X-Git-Url: https://git.mxchange.org/?a=blobdiff_plain;f=src%2FNetwork%2FProbe.php;h=91d894d2ee9ede193683747adaad0196f34a98a8;hb=977248f5105a2b3f127d0aa752ce27214605fbf0;hp=d6ea766312a226ad2ea7679330da3b726322a0e3;hpb=f8c0f24e34e124ab782d5dcf3d2d658d70e19e5c;p=friendica.git diff --git a/src/Network/Probe.php b/src/Network/Probe.php index d6ea766312..91d894d2ee 100644 --- a/src/Network/Probe.php +++ b/src/Network/Probe.php @@ -6,29 +6,29 @@ namespace Friendica\Network; /** * @file src/Network/Probe.php - * @brief Functions for probing URL + * Functions for probing URL */ use DOMDocument; use DomXPath; -use Friendica\Core\Cache; -use Friendica\Core\Config; +use Friendica\Core\Cache\Duration; 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\Profile; +use Friendica\Protocol\ActivityNamespace; use Friendica\Protocol\ActivityPub; use Friendica\Protocol\Email; use Friendica\Protocol\Feed; use Friendica\Util\Crypto; -use Friendica\Util\DateTimeFormat; use Friendica\Util\Network; use Friendica\Util\Strings; use Friendica\Util\XML; /** - * @brief This class contain functions for probing URL + * This class contain functions for probing URL * */ class Probe @@ -37,7 +37,7 @@ class Probe private static $istimeout; /** - * @brief Rearrange the array so that it always has the same order + * Rearrange the array so that it always has the same order * * @param array $data Unordered data * @@ -67,7 +67,7 @@ class Probe } /** - * @brief Check if the hostname belongs to the own server + * Check if the hostname belongs to the own server * * @param string $host The hostname that is to be checked * @@ -75,7 +75,7 @@ class Probe */ private static function ownHost($host) { - $own_host = \get_app()->getHostName(); + $own_host = DI::baseUrl()->getHostname(); $parts = parse_url($host); @@ -90,7 +90,7 @@ class Probe } /** - * @brief Probes for webfinger path via "host-meta" + * 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. @@ -108,25 +108,35 @@ class Probe $ssl_url = "https://".$host."/.well-known/host-meta"; $url = "http://".$host."/.well-known/host-meta"; - $xrd_timeout = Config::get('system', 'xrd_timeout', 20); + $xrd_timeout = DI::config()->get('system', 'xrd_timeout', 20); Logger::log("Probing for ".$host, Logger::DEBUG); $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; + } elseif ($curlResult->isTimeout()) { + Logger::info('Probing timeout', ['url' => $ssl_url], Logger::DEBUG); + self::$istimeout = true; + return false; } if (!is_object($xrd)) { $curlResult = Network::curl($url, false, ['timeout' => $xrd_timeout, 'accept_content' => 'application/xrd+xml']); + $connection_error = ($curlResult->getErrorNumber() == CURLE_COULDNT_CONNECT) || ($curlResult->getReturnCode() == 0); if ($curlResult->isTimeout()) { - Logger::log("Probing timeout for " . $url, Logger::DEBUG); + Logger::info('Probing timeout', ['url' => $url], Logger::DEBUG); + self::$istimeout = true; + return false; + } elseif ($connection_error && $ssl_connection_error) { self::$istimeout = true; return false; } + $xml = $curlResult->getBody(); $xrd = XML::parseString($xml, false); $host_url = 'http://'.$host; @@ -142,11 +152,7 @@ class Probe return []; } - $lrdd = []; - // The following webfinger path is defined in RFC 7033 https://tools.ietf.org/html/rfc7033 - // Problem is that Hubzilla currently doesn't provide all data in the JSON webfinger - // compared to the XML webfinger. So this is commented out by now. - // $lrdd = array("application/jrd+json" => $host_url.'/.well-known/webfinger?resource={uri}'); + $lrdd = ['application/jrd+json' => $host_url . '/.well-known/webfinger?resource={uri}']; foreach ($links["xrd"]["link"] as $value => $link) { if (!empty($link["@attributes"])) { @@ -172,7 +178,7 @@ class Probe } /** - * @brief Perform Webfinger lookup and return DFRN data + * Perform Webfinger lookup and return DFRN data * * Given an email style address, perform webfinger lookup and * return the resulting DFRN profile URL, or if no DFRN profile URL @@ -200,10 +206,10 @@ class Probe Logger::log('webfingerDfrn: '.