<?php
/**
- * @copyright Copyright (C) 2020, Friendica
+ * @copyright Copyright (C) 2010-2022, the Friendica project
*
* @license GNU AGPL version 3 or any later version
*
use DOMDocument;
use DomXPath;
+use Exception;
use Friendica\Core\Hook;
use Friendica\Core\Logger;
use Friendica\Core\Protocol;
use Friendica\Model\GServer;
use Friendica\Model\Profile;
use Friendica\Model\User;
+use Friendica\Network\HTTPClient\Client\HttpClientOptions;
use Friendica\Protocol\ActivityNamespace;
use Friendica\Protocol\ActivityPub;
use Friendica\Protocol\Email;
use Friendica\Util\Network;
use Friendica\Util\Strings;
use Friendica\Util\XML;
+use GuzzleHttp\Psr7\Uri;
/**
* This class contain functions for probing URL
/**
* Remove stuff from an URI that doesn't belong there
*
- * @param string $URI
+ * @param string $rawUri
* @return string Cleaned URI
*/
- public static function cleanURI(string $URI)
+ public static function cleanURI(string $rawUri): string
{
// At first remove leading and trailing junk
- $URI = trim($URI, "@#?:/ \t\n\r\0\x0B");
+ $rawUri = trim($rawUri, "@#?:/ \t\n\r\0\x0B");
- $parts = parse_url($URI);
-
- if (empty($parts['scheme'])) {
- return $URI;
+ $uri = new Uri($rawUri);
+ if (!$uri->getScheme()) {
+ return $uri->__toString();
}
// Remove the URL fragment, since these shouldn't be part of any profile URL
- unset($parts['fragment']);
-
- $URI = Network::unparseURL($parts);
+ $uri = $uri->withFragment('');
- return $URI;
+ return $uri->__toString();
}
/**
*/
private static function rearrangeData($data)
{
- $fields = ["name", "nick", "guid", "url", "addr", "alias", "photo", "account-type",
- "community", "keywords", "location", "about", "hide",
- "batch", "notify", "poll", "request", "confirm", "subscribe", "poco",
+ $fields = ["name", "nick", "guid", "url", "addr", "alias", "photo", "header",
+ "account-type", "community", "keywords", "location", "about", "xmpp", "matrix",
+ "hide", "batch", "notify", "poll", "request", "confirm", "subscribe", "poco",
"following", "followers", "inbox", "outbox", "sharedinbox",
"priority", "network", "pubkey", "manually-approve", "baseurl", "gsid"];
if (isset($data[$field])) {
if (in_array($field, $numeric_fields)) {
$newdata[$field] = (int)$data[$field];
- } else {
+ } else {
$newdata[$field] = $data[$field];
}
} elseif (!in_array($field, $numeric_fields)) {
Logger::info('Probing', ['host' => $host, 'ssl_url' => $ssl_url, 'url' => $url, 'callstack' => System::callstack(20)]);
$xrd = null;
- $curlResult = DI::httpRequest()->get($ssl_url, ['timeout' => $xrd_timeout, 'accept_content' => 'application/xrd+xml']);
+ $curlResult = DI::httpClient()->get($ssl_url, [HttpClientOptions::TIMEOUT => $xrd_timeout, HttpClientOptions::ACCEPT_CONTENT => ['application/xrd+xml']]);
$ssl_connection_error = ($curlResult->getErrorNumber() == CURLE_COULDNT_CONNECT) || ($curlResult->getReturnCode() == 0);
if ($curlResult->isSuccess()) {
$xml = $curlResult->getBody();
}
if (!is_object($xrd) && !empty($url)) {
- $curlResult = DI::httpRequest()->get($url, ['timeout' => $xrd_timeout, 'accept_content' => 'application/xrd+xml']);
+ $curlResult = DI::httpClient()->get($url, [HttpClientOptions::TIMEOUT => $xrd_timeout, HttpClientOptions::ACCEPT_CONTENT => ['application/xrd+xml']]);
$connection_error = ($curlResult->getErrorNumber() == CURLE_COULDNT_CONNECT) || ($curlResult->getReturnCode() == 0);
if ($curlResult->isTimeout()) {
Logger::info('Probing timeout', ['url' => $url]);
