friendica/src/Network/Probe.php

1580 lines
42 KiB
PHP
Raw Normal View History

2016-07-03 22:27:16 +02:00
<?php
2017-05-07 20:44:30 +02:00
namespace Friendica\Network;
2016-07-03 22:27:16 +02:00
/**
2017-05-07 20:44:30 +02:00
* @file src/Network/Probe.php
2016-07-08 22:55:39 +02:00
* @brief Functions for probing URL
2016-07-07 23:35:44 +02:00
*
2016-07-03 22:27:16 +02:00
*/
use Friendica\App;
2017-08-26 08:04:21 +02:00
use Friendica\Core\System;
2017-04-30 06:01:26 +02:00
use Friendica\Core\Config;
2016-07-03 22:27:16 +02:00
2017-08-30 20:05:41 +02:00
use dba;
2017-05-11 17:53:04 +02:00
use dbm;
use Cache;
use xml;
use DomXPath;
use DOMDocument;
2017-05-08 17:06:57 +02:00
require_once 'include/feed.php';
require_once 'include/email.php';
require_once 'include/network.php';
2016-07-03 22:27:16 +02:00
2016-07-08 22:55:39 +02:00
/**
* @brief This class contain functions for probing URL
*
*/
2016-07-03 22:27:16 +02:00
class Probe {
private static $baseurl;
2016-07-07 23:04:30 +02:00
/**
* @brief Rearrange the array so that it always has the same order
*
* @param array $data Unordered data
*
* @return array Ordered data
*/
private static function rearrangeData($data) {
2016-07-03 23:11:21 +02:00
$fields = array("name", "nick", "guid", "url", "addr", "alias",
"photo", "community", "keywords", "location", "about",
"batch", "notify", "poll", "request", "confirm", "poco",
"priority", "network", "pubkey", "baseurl");
2016-07-03 22:27:16 +02:00
$newdata = array();
2017-05-08 17:06:57 +02:00
foreach ($fields as $field) {
if (isset($data[$field])) {
2016-07-03 22:27:16 +02:00
$newdata[$field] = $data[$field];
2017-05-08 17:06:57 +02:00
} else {
2016-07-03 22:27:16 +02:00
$newdata[$field] = "";
2017-05-08 17:06:57 +02:00
}
}
2016-07-03 22:27:16 +02:00
// We don't use the "priority" field anymore and replace it with a dummy.
$newdata["priority"] = 0;
return $newdata;
}
/**
* @brief Check if the hostname belongs to the own server
*
* @param string $host The hostname that is to be checked
*
* @return bool Does the testes hostname belongs to the own server?
*/
private static function ownHost($host) {
$own_host = get_app()->get_hostname();
$parts = parse_url($host);
if (!isset($parts['scheme'])) {
$parts = parse_url('http://'.$host);
}
if (!isset($parts['host'])) {
return false;
}
return $parts['host'] == $own_host;
}
2016-07-03 22:27:16 +02:00
/**
* @brief Probes for XRD data
2017-05-08 17:06:57 +02:00
*
* @param string $host The host part of an url
2017-05-08 17:06:57 +02:00
*
2016-07-03 22:27:16 +02:00
* @return array
* 'lrdd' => Link to LRDD endpoint
* 'lrdd-xml' => Link to LRDD endpoint in XML format
* 'lrdd-json' => Link to LRDD endpoint in JSON format
*/
private static function xrd($host) {
2016-07-03 22:27:16 +02:00
2017-03-25 11:07:45 +01:00
// Reset the static variable
self::$baseurl = '';
2016-07-03 22:27:16 +02:00
$ssl_url = "https://".$host."/.well-known/host-meta";
$url = "http://".$host."/.well-known/host-meta";
2017-05-08 17:06:57 +02:00
$xrd_timeout = Config::get('system', 'xrd_timeout', 20);