$webbie.':'.print_r($links, true), Logger::DATA); if (!empty($links) && is_array($links)) { foreach ($links as $link) { - if ($link['@attributes']['rel'] === NAMESPACE_DFRN) { + if ($link['@attributes']['rel'] === ActivityNamespace::DFRN) { $profile_link = $link['@attributes']['href']; } - if (($link['@attributes']['rel'] === NAMESPACE_OSTATUSSUB) && ($profile_link == "")) { + if (($link['@attributes']['rel'] === ActivityNamespace::OSTATUSSUB) && ($profile_link == "")) { $profile_link = 'stat:'.$link['@attributes']['template']; } if ($link['@attributes']['rel'] === 'http://microformats.org/profile/hcard') { @@ -215,7 +221,7 @@ class Probe } /** - * @brief Check an URI for LRDD data + * 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. @@ -311,7 +317,7 @@ class Probe } /** - * @brief Fetch information (protocol endpoints and user information) about a given uri + * Fetch information (protocol endpoints and user information) about a given uri * * @param string $uri Address that should be probed * @param string $network Test for this specific network @@ -325,7 +331,7 @@ class Probe public static function uri($uri, $network = '', $uid = -1, $cache = true) { if ($cache) { - $result = Cache::get('Probe::uri:' . $network . ':' . $uri); + $result = DI::cache()->get('Probe::uri:' . $network . ':' . $uri); if (!is_null($result)) { return $result; } @@ -348,7 +354,7 @@ class Probe if (!self::$istimeout) { $ap_profile = ActivityPub::probeProfile($uri); - if (empty($data) || (!empty($ap_profile) && empty($network) && (($data['network'] ?? '') == Protocol::DFRN))) { + if (empty($data) || (!empty($ap_profile) && empty($network) && (($data['network'] ?? '') != Protocol::DFRN))) { $data = $ap_profile; } elseif (!empty($ap_profile)) { $ap_profile['batch'] = ''; @@ -362,10 +368,10 @@ class Probe $data['url'] = $uri; } - if (!empty($data['photo'])) { - $data['baseurl'] = Network::getUrlMatch(Strings::normaliseLink($data['baseurl'] ?? ''), Strings::normaliseLink($data['photo'])); - } else { - $data['photo'] = System::baseUrl() . '/images/person-300.jpg'; + if (!empty($data['photo']) && !empty($data['baseurl'])) { + $data['baseurl'] = Network::getUrlMatch(Strings::normaliseLink($data['baseurl']), Strings::normaliseLink($data['photo'])); + } elseif (empty($data['photo'])) { + $data['photo'] = DI::baseUrl() . '/images/person-300.jpg'; } if (empty($data['name'])) { @@ -394,6 +400,11 @@ class Probe $data['network'] = Protocol::PHANTOM; } + // Ensure that local connections always are DFRN + if (($network == '') && ($data['network'] != Protocol::PHANTOM) && (self::ownHost($data['baseurl'] ?? '') || self::ownHost($data['url']))) { + $data['network'] = Protocol::DFRN; + } + if (!isset($data['hide']) && in_array($data['network'], Protocol::FEDERATED)) { $data['hide'] = self::getHideStatus($data['url']); } @@ -402,7 +413,7 @@ class Probe // Only store into the cache if the value seems to be valid if (!in_array($data['network'], [Protocol::PHANTOM, Protocol::MAIL])) { - Cache::set('Probe::uri:' . $network . ':' . $uri, $data, Cache::DAY); + DI::cache()->set('Probe::uri:' . $network . ':' . $uri, $data, Duration::DAY); } return $data; @@ -473,7 +484,7 @@ class Probe } /** - * @brief Checks if a profile url should be OStatus but only provides partial information + * 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 @@ -492,7 +503,7 @@ class Probe $has_key = false; foreach ($webfinger['links'] as $link) { - if ($link['rel'] == NAMESPACE_OSTATUSSUB) { + if ($link['rel'] == ActivityNamespace::OSTATUSSUB) { $is_ostatus = true; } if ($link['rel'] == 'magic-public-key') { @@ -517,7 +528,7 @@ class Probe } /** - * @brief Fetch information (protocol endpoints and user information) about a given uri + * Fetch information (protocol endpoints and user information) about a given uri * * This function is only called by the "uri" function that adds caching and rearranging of data. * @@ -650,6 +661,9 @@ class Probe if ((!