}
}
+ if (Network::isUrlBlocked($host_url)) {
+ Logger::info('Domain is blocked', ['url' => $host]);
+ return [];
+ }
+
self::$baseurl = $host_url;
Logger::info('Probing successful', ['host' => $host]);
* @return string profile link
* @throws HTTPException\InternalServerErrorException
*/
- public static function webfingerDfrn($webbie, &$hcard_url)
+ public static function webfingerDfrn(string $webbie, string &$hcard_url)
{
$profile_link = '';
public static function uri($uri, $network = '', $uid = -1)
{
// Local profiles aren't probed via network
- if (empty($network) && strpos($uri, DI::baseUrl()->getHostname())) {
+ if (empty($network) && Contact::isLocal($uri)) {
$data = self::localProbe($uri);
if (!empty($data)) {
return $data;
*/
private static function getHideStatus($url)
{
- $curlResult = DI::httpRequest()->get($url);
+ $curlResult = DI::httpClient()->get($url, [HttpClientOptions::CONTENT_LENGTH => 1000000]);
if (!$curlResult->isSuccess()) {
return false;
}
- // If the file is too large then exit
- if (($curlResult->getInfo()['download_content_length'] ?? 0) > 1000000) {
- return false;
- }
-
// If it isn't a HTML file then exit
if (($curlResult->getContentType() != '') && !strstr(strtolower($curlResult->getContentType()), 'html')) {
return false;
}
$body = $curlResult->getBody();
+ if (empty($body)) {
+ return false;
+ }
$doc = new DOMDocument();
@$doc->loadHTML($body);
* Get webfinger data from a given URI
*
* @param string $uri
- * @return array Webfinger array
+ * @return array
+ * @throws HTTPException\InternalServerErrorException
*/
- private static function getWebfingerArray(string $uri)
+ private static function getWebfingerArray(string $uri): array
{
$parts = parse_url($uri);
if (!empty($parts['scheme']) && !empty($parts['host'])) {
$host = $parts['host'];
if (!empty($parts['port'])) {
- $host .= ':'.$parts['port'];
+ $host .= ':' . $parts['port'];
}
$baseurl = $parts['scheme'] . '://' . $host;
$nick = '';
$addr = '';
- $path_parts = explode("/", trim($parts['path'] ?? '', "/"));
+ $path_parts = explode('/', trim($parts['path'] ?? '', '/'));
if (!empty($path_parts)) {
$nick = ltrim(end($path_parts), '@');
- // When the last part of the URI is numeric then it is most likely an ID and not a nick name
- if (!is_numeric($nick)) {
- $addr = $nick."@".$host;
- } else {
- $nick = '';
- }
+ $addr = $nick . '@' . $host;
}
$webfinger = self::getWebfinger($parts['scheme'] . '://' . $host . self::WEBFINGER, 'application/jrd+json', $uri, $addr);
if (empty($webfinger) && empty($lrdd)) {
while (empty($lrdd) && empty($webfinger) && (sizeof($path_parts) > 1)) {
- $host .= "/".array_shift($path_parts);
+ $host .= '/' . array_shift($path_parts);
$baseurl = $parts['scheme'] . '://' . $host;
if (!empty($nick)) {
- $addr = $nick."@".$host;
+ $addr = $nick . '@' . $host;
}
$webfinger = self::getWebfinger($parts['scheme'] . '://' . $host . self::WEBFINGER, 'application/jrd+json', $uri, $addr);
*/
private static function getWebfinger(string $template, string $type, string $uri, string $addr)
{
+ if (Network::isUrlBlocked($template)) {
+ Logger::info('Domain is blocked', ['url' => $template]);
+ return [];
+ }
+
// First try the address because this is the primary purpose of webfinger
if (!empty($addr)) {
$detected = $addr;
'uri' => $uri,
'network' => $network,
'uid' => $uid,
- 'result' => [],
+ 'result' => null,
];
Hook::callAll('probe_detect', $hookData);
- if ($hookData['result']) {
