use DOMDocument;
use DomXPath;
use Friendica\Core\Cache\Duration;
+use Friendica\Core\Hook;
use Friendica\Core\Logger;
use Friendica\Core\Protocol;
+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;
{
$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 {
+ if (in_array($field, ["gsid", "hide", "account-type"])) {
+ $newdata[$field] = (int)$data[$field];
+ } else {
+ $newdata[$field] = $data[$field];
+ }
+ } elseif ($field != "gsid") {
$newdata[$field] = "";
+ } else {
+ $newdata[$field] = null;
}
}
// Reset the static variable
self::$baseurl = '';
- $ssl_url = "https://".$host."/.well-known/host-meta";
- $url = "http://".$host."/.well-known/host-meta";
+ // Handles the case when the hostname contains the scheme
+ if (!parse_url($host, PHP_URL_SCHEME)) {
+ $ssl_url = "https://" . $host . "/.well-known/host-meta";
+ $url = "http://" . $host . "/.well-known/host-meta";
+ } else {
+ $ssl_url = $host . "/.well-known/host-meta";
+ $url = '';
+ }
$xrd_timeout = DI::config()->get('system', 'xrd_timeout', 20);
- Logger::log("Probing for ".$host, Logger::DEBUG);
+ Logger::info('Probing', ['host' => $host, 'ssl_url' => $ssl_url, 'url' => $url, 'callstack' => System::callstack(20)]);
$xrd = null;
$curlResult = Network::curl($ssl_url, false, ['timeout' => $xrd_timeout, 'accept_content' => 'application/xrd+xml']);
$ssl_connection_error = ($curlResult->getErrorNumber() == CURLE_COULDNT_CONNECT) || ($curlResult->getReturnCode() == 0);
if ($curlResult->isSuccess()) {
$xml = $curlResult->getBody();
- $xrd = XML::parseString($xml, false);
- $host_url = 'https://'.$host;
+ $xrd = XML::parseString($xml, true);
+ if (!empty($url)) {
+ $host_url = 'https://' . $host;
+ } else {
+ $host_url = $host;
+ }
} elseif ($curlResult->isTimeout()) {
Logger::info('Probing timeout', ['url' => $ssl_url], Logger::DEBUG);
self::$istimeout = true;
- return false;
+ return [];
}
- if (!is_object($xrd)) {
+ if (!is_object($xrd) && !empty($url)) {
$curlResult = Network::curl($url, false, ['timeout' => $xrd_timeout, 'accept_content' => 'application/xrd+xml']);
$connection_error = ($curlResult->getErrorNumber() == CURLE_COULDNT_CONNECT) || ($curlResult->getReturnCode() == 0);
if ($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();
- $xrd = XML::parseString($xml, false);
+ $xrd = XML::parseString($xml, true);
$host_url = 'http://'.$host;
}
if (!is_object($xrd)) {
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
*
- * this is a replacement for the "lrdd" function.
- * It isn't used in this class and has some redundancies in the code.
- * When time comes we can check the existing calls for "lrdd" if we can rework them.
- *
- * @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;
return [];
}
- $host = $parts["host"];
+ $host = $parts['scheme'] . '://' . $parts["host"];
if (!empty($parts["port"])) {
$host .= ':'.$parts["port"];
}
}
}
- if (!is_array($webfinger["links"])) {
+ if (empty($webfinger["links"])) {
Logger::log("No webfinger links found for ".$uri, Logger::DEBUG);
- return false;
+ return [];
}
$data = [];
$data[] = ["@attributes" => $link];
}
- if (is_array($webfinger["aliases"])) {
+ if (!empty($webfinger["aliases"]) && is_array($webfinger["aliases"])) {
foreach ($webfinger["aliases"] as $alias) {
$data[] = ["@attributes" =>
["rel" => "alias",
// When the previous detection process had got a time out
// we could falsely detect a Friendica profile as AP profile.