$result && ($network == "")) || ($network == Protocol::OSTATUS)) { $result = self::ostatus($webfinger); } + if (in_array($network, ['', Protocol::ZOT])) { + $result = self::zot($webfinger, $result); + } if ((!$result && ($network == "")) || ($network == Protocol::PUMPIO)) { $result = self::pumpio($webfinger, $addr); } @@ -677,7 +691,7 @@ class Probe Logger::log($uri." is ".$result["network"], Logger::DEBUG); - if (empty($result["baseurl"])) { + if (empty($result["baseurl"]) && ($result["network"] != Protocol::PHANTOM)) { $pos = strpos($result["url"], $host); if ($pos) { $result["baseurl"] = substr($result["url"], 0, $pos).$host; @@ -687,7 +701,149 @@ class Probe } /** - * @brief Perform a webfinger request. + * Check for Zot contact + * + * @param array $webfinger Webfinger data + * @param array $data previously probed data + * + * @return array Zot data + * @throws HTTPException\InternalServerErrorException + */ + private static function zot($webfinger, $data) + { + if (!empty($webfinger["aliases"]) && is_array($webfinger["aliases"])) { + foreach ($webfinger["aliases"] as $alias) { + if (substr($alias, 0, 5) == 'acct:') { + $data["addr"] = substr($alias, 5); + } + } + } + + if (!empty($webfinger["subject"]) && (substr($webfinger["subject"], 0, 5) == "acct:")) { + $data["addr"] = substr($webfinger["subject"], 5); + } + + $zot_url = ''; + foreach ($webfinger['links'] as $link) { + if (($link['rel'] == 'http://purl.org/zot/protocol') && !empty($link['href'])) { + $zot_url = $link['href']; + } + } + + if (empty($zot_url) && !empty($data['addr']) && !empty(self::$baseurl)) { + $condition = ['nurl' => Strings::normaliseLink(self::$baseurl), 'platform' => ['hubzilla']]; + if (!DBA::exists('gserver', $condition)) { + return $data; + } + $zot_url = self::$baseurl . '/.well-known/zot-info?address=' . $data['addr']; + } + + if (empty($zot_url)) { + return $data; + } + + $data = self::pollZot($zot_url, $data); + + if (!empty($data['url']) && !empty($webfinger['aliases']) && is_array($webfinger['aliases'])) { + foreach ($webfinger['aliases'] as $alias) { + if (!strstr($alias, '@') && Strings::normaliseLink($alias) != Strings::normaliseLink($data['url'])) { + $data['alias'] = $alias; + } + } + } + + return $data; + } + + public static function pollZot($url, $data) + { + $curlResult = Network::curl($url); + if ($curlResult->isTimeout()) { + return $data; + } + $content = $curlResult->getBody(); + if (!$content) { + return $data; + } + + $json = json_decode($content, true); + if (!is_array($json)) { + return $data; + } + + if (empty($data['network'])) { + if (!empty($json['protocols']) && in_array('zot', $json['protocols'])) { + $data['network'] = Protocol::ZOT; + } elseif (!isset($json['protocols'])) { + $data['network'] = Protocol::ZOT; + } + } + + if (!empty($json['guid']) && empty($data['guid'])) { + $data['guid'] = $json['guid']; + } + if (!empty($json['key']) && empty($data['pubkey'])) { + $data['pubkey'] = $json['key']; + } + if (!empty($json['name'])) { + $data['name'] = $json['name']; + } + if (!empty($json['photo'])) { + $data['photo'] = $json['photo']; + if (!empty($json['photo_updated'])) { + $data['photo'] .= '?rev=' . urlencode($json['photo_updated']); + } + } + if (!empty($json['address'])) { + $data['addr'] = $json['address']; + } + if (!empty($json['url'])) { + $data['url'] = $json['url']; + } + if (!empty($json['connections_url'])) { + $data['poco'] = $json['connections_url']; + } + if (isset($json['searchable'])) { + $data['hide'] = !