn = false)
1037 $ret = Network::curl($hcard_url);
1038 if (!empty($ret["errno"]) && ($ret['errno'] == CURLE_OPERATION_TIMEDOUT)) {
1041 $content = $ret['body'];
1046 $doc = new DOMDocument();
1047 if (!@$doc->loadHTML($content)) {
1051 $xpath = new DomXPath($doc);
1053 $vcards = $xpath->query("//div[contains(concat(' ', @class, ' '), ' vcard ')]");
1054 if (!is_object($vcards)) {
1058 if (!isset($data["baseurl"])) {
1059 $data["baseurl"] = "";
1062 if ($vcards->length > 0) {
1063 $vcard = $vcards->item(0);
1065 // We have to discard the guid from the hcard in favour of the guid from lrdd
1066 // Reason: Hubzilla doesn't use the value "uid" in the hcard like Diaspora does.
1067 $search = $xpath->query("//*[contains(concat(' ', @class, ' '), ' uid ')]", $vcard); // */
1068 if (($search->length > 0) && empty($data["guid"])) {
1069 $data["guid"] = $search->item(0)->nodeValue;
1072 $search = $xpath->query("//*[contains(concat(' ', @class, ' '), ' nickname ')]", $vcard); // */
1073 if ($search->length > 0) {
1074 $data["nick"] = $search->item(0)->nodeValue;
1077 $search = $xpath->query("//*[contains(concat(' ', @class, ' '), ' fn ')]", $vcard); // */
1078 if ($search->length > 0) {
1079 $data["name"] = $search->item(0)->nodeValue;
1082 $search = $xpath->query("//*[contains(concat(' ', @class, ' '), ' searchable ')]", $vcard); // */
1083 if ($search->length > 0) {
1084 $data["searchable"] = $search->item(0)->nodeValue;
1087 $search = $xpath->query("//*[contains(concat(' ', @class, ' '), ' key ')]", $vcard); // */
1088 if ($search->length > 0) {
1089 $data["pubkey"] = $search->item(0)->nodeValue;
1090 if (strstr($data["pubkey"], 'RSA ')) {
1091 $data["pubkey"] = Crypto::rsaToPem($data["pubkey"]);
1095 $search = $xpath->query("//*[@id='pod_location']", $vcard); // */
1096 if ($search->length > 0) {
1097 $data["baseurl"] = trim($search->item(0)->nodeValue, "/");
1102 if (!empty($vcard)) {
1103 $photos = $xpath->query("//*[contains(concat(' ', @class, ' '), ' photo ') or contains(concat(' ', @class, ' '), ' avatar ')]", $vcard); // */
1104 foreach ($photos as $photo) {
1106 foreach ($photo->attributes as $attribute) {
1107 $attr[$attribute->name] = trim($attribute->value);
1110 if (isset($attr["src"]) && isset($attr["width"])) {
1111 $avatar[$attr["width"]] = $attr["src"];
1114 // We don't have a width. So we just take everything that we got.
1115 // This is a Hubzilla workaround which doesn't send a width.