- if (!is_array($hookData['result'])) {
- return [];
- } else {
- return $hookData['result'];
- }
+ if (isset($hookData['result'])) {
+ return is_array($hookData['result']) ? $hookData['result'] : [];
}
$parts = parse_url($uri);
-
- if (!empty($parts['scheme']) && !empty($parts['host'])) {
- if (in_array($parts['host'], ['twitter.com', 'mobile.twitter.com'])) {
- return self::twitter($uri);
- }
- } elseif (strstr($uri, '@')) {
- // If the URI starts with "mailto:" then jump directly to the mail detection
- if (strpos($uri, 'mailto:') !== false) {
- $uri = str_replace('mailto:', '', $uri);
- return self::mail($uri, $uid);
- }
-
- if ($network == Protocol::MAIL) {
- return self::mail($uri, $uid);
- }
-
- if (Strings::endsWith($uri, '@twitter.com')
- || Strings::endsWith($uri, '@mobile.twitter.com')
- ) {
- return self::twitter($uri);
- }
- } else {
+ if (empty($parts['scheme']) && empty($parts['host']) && !strstr($parts['path'], '@')) {
Logger::info('URI was not detectable', ['uri' => $uri]);
return [];
}
+ // If the URI starts with "mailto:" then jump directly to the mail detection
+ if (strpos($uri, 'mailto:') !== false) {
+ $uri = str_replace('mailto:', '', $uri);
+ return self::mail($uri, $uid);
+ }
+
+ if ($network == Protocol::MAIL) {
+ return self::mail($uri, $uid);
+ }
+
Logger::info('Probing start', ['uri' => $uri]);
if (!empty($ap_profile['addr']) && ($ap_profile['addr'] != $uri)) {
public static function pollZot($url, $data)
{
- $curlResult = DI::httpRequest()->get($url);
+ $curlResult = DI::httpClient()->get($url);
if ($curlResult->isTimeout()) {
return $data;
}
{
$xrd_timeout = DI::config()->get('system', 'xrd_timeout', 20);
- $curlResult = DI::httpRequest()->get($url, ['timeout' => $xrd_timeout, 'accept_content' => $type]);
+ $curlResult = DI::httpClient()->get($url, [HttpClientOptions::TIMEOUT => $xrd_timeout, HttpClientOptions::ACCEPT_CONTENT => [$type]]);
if ($curlResult->isTimeout()) {
self::$istimeout = true;
return [];
*/
private static function pollNoscrape($noscrape_url, $data)
{
- $curlResult = DI::httpRequest()->get($noscrape_url);
+ $curlResult = DI::httpClient()->get($noscrape_url);
if ($curlResult->isTimeout()) {
self::$istimeout = true;
- return [];
+ return $data;
}
$content = $curlResult->getBody();
if (!$content) {
Logger::info('Empty body', ['url' => $noscrape_url]);
- return [];
+ return $data;
}
$json = json_decode($content, true);
if (!is_array($json)) {
Logger::info('No json data', ['url' => $noscrape_url]);
- return [];
+ return $data;
}
if (!empty($json["fn"])) {
$data["about"] = $json["about"];
}
+ if (!empty($json["xmpp"])) {
+ $data["xmpp"] = $json["xmpp"];
+ }
+
+ if (!empty($json["matrix"])) {
+ $data["matrix"] = $json["matrix"];
+ }
+
if (!empty($json["key"])) {
$data["pubkey"] = $json["key"];
}
*/
private static function pollHcard($hcard_url, $data, $dfrn = false)
{
- $curlResult = DI::httpRequest()->get($hcard_url);
+ $curlResult = DI::httpClient()->get($hcard_url);
if ($curlResult->isTimeout()) {
self::$istimeout = true;
return [];
}
$content = $curlResult->getBody();
- if (!$content) {
+ if (empty($content)) {
return [];
}
$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"] == "http://webfinger.net/rel/profile-page") && empty($link["type"]) && !empty($link["href"])) {
+ $profile_url = $link["href"];
} elseif (($link["rel"] == ActivityNamespace::FEED) && !empty($link["href"])) {
$data["poll"] = $link["href"];