if (!self::$istimeout) {
- $ap_profile = ActivityPub::probeProfile($uri);
+ $ap_profile = ActivityPub::probeProfile($uri, !$cache);
if (empty($data) || (!empty($ap_profile) && empty($network) && (($data['network'] ?? '') != Protocol::DFRN))) {
$data = $ap_profile;
$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'])) {
$data['photo'] = DI::baseUrl() . '/images/person-300.jpg';
}
}
}
- if (!empty(self::$baseurl)) {
+ if (empty($data['baseurl']) && !empty(self::$baseurl)) {
$data['baseurl'] = self::$baseurl;
}
+ if (!empty($data['baseurl']) && empty($data['gsid'])) {
+ $data['gsid'] = GServer::getID($data['baseurl']);
+ }
+
if (empty($data['network'])) {
$data['network'] = Protocol::PHANTOM;
}
}
/**
- * Checks if a profile url should be OStatus but only provides partial information
- *
- * @param array $webfinger Webfinger data
- * @param string $lrdd Path template for webfinger request
- * @param string $type type
+ * Fetch the "subscribe" and add it to the result
*
- * @return array fixed webfinger data
- * @throws HTTPException\InternalServerErrorException
+ * @param array $result
+ * @param array $webfinger
+ * @return array result
*/
- private static function fixOStatus($webfinger, $lrdd, $type)
+ private static function getSubscribeLink(array $result, array $webfinger)
{
- if (empty($webfinger['links']) || empty($webfinger['subject'])) {
- return $webfinger;
+ if (empty($webfinger['links'])) {
+ return $result;
}
- $is_ostatus = false;
- $has_key = false;
-
foreach ($webfinger['links'] as $link) {
- if ($link['rel'] == ActivityNamespace::OSTATUSSUB) {
- $is_ostatus = true;
- }
- if ($link['rel'] == 'magic-public-key') {
- $has_key = true;
+ if (!empty($link['template']) && ($link['rel'] === ActivityNamespace::OSTATUSSUB)) {
+ $result['subscribe'] = $link['template'];
}
}
- if (!$is_ostatus || $has_key) {
- return $webfinger;
- }
-
- $url = Network::switchScheme($webfinger['subject']);
- $path = str_replace('{uri}', urlencode($url), $lrdd);
- $webfinger2 = self::webfinger($path, $type);
-
- // Is the new webfinger detectable as OStatus?
- if (self::ostatus($webfinger2, true)) {
- $webfinger = $webfinger2;
- }
-
- return $webfinger;
+ return $result;
}
/**
{
$parts = parse_url($uri);
+ $hookData = [
+ 'uri' => $uri,
+ 'network' => $network,
+ 'uid' => $uid,
+ 'result' => [],
+ ];
+
+ Hook::callAll('probe_detect', $hookData);
+
+ if ($hookData['result']) {
+ return $hookData['result'];
+ }
+
if (!empty($parts["scheme"]) && !empty($parts["host"])) {
$host = $parts["host"];
if (!empty($parts["port"])) {
$addr = $uri;
} else {
Logger::log("Uri ".$uri." was not detectable", Logger::DEBUG);
- return false;
+ return [];
}
$webfinger = false;
continue;
}
- // At first try it with the given uri
- $path = str_replace('{uri}', urlencode($uri), $template);
- $webfinger = self::webfinger($path, $type);
-
- // Fix possible problems with GNU Social probing to wrong scheme
- $webfinger = self::fixOStatus($webfinger, $template, $type);
-
- // We cannot be sure that the detected address was correct, so we don't use the values
- if ($webfinger && ($uri != $addr)) {
- $nick = "";
- $addr = "";
+ // Try the URI first
+ if ($uri != $addr) {
+ $path = str_replace('{uri}', urlencode($uri), $template);
+ $webfinger = self::webfinger($path, $type);
}
- // Try webfinger with the address (user@domain.tld)
+ // Then try the address
if (!$webfinger) {
- $path = str_replace('{uri}', urlencode($addr), $template);
+ $path = str_replace('{uri}', urlencode("acct:" . $addr), $template);
$webfinger = self::webfinger($path, $type);