1116 if ((sizeof($avatar) == 0) && !empty($attr["src"])) {
1117 $avatar[] = $attr["src"];
1122 if (sizeof($avatar)) {
1124 $data["photo"] = self::fixAvatar(array_pop($avatar), $data["baseurl"]);
1128 // Poll DFRN specific data
1129 $search = $xpath->query("//link[contains(concat(' ', @rel), ' dfrn-')]");
1130 if ($search->length > 0) {
1131 foreach ($search as $link) {
1132 //$data["request"] = $search->item(0)->nodeValue;
1134 foreach ($link->attributes as $attribute) {
1135 $attr[$attribute->name] = trim($attribute->value);
1138 $data[substr($attr["rel"], 5)] = $attr["href"];
1142 // Older Friendica versions had used the "uid" field differently than newer versions
1143 if (!empty($data["nick"]) && !empty($data["guid"]) && ($data["nick"] == $data["guid"])) {
1144 unset($data["guid"]);
1153 * @brief Check for Diaspora contact
1155 * @param array $webfinger Webfinger data
1157 * @return array Diaspora data
1159 private static function diaspora($webfinger)
1163 foreach ($webfinger["links"] as $link) {
1164 if (($link["rel"] == "http://microformats.org/profile/hcard") && ($link["href"] != "")) {
1165 $hcard_url = $link["href"];
1166 } elseif (($link["rel"] == "http://joindiaspora.com/seed_location") && ($link["href"] != "")) {
1167 $data["baseurl"] = trim($link["href"], '/');
1168 } elseif (($link["rel"] == "http://joindiaspora.com/guid") && ($link["href"] != "")) {
1169 $data["guid"] = $link["href"];
1170 } elseif (($link["rel"] == "http://webfinger.net/rel/profile-page") && ($link["type"] == "text/html") && ($link["href"] != "")) {
1171 $data["url"] = $link["href"];
1172 } elseif (($link["rel"] == NAMESPACE_FEED) && ($link["href"] != "")) {
1173 $data["poll"] = $link["href"];
1174 } elseif (($link["rel"] == NAMESPACE_POCO) && ($link["href"] != "")) {
1175 $data["poco"] = $link["href"];
1176 } elseif (($link["rel"] == "salmon") && ($link["href"] != "")) {
1177 $data["notify"] = $link["href"];
1178 } elseif (($link["rel"] == "diaspora-public-key") && ($link["href"] != "")) {
1179 $data["pubkey"] = base64_decode($link["href"]);
1181 //if (strstr($data["pubkey"], 'RSA ') || ($link["type"] == "RSA"))
1182 if (strstr($data["pubkey"], 'RSA ')) {
1183 $data["pubkey"] = Crypto::rsaToPem($data["pubkey"]);
1188 if (!isset($data["url"]) || ($hcard_url == "")) {
1192 if (!empty($webfinger["aliases"]) && is_array($webfinger["aliases"])) {
1193 foreach ($webfinger["aliases"] as $alias) {
1194 if (normalise_link($alias) != normalise_link($data["url"]) && ! strstr($alias, "@")) {
1195 $data["alias"] = $alias;
1196 } elseif (substr($alias, 0, 5) == 'acct:') {
1197 $data["addr"] = substr($alias, 5);
1202 if (!empty($webfinger["subject"]) && (substr($webfinger["subject"], 0, 5) == 'acct:')) {
1203 $data["addr"] = substr($webfinger["subject"], 5);
1206 // Fetch further information from the hcard
1207 $data = self::pollHcard($hcard_url, $data);
1213 if (isset($data["url"])
1214 && isset($data["guid"])
1215 && isset($data["baseurl"])
1216 && isset($data["pubkey"])
1217 && ($hcard_url != "")
1219 $data["network"] = NETWORK_DIASPORA;
1221 // The Diaspora handle must always be lowercase
1222 if (!empty($data["addr"])) {
1223 $data["addr"] = strtolower($data["addr"]);
1226 // We have to overwrite the detected value for "notify" since Hubzilla doesn't send it
1227 $data["notify"] = $data["baseurl"] . "/receive/users/" . $data["guid"];
1228 $data["batch"] = $data["baseurl"] . "/receive/public";
1237 * @brief Check for OStatus contact
1239 * @param array $webfinger Webfinger data
1240 * @param bool $short Short detection mode
1242 * @return array|bool OStatus data or "false" on error or "true" on short mode
1244 private static function ostatus($webfinger, $short = false)