$json['searchable']; + } + if (!empty($json['public_forum'])) { + $data['community'] = $json['public_forum']; + $data['account-type'] = Contact::PAGE_COMMUNITY; + } + + if (!empty($json['profile'])) { + $profile = $json['profile']; + if (!empty($profile['description'])) { + $data['about'] = $profile['description']; + } + if (!empty($profile['gender'])) { + $data['gender'] = $profile['gender']; + } + if (!empty($profile['keywords'])) { + $keywords = implode(', ', $profile['keywords']); + if (!empty($keywords)) { + $data['keywords'] = $keywords; + } + } + + $loc = []; + if (!empty($profile['region'])) { + $loc['region'] = $profile['region']; + } + if (!empty($profile['country'])) { + $loc['country-name'] = $profile['country']; + } + $location = Profile::formatLocation($loc); + if (!empty($location)) { + $data['location'] = $location; + } + } + + return $data; + } + + /** + * Perform a webfinger request. * * For details see RFC 7033: * @@ -699,7 +855,7 @@ class Probe */ private static function webfinger($url, $type) { - $xrd_timeout = Config::get('system', 'xrd_timeout', 20); + $xrd_timeout = DI::config()->get('system', 'xrd_timeout', 20); $curlResult = Network::curl($url, false, ['timeout' => $xrd_timeout, 'accept_content' => $type]); if ($curlResult->isTimeout()) { @@ -757,7 +913,7 @@ class Probe } /** - * @brief Poll the Friendica specific noscrape page. + * Poll the Friendica specific noscrape page. * * "noscrape" is a faster alternative to fetch the data from the hcard. * This functionality was originally created for the directory. @@ -861,7 +1017,7 @@ class Probe } /** - * @brief Check for valid DFRN data + * Check for valid DFRN data * * @param array $data DFRN data * @@ -889,7 +1045,7 @@ class Probe } /** - * @brief Fetch data from a DFRN profile page and via "noscrape" + * Fetch data from a DFRN profile page and via "noscrape" * * @param string $profile_link Link to the profile page * @@ -941,7 +1097,7 @@ class Probe } /** - * @brief Check for DFRN contact + * Check for DFRN contact * * @param array $webfinger Webfinger data * @@ -955,15 +1111,15 @@ class Probe // The array is reversed to take into account the order of preference for same-rel links // See: https://tools.ietf.org/html/rfc7033#section-4.4.4 foreach (array_reverse($webfinger["links"]) as $link) { - if (($link["rel"] == NAMESPACE_DFRN) && !empty($link["href"])) { + if (($link["rel"] == ActivityNamespace::DFRN) && !empty($link["href"])) { $data["network"] = Protocol::DFRN; - } elseif (($link["rel"] == NAMESPACE_FEED) && !empty($link["href"])) { + } elseif (($link["rel"] == ActivityNamespace::FEED) && !empty($link["href"])) { $data["poll"] = $link["href"]; } elseif (($link["rel"] == "http://webfinger.net/rel/profile-page") && (($link["type"] ?? "") == "text/html") && !empty($link["href"])) { $data["url"] = $link["href"]; } elseif (($link["rel"] == "http://microformats.org/profile/hcard") && !empty($link["href"])) { $hcard_url = $link["href"]; - } elseif (($link["rel"] == NAMESPACE_POCO) && !empty($link["href"])) { + } elseif (($link["rel"] == ActivityNamespace::POCO) && !empty($link["href"])) { $data["poco"] = $link["href"]; } elseif (($link["rel"] == "http://webfinger.net/rel/avatar") && !empty($link["href"])) { $data["photo"] = $link["href"]; @@ -1021,7 +1177,7 @@ class Probe } /** - * @brief Poll the hcard page (Diaspora and Friendica specific) + * Poll the hcard page (Diaspora and Friendica specific) * * @param string $hcard_url Link to the hcard page * @param array $data The already fetched data @@ -1149,7 +1305,7 @@ class Probe } /** - * @brief Check for Diaspora contact + * Check for Diaspora contact * * @param array $webfinger Webfinger data * @@ -1160,6 +1316,7 @@ class Probe { $hcard_url = ""; $data = []; + // The array is reversed to take into account the order of preference for same-rel links // See: https://tools.ietf.org/html/rfc7033#section-4.4.4 foreach (array_reverse($webfinger["links"]) as $link) { @@ -1171,9 +1328,9 @@ class Probe $data["guid"] = $link["href"]; } elseif (($link["rel"] == "http://webfinger.net/rel/profile-page") && (($link["type"] ?? "") == "text/html") && !empty($link["href"])) { $data["url"] = $link["href"]; - } elseif (($link["rel"] == NAMESPACE_FEED) && !empty($link["href"])) { + } elseif (($link["rel"] == ActivityNamespace::FEED) && !empty($link["href"])) { $data["poll"] = $link["href"]; - } elseif (($link["rel"] == NAMESPACE_POCO) && !empty($link["href"])) { + } elseif (($link["rel"] == ActivityNamespace::POCO) && !empty($link["href"])) { $data["poco"] = $link["href"]; } elseif (($link["rel"] == "salmon") && !empty($link["href"])) { $data["notify"] = $link["href"]; @@ -1187,7 +1344,7 @@ class Probe } } - if (!isset($data["url"]) || ($hcard_url == "")) { + if (empty($data["url"]) || empty($hcard_url)) { return false; } @@ -1212,11 +1369,11 @@ class Probe return false; } - if (isset($data["url"]) - && isset($data["guid"]) - && isset($data["baseurl"]) - && isset($data["pubkey"]) - && ($hcard_url != "") + if (!empty($data["url"]) + && !empty($data["guid"]) + && !empty($data["baseurl"]) + && !empty($data["pubkey"]) + && !empty($hcard_url) ) { $data["network"] = Protocol::DIASPORA; @@ -1236,7 +1393,7 @@ class Probe } /** - * @brief Check for OStatus contact + * Check for OStatus contact * * @param array $webfinger Webfinger data * @param bool $short Short detection mode @@ -1273,7 +1430,7 @@ class Probe $data["url"] = $link["href"]; } elseif (($link["rel"] == "salmon") && !empty($link["href"])) { $data["notify"] = $link["href"]; - } elseif (($link["rel"] == NAMESPACE_FEED) && !empty($link["href"])) { + } elseif (($link["rel"] == ActivityNamespace::FEED) && !empty($link["href"])) { $data["poll"] = $link["href"]; } elseif (($link["rel"] == "magic-public-key") && !empty($link["href"])) { $pubkey = $link["href"]; @@ -1324,10 +1481,7 @@ class Probe return false; } $feed = $curlResult->getBody(); - $dummy1 = null; - $dummy2 = null; - $dummy2 = null; - $feed_data = Feed::import($feed, $dummy1, $dummy2, $dummy3, true); + $feed_data = Feed::import($feed); if (!$feed_data) { return false; } @@ -1366,7 +1520,7 @@ class Probe } /** - * @brief Fetch data from a pump.io profile page + * Fetch data from a pump.io profile page * * @param string $profile_link Link to the profile page * @@ -1374,8 +1528,13 @@ class Probe */ private static function pumpioProfileData($profile_link) { + $curlResult = Network::curl($profile_link); + if (!$curlResult->isSuccess()) { + return false; + } + $doc = new DOMDocument(); - if (!@$doc->loadHTMLFile($profile_link)) { + if (!@$doc->loadHTML($curlResult->getBody())) { return false; } @@ -1422,11 +1581,10 @@ class Probe } /** - * @brief Check for pump.io contact + * Check for pump.io contact * - * @param array $webfinger Webfinger data - * - * @param $addr + * @param array $webfinger Webfinger data + * @param string $addr * @return array pump.io data */ private static function pumpio($webfinger, $addr) @@ -1480,7 +1638,7 @@ class Probe } /** - * @brief Check for twitter contact + * Check for twitter contact * * @param string $uri * @@ -1540,7 +1698,7 @@ class Probe } /** - * @brief Check page for feed link + * Check page for feed link * * @param string $url Page link * @@ -1548,9 +1706,13 @@ class Probe */ private static function getFeedLink($url) { - $doc = new DOMDocument(); + $curlResult = Network::curl($url); + if (!$curlResult->isSuccess()) { + return false; + } - if (!@$doc->loadHTMLFile($url)) { + $doc = new DOMDocument(); + if (!@$doc->loadHTML($curlResult->getBody())) { return false; } @@ -1574,7 +1736,7 @@ class Probe $attr[$attribute->name] = trim($attribute->value); } - if ($feed_url == "") { + if (empty($feed_url) && !empty($attr['href'])) { $feed_url = $attr["href"]; } } @@ -1583,7 +1745,7 @@ class Probe } /** - * @brief Check for feed contact + * Check for feed contact * * @param string $url Profile link * @param boolean $probe Do a probe if the page contains a feed link @@ -1599,8 +1761,7 @@ class Probe return false; } $feed = $curlResult->getBody(); - $dummy1 = $dummy2 = $dummy3 = null; - $feed_data = Feed::import($feed, $dummy1, $dummy2, $dummy3, true); + $feed_data = Feed::import($feed); if (!$feed_data) { if (!$probe) { @@ -1647,7 +1808,7 @@ class Probe } /** - * @brief Check for mail contact + * Check for mail contact * * @param string $uri Profile link * @param integer $uid User ID @@ -1735,7 +1896,7 @@ class Probe } /** - * @brief Mix two paths together to possibly fix missing parts + * Mix two paths together to possibly fix missing parts * * @param string $avatar Path to the avatar * @param string $base Another path that is hopefully complete