}
- // Mastodon needs to have it with "acct:"
+ // Finally try without the "acct"
if (!$webfinger) {
- $path = str_replace('{uri}', urlencode("acct:".$addr), $template);
+ $path = str_replace('{uri}', urlencode($addr), $template);
$webfinger = self::webfinger($path, $type);
}
+
+ // We cannot be sure that the detected address was correct, so we don't use the values
+ if ($webfinger && ($uri != $addr)) {
+ $nick = "";
+ $addr = "";
+ }
}
if (!$webfinger) {
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);
}
}
+ $result = self::getSubscribeLink($result, $webfinger);
+
if (empty($result["network"])) {
$result["network"] = Protocol::PHANTOM;
}
Logger::log($uri." is ".$result["network"], Logger::DEBUG);
- if (empty($result["baseurl"]) && ($result["network"] != Protocol::PHANTOM)) {
- $pos = strpos($result["url"], $host);
- if ($pos) {
- $result["baseurl"] = substr($result["url"], 0, $pos).$host;
- }
- }
return $result;
}
* @return array webfinger data
* @throws HTTPException\InternalServerErrorException
*/
- private static function webfinger($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;
}
// If it is not JSON, maybe it is XML
- $xrd = XML::parseString($data, false);
+ $xrd = XML::parseString($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 = [];
$curlResult = Network::curl($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($json["fn"])) {
}
if (!isset($data["network"]) || ($hcard_url == "")) {
- return false;
+ return [];
}
// Fetch data via noscrape - this is faster
$curlResult = Network::curl($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($data["baseurl"])) {
}
if (empty($data["url"]) || empty($hcard_url)) {
- return false;
+ return [];
}
if (!empty($webfinger["aliases"]) && is_array($webfinger["aliases"])) {
$data = self::pollHcard($hcard_url, $data);
if (!$data) {
- return false;
+ return [];
}
if (!empty($data["url"])
$data["notify"] = $data["baseurl"] . "/receive/users/" . $data["guid"];
$data["batch"] = $data["baseurl"] . "/receive/public";
} else {
- return false;
+ return [];
}
return $data;
$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 $link) {
&& (($link["type"] ?? "") == "text/html")
&& ($link["href"] != "")
) {
- $data["url"] = $link["href"];
+ $data["url"] = $data["alias"] = $link["href"];
} elseif (($link["rel"] == "salmon") && !empty($link["href"])) {
$data["notify"] = $link["href"];
} elseif (($link["rel"] == ActivityNamespace::FEED) && !empty($link["href"])) {
$curlResult = Network::curl($pubkey);
if ($curlResult->isTimeout()) {
self::$istimeout = true;
- return false;
+ return $short ? false : [];
}
$pubkey = $curlResult->getBody();
}
) {
$data["network"] = Protocol::OSTATUS;
} else {
- return false;
+ return $short ? false : [];
}
if ($short) {
$curlResult = Network::curl($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"])) {
$data["url"] = $feed_data["header"]["author-link"];
}
- if (($data['poll'] == $data['url']) && ($data["alias"] != '')) {
- $data['url'] = $data["alias"];
+ if ($data["url"] == $data["alias"]) {
$data["alias"] = '';
}
{
$curlResult = Network::curl($profile_link);
if (!$curlResult->isSuccess()) {
- return false;
+ return [];
}
$doc = new DOMDocument();
if (!@$doc->loadHTML($curlResult->getBody())) {
- return false;
+ return [];
}
$xpath = new DomXPath($doc);
$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);
$data['network'] = Protocol::TWITTER;
$data['baseurl'] = 'https://twitter.com';
- $curlResult = Network::curl($data['url'], false);
- if (!$curlResult->isSuccess()) {
- return [];
- }
+ return $data;