1248 if (!empty($webfinger["aliases"]) && is_array($webfinger["aliases"])) {
1249 foreach ($webfinger["aliases"] as $alias) {
1250 if (strstr($alias, "@") && !strstr(normalise_link($alias), "http://")) {
1251 $data["addr"] = str_replace('acct:', '', $alias);
1256 if (!empty($webfinger["subject"]) && strstr($webfinger["subject"], "@")
1257 && !strstr(normalise_link($webfinger["subject"]), "http://")
1259 $data["addr"] = str_replace('acct:', '', $webfinger["subject"]);
1263 if (is_array($webfinger["links"])) {
1264 foreach ($webfinger["links"] as $link) {
1265 if (($link["rel"] == "http://webfinger.net/rel/profile-page")
1266 && ($link["type"] == "text/html")
1267 && ($link["href"] != "")
1269 $data["url"] = $link["href"];
1270 } elseif (($link["rel"] == "salmon") && ($link["href"] != "")) {
1271 $data["notify"] = $link["href"];
1272 } elseif (($link["rel"] == NAMESPACE_FEED) && ($link["href"] != "")) {
1273 $data["poll"] = $link["href"];
1274 } elseif (($link["rel"] == "magic-public-key") && ($link["href"] != "")) {
1275 $pubkey = $link["href"];
1277 if (substr($pubkey, 0, 5) === 'data:') {
1278 if (strstr($pubkey, ',')) {
1279 $pubkey = substr($pubkey, strpos($pubkey, ',') + 1);
1281 $pubkey = substr($pubkey, 5);
1283 } elseif (normalise_link($pubkey) == 'http://') {
1284 $ret = Network::curl($pubkey);
1285 if (!empty($ret["errno"]) && ($ret['errno'] == CURLE_OPERATION_TIMEDOUT)) {
1288 $pubkey = $ret['body'];
1291 $key = explode(".", $pubkey);
1293 if (sizeof($key) >= 3) {
1294 $m = base64url_decode($key[1]);
1295 $e = base64url_decode($key[2]);
1296 $data["pubkey"] = Crypto::meToPem($m, $e);
1302 if (isset($data["notify"]) && isset($data["pubkey"])
1303 && isset($data["poll"])
1304 && isset($data["url"])
1306 $data["network"] = NETWORK_OSTATUS;
1315 // Fetch all additional data from the feed
1316 $ret = Network::curl($data["poll"]);
1317 if (!empty($ret["errno"]) && ($ret['errno'] == CURLE_OPERATION_TIMEDOUT)) {
1320 $feed = $ret['body'];
1324 $feed_data = Feed::import($feed, $dummy1, $dummy2, $dummy3, true);
1329 if (!empty($feed_data["header"]["author-name"])) {
1330 $data["name"] = $feed_data["header"]["author-name"];
1332 if (!empty($feed_data["header"]["author-nick"])) {
1333 $data["nick"] = $feed_data["header"]["author-nick"];
1335 if (!empty($feed_data["header"]["author-avatar"])) {
1336 $data["photo"] = self::fixAvatar($feed_data["header"]["author-avatar"], $data["url"]);
1338 if (!empty($feed_data["header"]["author-id"])) {
1339 $data["alias"] = $feed_data["header"]["author-id"];
1341 if (!empty($feed_data["header"]["author-location"])) {
1342 $data["location"] = $feed_data["header"]["author-location"];
1344 if (!empty($feed_data["header"]["author-about"])) {
1345 $data["about"] = $feed_data["header"]["author-about"];
1347 // OStatus has serious issues when the the url doesn't fit (ssl vs. non ssl)
1348 // So we take the value that we just fetched, although the other one worked as well
1349 if (!empty($feed_data["header"]["author-link"])) {
1350 $data["url"] = $feed_data["header"]["author-link"];
1353 if (($data['poll'] == $data['url']) && ($data["alias"] != '')) {
1354 $data['url'] = $data["alias"];
1355 $data["alias"] = '';
1358 /// @todo Fetch location and "about" from the feed as well
1363 * @brief Fetch data from a pump.io profile page
1365 * @param string $profile_link Link to the profile page
1367 * @return array profile data
1369 private static function pumpioProfileData($profile_link)
1371 $doc = new DOMDocument();
1372 if (!@$doc->loadHTMLFile($profile_link)) {
1376 $xpath = new DomXPath($doc);
1380 $data["name"] = $xpath->query("//span[contains(@class, 'p-name')]")->item(0)->nodeValue;
1382 if ($data["name"] == '') {