} elseif (($link["rel"] == ActivityNamespace::POCO) && !empty($link["href"])) {
}
}
+ if (empty($data["url"]) && !empty($profile_url)) {
+ $data["url"] = $profile_url;
+ }
+
if (empty($data["url"]) || empty($hcard_url)) {
return [];
}
$pubkey = substr($pubkey, 5);
}
} elseif (Strings::normaliseLink($pubkey) == 'http://') {
- $curlResult = DI::httpRequest()->get($pubkey);
+ $curlResult = DI::httpClient()->get($pubkey);
if ($curlResult->isTimeout()) {
self::$istimeout = true;
return $short ? false : [];
}
// Fetch all additional data from the feed
- $curlResult = DI::httpRequest()->get($data["poll"]);
+ $curlResult = DI::httpClient()->get($data["poll"]);
if ($curlResult->isTimeout()) {
self::$istimeout = true;
return [];
*/
private static function pumpioProfileData($profile_link)
{
- $curlResult = DI::httpRequest()->get($profile_link);
- if (!$curlResult->isSuccess()) {
+ $curlResult = DI::httpClient()->get($profile_link);
+ if (!$curlResult->isSuccess() || empty($curlResult->getBody())) {
return [];
}
return $data;
}
- /**
- * Check for twitter contact
- *
- * @param string $uri
- *
- * @return array twitter data
- */
- private static function twitter($uri)
- {
- if (preg_match('=([^@]+)@(?:mobile\.)?twitter\.com$=i', $uri, $matches)) {
- $nick = $matches[1];
- } elseif (preg_match('=^https?://(?:mobile\.)?twitter\.com/(.+)=i', $uri, $matches)) {
- $nick = $matches[1];
- } else {
- return [];
- }
-
- $data = [];
- $data['url'] = 'https://twitter.com/' . $nick;
- $data['addr'] = $nick . '@twitter.com';
- $data['nick'] = $data['name'] = $nick;
- $data['network'] = Protocol::TWITTER;
- $data['baseurl'] = 'https://twitter.com';
-
- return $data;
- }
-
/**
* Checks HTML page for RSS feed link
*
*/
public static function getFeedLink(string $url, string $body)
{
+ if (empty($body)) {
+ return '';
+ }
+
$doc = new DOMDocument();
if (!@$doc->loadHTML($body)) {
return false;
$hrefParts = parse_url($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[] = $part;
+ if (!empty($hrefParts['path'])) {
+ // 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);
+ $path = '/' . implode('/', $absolutes);
+ }
}
// Relative scheme case (//host/path)
*/
private static function feed($url, $probe = true)
{
- $curlResult = DI::httpRequest()->get($url);
+ $curlResult = DI::httpClient()->get($url);
if ($curlResult->isTimeout()) {
self::$istimeout = true;
return [];
$data["name"] .= $perspart->text;
}
}
-
- $data["name"] = Strings::escapeTags($data["name"]);
}
}
}
return '';
}
- $curlResult = DI::httpRequest()->get($gserver['noscrape'] . '/' . $data['nick']);
+ $curlResult = DI::httpClient()->get($gserver['noscrape'] . '/' . $data['nick']);
if ($curlResult->isSuccess() && !empty($curlResult->getBody())) {
$noscrape = json_decode($curlResult->getBody(), true);
private static function updateFromFeed(array $data)
{
// Search for the newest entry in the feed
- $curlResult = DI::httpRequest()->get($data['poll']);
- if (!$curlResult->isSuccess()) {
+ $curlResult = DI::httpClient()->get($data['poll']);
+ if (!$curlResult->isSuccess() || !$curlResult->getBody()) {
return '';
}
*
* @param string $url
* @return array probed data
+ * @throws HTTPException\InternalServerErrorException
+ * @throws HTTPException\NotFoundException
*/
- private static function localProbe(string $url)
+ private static function localProbe(string $url): array
{
- $uid = User::getIdForURL($url);
- if (empty($uid)) {
- return [];
- }
-
- $profile = User::getOwnerDataById($uid);