+ }
- $body = $curlResult->getBody();
+ /**
+ * Checks HTML page for RSS 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
+ */
+ public static function getFeedLink(string $url, string $body)
+ {
$doc = new DOMDocument();
- @$doc->loadHTML($body);
- $xpath = new DOMXPath($doc);
+ if (!@$doc->loadHTML($body)) {
+ return false;
+ }
- $list = $xpath->query('//img[@class]');
- foreach ($list as $node) {
- $img_attr = [];
- if ($node->attributes->length) {
- foreach ($node->attributes as $attribute) {
- $img_attr[$attribute->name] = $attribute->value;
- }
- }
+ $xpath = new DOMXPath($doc);
- if (empty($img_attr['class'])) {
- continue;
- }
+ $feedUrl = $xpath->evaluate('string(/html/head/link[@type="application/rss+xml" and @rel="alternate"]/@href)');
- if (strpos($img_attr['class'], 'ProfileAvatar-image') !== false) {
- if (!empty($img_attr['src'])) {
- $data['photo'] = $img_attr['src'];
- }
- if (!empty($img_attr['alt'])) {
- $data['name'] = $img_attr['alt'];
- }
- }
- }
+ $feedUrl = $feedUrl ? self::ensureAbsoluteLinkFromHTMLDoc($feedUrl, $url, $xpath) : '';
- return $data;
+ return $feedUrl;
}
/**
- * Check page for feed link
+ * Return an absolute URL in the context of a HTML document retrieved from the provided URL.
+ *
+ * Loosely based on RFC 1808
*
- * @param string $url Page link
+ * @see https://tools.ietf.org/html/rfc1808
*
- * @return string feed link
+ * @param string $href The potential relative href found in the HTML document
+ * @param string $base The HTML document URL
+ * @param DOMXPath $xpath The HTML document XPath
+ * @return string
*/
- private static function getFeedLink($url)
+ private static function ensureAbsoluteLinkFromHTMLDoc(string $href, string $base, DOMXPath $xpath)
{
- $curlResult = Network::curl($url);
- if (!$curlResult->isSuccess()) {
- return false;
+ if (filter_var($href, FILTER_VALIDATE_URL)) {
+ return $href;
}
- $doc = new DOMDocument();
- if (!@$doc->loadHTML($curlResult->getBody())) {
- return false;
- }
+ $base = $xpath->evaluate('string(/html/head/base/@href)') ?: $base;
- $xpath = new DomXPath($doc);
+ $baseParts = parse_url($base);
- //$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;
- }
+ // Naked domain case (scheme://basehost)
+ $path = $baseParts['path'] ?? '/';
- if ($feeds->length == 0) {
- return false;
- }
+ // Remove the filename part of the path if it exists (/base/path/file)
+ $path = implode('/', array_slice(explode('/', $path), 0, -1));
- $feed_url = "";
+ $hrefParts = parse_url($href);
- foreach ($feeds as $feed) {
- $attr = [];
- foreach ($feed->attributes as $attribute) {
- $attr[$attribute->name] = trim($attribute->value);
+ // Root path case (/path) including relative scheme case (//host/path)
+ if ($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[] = $part;
+ }
}
- if (empty($feed_url) && !empty($attr['href'])) {
- $feed_url = $attr["href"];
- }
+ $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);
}
/**
$curlResult = Network::curl($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($feed_url, false);
private static function mail($uri, $uid)
{
if (!Network::isEmailDomainValid($uri)) {
- return false;
+ return [];
}
if ($uid == 0) {
- return false;
+ return [];
}
$user = DBA::selectFirst('user', ['prvkey'], ['uid' => $uid]);
$mailacct = DBA::selectFirst('mailacct', $fields, $condition);
if (!DBA::isResult($user) || !DBA::isResult($mailacct)) {
- return false;
+ return [];
}
$mailbox = Email::constructMailboxName($mailacct);
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);
projects / friendica.git / blobdiff