2016-07-03 22:27:16 +02:00
$redirects = 0;
2017-05-23 22:19:39 +02:00
logger("Probing for ".$host, LOGGER_DEBUG);
2017-02-01 22:35:01 +01:00
$ret = z_fetch_url($ssl_url, false, $redirects, array('timeout' => $xrd_timeout, 'accept_content' => 'application/xrd+xml'));
if ($ret['success']) {
$xml = $ret['body'];
$xrd = parse_xml_string($xml, false);
2017-02-01 22:35:01 +01:00
}
2016-07-03 22:27:16 +02:00
if (!is_object($xrd)) {
$ret = z_fetch_url($url, false, $redirects, array('timeout' => $xrd_timeout, 'accept_content' => 'application/xrd+xml'));
if ($ret['errno'] == CURLE_OPERATION_TIMEDOUT) {
2017-05-28 07:38:12 +02:00
logger("Probing timeout for ".$url, LOGGER_DEBUG);
return false;
}
$xml = $ret['body'];
2016-07-03 22:27:16 +02:00
$xrd = parse_xml_string($xml, false);
}
2017-05-08 17:06:57 +02:00
if (!is_object($xrd)) {
2017-05-28 07:38:12 +02:00
logger("No xrd object found for ".$host, LOGGER_DEBUG);
return array();
2017-05-08 17:06:57 +02:00
}
2016-07-03 22:27:16 +02:00
2017-05-11 17:53:04 +02:00
$links = xml::element_to_array($xrd);
2017-05-08 17:06:57 +02:00
if (!isset($links["xrd"]["link"])) {
2017-05-23 22:19:39 +02:00
logger("No xrd data found for ".$host, LOGGER_DEBUG);
return array();
2017-05-08 17:06:57 +02:00
}
2016-07-03 22:27:16 +02:00
$xrd_data = array();
2016-07-09 20:09:09 +02:00
2017-05-08 17:06:57 +02:00
foreach ($links["xrd"]["link"] as $value => $link) {
if (!empty($link["@attributes"])) {
2016-07-03 22:27:16 +02:00
$attributes = $link["@attributes"];
2017-05-08 17:06:57 +02:00
} elseif ($value == "@attributes") {
2016-07-03 22:27:16 +02:00
$attributes = $link;
2017-05-08 17:06:57 +02:00
} else {
2016-07-03 22:27:16 +02:00
continue;
2017-05-08 17:06:57 +02:00
}
2016-07-03 22:27:16 +02:00
2017-05-08 17:36:04 +02:00
if (($attributes["rel"] == "lrdd")
&& ($attributes["type"] == "application/xrd+xml")
2017-05-08 17:36:04 +02:00
) {
2016-07-03 22:27:16 +02:00
$xrd_data["lrdd-xml"] = $attributes["template"];
2017-05-08 17:36:04 +02:00
} elseif (($attributes["rel"] == "lrdd")
&& ($attributes["type"] == "application/json")
2017-05-08 17:36:04 +02:00
) {
2016-07-03 22:27:16 +02:00
$xrd_data["lrdd-json"] = $attributes["template"];
2017-05-08 17:06:57 +02:00
} elseif ($attributes["rel"] == "lrdd") {
2016-07-03 22:27:16 +02:00
$xrd_data["lrdd"] = $attributes["template"];
2017-05-08 17:06:57 +02:00
}
2016-07-03 22:27:16 +02:00
}
self::$baseurl = "http://".$host;
2017-05-28 07:38:12 +02:00
logger("Probing successful for ".$host, LOGGER_DEBUG);
2016-07-03 22:27:16 +02:00
return $xrd_data;
}
2016-07-09 20:09:09 +02:00
/**
* @brief 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
* is located, returns an OStatus subscription template (prefixed
* with the string 'stat:' to identify it as on OStatus template).
* If this isn't an email style address just return $webbie.
* Return an empty string if email-style addresses but webfinger fails,
* or if the resultant personal XRD doesn't contain a supported
* subscription/friend-request attribute.