- if (empty($profile)) {
- return [];
+ try {
+ $uid = User::getIdForURL($url);
+ if (!$uid) {
+ throw new HTTPException\NotFoundException('User not found.');
+ }
+
+ $owner = User::getOwnerDataById($uid);
+ $approfile = ActivityPub\Transmitter::getProfile($uid);
+
+ if (empty($owner['gsid'])) {
+ $owner['gsid'] = GServer::getID($approfile['generator']['url']);
+ }
+
+ $data = [
+ 'name' => $owner['name'], 'nick' => $owner['nick'], 'guid' => $approfile['diaspora:guid'] ?? '',
+ 'url' => $owner['url'], 'addr' => $owner['addr'], 'alias' => $owner['alias'],
+ 'photo' => User::getAvatarUrl($owner),
+ 'header' => $owner['header'] ? Contact::getHeaderUrlForId($owner['id'], $owner['updated']) : '',
+ 'account-type' => $owner['contact-type'], 'community' => ($owner['contact-type'] == User::ACCOUNT_TYPE_COMMUNITY),
+ 'keywords' => $owner['keywords'], 'location' => $owner['location'], 'about' => $owner['about'],
+ 'xmpp' => $owner['xmpp'], 'matrix' => $owner['matrix'],
+ 'hide' => !$owner['net-publish'], 'batch' => '', 'notify' => $owner['notify'],
+ 'poll' => $owner['poll'], 'request' => $owner['request'], 'confirm' => $owner['confirm'],
+ 'subscribe' => $approfile['generator']['url'] . '/follow?url={uri}', 'poco' => $owner['poco'],
+ 'following' => $approfile['following'], 'followers' => $approfile['followers'],
+ 'inbox' => $approfile['inbox'], 'outbox' => $approfile['outbox'],
+ 'sharedinbox' => $approfile['endpoints']['sharedInbox'], 'network' => Protocol::DFRN,
+ 'pubkey' => $owner['upubkey'], 'baseurl' => $approfile['generator']['url'], 'gsid' => $owner['gsid'],
+ 'manually-approve' => in_array($owner['page-flags'], [User::PAGE_FLAGS_NORMAL, User::PAGE_FLAGS_PRVGROUP])
+ ];
+ } catch (Exception $e) {
+ // Default values for non existing targets
+ $data = [
+ 'name' => $url, 'nick' => $url, 'url' => $url, 'network' => Protocol::PHANTOM,
+ 'photo' => DI::baseUrl() . Contact::DEFAULT_AVATAR_PHOTO
+ ];
}
- $approfile = ActivityPub\Transmitter::getProfile($uid);
- if (empty($approfile)) {
- return [];
- }
-
- if (empty($profile['gsid'])) {
- $profile['gsid'] = GServer::getID($approfile['generator']['url']);
- }
-
- $data = ['name' => $profile['name'], 'nick' => $profile['nick'], 'guid' => $approfile['diaspora:guid'] ?? '',
- 'url' => $profile['url'], 'addr' => $profile['addr'], 'alias' => $profile['alias'],
- 'photo' => $profile['photo'], 'account-type' => $profile['contact-type'],
- 'community' => ($profile['contact-type'] == User::ACCOUNT_TYPE_COMMUNITY),
- 'keywords' => $profile['keywords'], 'location' => $profile['location'], 'about' => $profile['about'],
- 'hide' => !$profile['net-publish'], 'batch' => '', 'notify' => $profile['notify'],
- 'poll' => $profile['poll'], 'request' => $profile['request'], 'confirm' => $profile['confirm'],
- 'subscribe' => $approfile['generator']['url'] . '/follow?url={uri}', 'poco' => $profile['poco'],
- 'following' => $approfile['following'], 'followers' => $approfile['followers'],
- 'inbox' => $approfile['inbox'], 'outbox' => $approfile['outbox'],
- 'sharedinbox' => $approfile['endpoints']['sharedInbox'], 'network' => Protocol::DFRN,
- 'pubkey' => $profile['upubkey'], 'baseurl' => $approfile['generator']['url'], 'gsid' => $profile['gsid'],
- 'manually-approve' => in_array($profile['page-flags'], [User::PAGE_FLAGS_NORMAL, User::PAGE_FLAGS_PRVGROUP])];
- return self::rearrangeData($data);
+ return self::rearrangeData($data);
}
}
projects / friendica.git / blobdiff