1383 // This is ugly - but pump.io doesn't seem to know a better way for it
1384 $data["name"] = trim($xpath->query("//h1[@class='media-header']")->item(0)->nodeValue);
1385 $pos = strpos($data["name"], chr(10));
1387 $data["name"] = trim(substr($data["name"], 0, $pos));
1391 $data["location"] = XML::getFirstNodeValue($xpath, "//p[contains(@class, 'p-locality')]");
1393 if ($data["location"] == '') {
1394 $data["location"] = XML::getFirstNodeValue($xpath, "//p[contains(@class, 'location')]");
1397 $data["about"] = XML::getFirstNodeValue($xpath, "//p[contains(@class, 'p-note')]");
1399 if ($data["about"] == '') {
1400 $data["about"] = XML::getFirstNodeValue($xpath, "//p[contains(@class, 'summary')]");
1403 $avatar = $xpath->query("//img[contains(@class, 'u-photo')]")->item(0);
1405 $avatar = $xpath->query("//img[@class='img-rounded media-object']")->item(0);
1408 foreach ($avatar->attributes as $attribute) {
1409 if ($attribute->name == "src") {
1410 $data["photo"] = trim($attribute->value);
1419 * @brief Check for pump.io contact
1421 * @param array $webfinger Webfinger data
1423 * @return array pump.io data
1425 private static function pumpio($webfinger, $addr)
1428 foreach ($webfinger["links"] as $link) {
1429 if (($link["rel"] == "http://webfinger.net/rel/profile-page")
1430 && ($link["type"] == "text/html")
1431 && ($link["href"] != "")
1433 $data["url"] = $link["href"];
1434 } elseif (($link["rel"] == "activity-inbox") && ($link["href"] != "")) {
1435 $data["notify"] = $link["href"];
1436 } elseif (($link["rel"] == "activity-outbox") && ($link["href"] != "")) {
1437 $data["poll"] = $link["href"];
1438 } elseif (($link["rel"] == "dialback") && ($link["href"] != "")) {
1439 $data["dialback"] = $link["href"];
1442 if (isset($data["poll"]) && isset($data["notify"])
1443 && isset($data["dialback"])
1444 && isset($data["url"])
1446 // by now we use these fields only for the network type detection
1447 // So we unset all data that isn't used at the moment
1448 unset($data["dialback"]);
1450 $data["network"] = NETWORK_PUMPIO;
1455 $profile_data = self::pumpioProfileData($data["url"]);
1457 if (!$profile_data) {
1461 $data = array_merge($data, $profile_data);
1463 if (($addr != '') && ($data['name'] != '')) {
1464 $name = trim(str_replace($addr, '', $data['name']));
1466 $data['name'] = $name;
1474 * @brief Check page for feed link
1476 * @param string $url Page link
1478 * @return string feed link
1480 private static function getFeedLink($url)
1482 $doc = new DOMDocument();
1484 if (!@$doc->loadHTMLFile($url)) {
1488 $xpath = new DomXPath($doc);
1490 //$feeds = $xpath->query("/html/head/link[@type='application/rss+xml']");
1491 $feeds = $xpath->query("/html/head/link[@type='application/rss+xml' and @rel='alternate']");
1492 if (!is_object($feeds)) {
1496 if ($feeds->length == 0) {
1502 foreach ($feeds as $feed) {
1504 foreach ($feed->attributes as $attribute) {
1505 $attr[$attribute->name] = trim($attribute->value);
1508 if ($feed_url == "") {
1509 $feed_url = $attr["href"];
1517 * @brief Check for feed contact
1519 * @param string $url Profile link
1520 * @param boolean $probe Do a probe if the page contains a feed link
1522 * @return array feed data
1524 private static function feed($url, $probe = true)
1526 $ret = Network::curl($url);
1527 if (!empty($ret["errno"]) && ($ret['errno'] == CURLE_OPERATION_TIMEDOUT)) {
1530 $feed = $ret['body'];
1531 $dummy1 = $dummy2 = $dummy3 = null;
1532 $feed_data = Feed::import($feed, $dummy1, $dummy2, $dummy3, true);
1539 $feed_url = self::getFeedLink($url);
1545 return self::feed($feed_url, false);
1548 if (!empty($feed_data["header"]["author-name"])) {
1549 $data["name"] = $feed_data["header"]["author-name"];