*
* amended 7/9/2011 to return an hcard which could save potentially loading
* a lengthy content page to scrape dfrn attributes
*
* @param string $webbie Address that should be probed
* @param string $hcard_url Link to the hcard - is returned by reference
2016-07-09 20:09:09 +02:00
*
* @return string profile link
*/
public static function webfingerDfrn($webbie, &$hcard_url) {
2016-07-09 20:09:09 +02:00
$profile_link = '';
2016-10-09 12:47:47 +02:00
$links = self::lrdd($webbie);
2017-05-08 17:06:57 +02:00
logger('webfingerDfrn: '.$webbie.':'.print_r($links, true), LOGGER_DATA);
2016-07-09 20:09:09 +02:00
if (count($links)) {
foreach ($links as $link) {
2017-05-08 17:06:57 +02:00
if ($link['@attributes']['rel'] === NAMESPACE_DFRN) {
2016-07-09 20:09:09 +02:00
$profile_link = $link['@attributes']['href'];
2017-05-08 17:06:57 +02:00
}
if (($link['@attributes']['rel'] === NAMESPACE_OSTATUSSUB) && ($profile_link == "")) {
2016-07-09 20:09:09 +02:00
$profile_link = 'stat:'.$link['@attributes']['template'];
2017-05-08 17:06:57 +02:00
}
if ($link['@attributes']['rel'] === 'http://microformats.org/profile/hcard') {
$hcard_url = $link['@attributes']['href'];
2017-05-08 17:06:57 +02:00
}
2016-07-09 20:09:09 +02:00
}
}
return $profile_link;
}
/**
* @brief Check an URI for LRDD data
*
* this is a replacement for the "lrdd" function in include/network.php.
* 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 $uri Address that should be probed
*
* @return array uri data
*/
public static function lrdd($uri) {
$lrdd = self::xrd($uri);
$webfinger = null;
2016-07-09 20:09:09 +02:00
if (is_bool($lrdd)) {
return array();
}
2016-07-09 20:09:09 +02:00
if (!$lrdd) {
$parts = @parse_url($uri);
2017-05-08 17:06:57 +02:00
if (!$parts) {
2016-07-09 20:09:09 +02:00
return array();
2017-05-08 17:06:57 +02:00
}
2016-07-09 20:09:09 +02:00
$host = $parts["host"];
if (!empty($parts["port"])) {
2017-05-07 15:11:11 +02:00
$host .= ':'.$parts["port"];
}
2016-07-09 20:09:09 +02:00
$path_parts = explode("/", trim($parts["path"], "/"));
$nick = array_pop($path_parts);
2016-07-09 20:09:09 +02:00
do {
$lrdd = self::xrd($host);
$host .= "/".array_shift($path_parts);
} while (!$lrdd && (sizeof($path_parts) > 0));
2016-07-09 20:09:09 +02:00
}
2017-05-08 17:06:57 +02:00
if (!$lrdd) {
2017-05-23 22:19:39 +02:00
logger("No lrdd data found for ".$uri, LOGGER_DEBUG);
2016-07-09 20:09:09 +02:00
return array();
2017-05-08 17:06:57 +02:00
}
2016-07-09 20:09:09 +02:00
2017-05-08 17:06:57 +02:00
foreach ($lrdd as $key => $link) {
if ($webfinger) {
2016-07-09 20:09:09 +02:00
continue;
2017-05-08 17:06:57 +02:00
}
2016-07-09 20:09:09 +02:00
2017-05-08 17:06:57 +02:00
if (!in_array($key, array("lrdd", "lrdd-xml", "lrdd-json"))) {
2016-07-09 20:09:09 +02:00
continue;
2017-05-08 17:06:57 +02:00
}
2016-07-09 20:09:09 +02:00
$path = str_replace('{uri}', urlencode($uri), $link);
$webfinger = self::webfinger($path);
if (!$webfinger && (strstr($uri, "@"))) {
$path = str_replace('{uri}', urlencode("acct:".$uri), $link);
$webfinger = self::webfinger($path);
}
// Special treatment for Mastodon
// Problem is that Mastodon uses an URL format like http://domain.tld/@nick
// But the webfinger for this format fails.