1552 if (!empty($feed_data["header"]["author-nick"])) {
1553 $data["nick"] = $feed_data["header"]["author-nick"];
1556 if (!empty($feed_data["header"]["author-avatar"])) {
1557 $data["photo"] = $feed_data["header"]["author-avatar"];
1560 if (!empty($feed_data["header"]["author-id"])) {
1561 $data["alias"] = $feed_data["header"]["author-id"];
1564 $data["url"] = $url;
1565 $data["poll"] = $url;
1567 if (!empty($feed_data["header"]["author-link"])) {
1568 $data["baseurl"] = $feed_data["header"]["author-link"];
1570 $data["baseurl"] = $data["url"];
1573 $data["network"] = NETWORK_FEED;
1579 * @brief Check for mail contact
1581 * @param string $uri Profile link
1582 * @param integer $uid User ID
1584 * @return array mail data
1586 private static function mail($uri, $uid)
1588 if (!Network::isEmailDomainValid($uri)) {
1596 $x = q("SELECT `prvkey` FROM `user` WHERE `uid` = %d LIMIT 1", intval($uid));
1598 $r = q("SELECT * FROM `mailacct` WHERE `uid` = %d AND `server` != '' LIMIT 1", intval($uid));
1600 if (DBA::is_result($x) && DBA::is_result($r)) {
1601 $mailbox = Email::constructMailboxName($r[0]);
1603 openssl_private_decrypt(hex2bin($r[0]['pass']), $password, $x[0]['prvkey']);
1604 $mbox = Email::connect($mailbox, $r[0]['user'], $password);
1610 $msgs = Email::poll($mbox, $uri);
1611 logger('searching '.$uri.', '.count($msgs).' messages found.', LOGGER_DEBUG);
1613 if (!count($msgs)) {
1617 $phost = substr($uri, strpos($uri, '@') + 1);
1620 $data["addr"] = $uri;
1621 $data["network"] = NETWORK_MAIL;
1622 $data["name"] = substr($uri, 0, strpos($uri, '@'));
1623 $data["nick"] = $data["name"];
1624 $data["photo"] = Network::lookupAvatarByEmail($uri);
1625 $data["url"] = 'mailto:'.$uri;
1626 $data["notify"] = 'smtp '.random_string();
1627 $data["poll"] = 'email '.random_string();
1629 $x = Email::messageMeta($mbox, $msgs[0]);
1630 if (stristr($x[0]->from, $uri)) {
1631 $adr = imap_rfc822_parse_adrlist($x[0]->from, '');
1632 } elseif (stristr($x[0]->to, $uri)) {
1633 $adr = imap_rfc822_parse_adrlist($x[0]->to, '');
1636 foreach ($adr as $feadr) {
1637 if ((strcasecmp($feadr->mailbox, $data["name"]) == 0)
1638 &&(strcasecmp($feadr->host, $phost) == 0)
1639 && (strlen($feadr->personal))
1641 $personal = imap_mime_header_decode($feadr->personal);
1643 foreach ($personal as $perspart) {
1644 if ($perspart->charset != "default") {
1645 $data["name"] .= iconv($perspart->charset, 'UTF-8//IGNORE', $perspart->text);
1647 $data["name"] .= $perspart->text;
1651 $data["name"] = notags($data["name"]);
1655 if (!empty($mbox)) {
1663 * @brief Mix two paths together to possibly fix missing parts
1665 * @param string $avatar Path to the avatar
1666 * @param string $base Another path that is hopefully complete
1668 * @return string fixed avatar path
1670 public static function fixAvatar($avatar, $base)
1672 $base_parts = parse_url($base);
1674 // Remove all parts that could create a problem
1675 unset($base_parts['path']);
1676 unset($base_parts['query']);
1677 unset($base_parts['fragment']);
1679 $avatar_parts = parse_url($avatar);
1682 $parts = array_merge($base_parts, $avatar_parts);
1684 // And put them together again
1685 $scheme = isset($parts['scheme']) ? $parts['scheme'] . '://' : '';
1686 $host = isset($parts['host']) ? $parts['host'] : '';
1687 $port = isset($parts['port']) ? ':' . $parts['port'] : '';
1688 $path = isset($parts['path']) ? $parts['path'] : '';
1689 $query = isset($parts['query']) ? '?' . $parts['query'] : '';
1690 $fragment = isset($parts['fragment']) ? '#' . $parts['fragment'] : '';
1692 $fixed = $scheme.$host.$port.$path.$query.$fragment;
1694 logger('Base: '.$base.' - Avatar: '.$avatar.' - Fixed: '.$fixed, LOGGER_DATA);