if (!$webfinger && !empty($nick)) {
// Mastodon uses a "@" as prefix for usernames in their url format
$nick = ltrim($nick, '@');
$addr = $nick."@".$host;
$path = str_replace('{uri}', urlencode("acct:".$addr), $link);
$webfinger = self::webfinger($path);
}
2016-07-09 20:09:09 +02:00
}
2017-05-08 17:06:57 +02:00
if (!is_array($webfinger["links"])) {
2017-05-23 22:19:39 +02:00
logger("No webfinger links found for ".$uri, LOGGER_DEBUG);
2016-07-09 20:09:09 +02:00
return false;
2017-05-08 17:06:57 +02:00
}
2016-07-09 20:09:09 +02:00
$data = array();
2017-05-08 17:06:57 +02:00
foreach ($webfinger["links"] as $link) {
2016-07-09 20:09:09 +02:00
$data[] = array("@attributes" => $link);
2017-05-08 17:06:57 +02:00
}
2016-07-09 20:09:09 +02:00
2017-05-08 17:06:57 +02:00
if (is_array($webfinger["aliases"])) {
foreach ($webfinger["aliases"] as $alias) {
2016-07-09 20:09:09 +02:00
$data[] = array("@attributes" =>
array("rel" => "alias",
"href" => $alias));
2017-05-08 17:06:57 +02:00
}
}
2016-07-09 20:09:09 +02:00
return $data;
}
2016-07-07 23:04:30 +02:00
/**
2016-07-10 18:44:48 +02:00
* @brief Fetch information (protocol endpoints and user information) about a given uri
2016-07-07 23:04:30 +02:00
*
* @param string $uri Address that should be probed
* @param string $network Test for this specific network
* @param integer $uid User ID for the probe (only used for mails)
* @param boolean $cache Use cached values?
*
* @return array uri data
*/
public static function uri($uri, $network = "", $uid = 0, $cache = true) {
2016-07-03 22:27:16 +02:00
if ($cache) {
2017-05-11 17:53:04 +02:00
$result = Cache::get("probe_url:".$network.":".$uri);
if (!is_null($result)) {
return $result;
}
2016-07-04 00:14:08 +02:00
}
2017-05-08 17:06:57 +02:00
if ($uid == 0) {
$uid = local_user();
2017-05-08 17:06:57 +02:00
}
$data = self::detect($uri, $network, $uid);
2016-07-03 22:27:16 +02:00
2017-05-08 17:06:57 +02:00
if (!isset($data["url"])) {
2016-07-03 22:27:16 +02:00
$data["url"] = $uri;
2017-05-08 17:06:57 +02:00
}
2016-07-03 22:27:16 +02:00
2017-05-08 17:06:57 +02:00
if ($data["photo"] != "") {
2016-07-03 22:27:16 +02:00
$data["baseurl"] = matching_url(normalise_link($data["baseurl"]), normalise_link($data["photo"]));
2017-05-08 17:06:57 +02:00
} else {
$data["photo"] = System::baseUrl().'/images/person-175.jpg';
2017-05-08 17:06:57 +02:00
}
2016-07-03 22:27:16 +02:00
if (empty($data["name"])) {
if (!empty($data["nick"])) {
2016-07-10 09:27:40 +02:00
$data["name"] = $data["nick"];
2017-05-08 17:06:57 +02:00
}
2016-07-03 22:27:16 +02:00
2017-05-08 17:06:57 +02:00
if ($data["name"] == "") {
2016-07-10 09:27:40 +02:00
$data["name"] = $data["url"];
2017-05-08 17:06:57 +02:00
}
2016-07-10 09:27:40 +02:00
}
if (empty($data["nick"])) {
2016-07-03 22:27:16 +02:00
$data["nick"] = strtolower($data["name"]);
2017-05-08 17:06:57 +02:00
if (strpos($data['nick'], ' ')) {
$data['nick'] = trim(substr($data['nick'], 0, strpos($data['nick'], ' ')));
2017-05-08 17:06:57 +02:00
}
}
if (self::$baseurl != "") {
$data["baseurl"] = self::$baseurl;
}
if (!isset($data["network"])) {
2016-07-03 22:27:16 +02:00
$data["network"] = NETWORK_PHANTOM;
}
2016-07-03 22:27:16 +02:00
$data = self::rearrangeData($data);
2016-07-03 22:27:16 +02:00
2016-07-04 00:14:08 +02:00
// Only store into the cache if the value seems to be valid
if (!in_array($data['network'], array(NETWORK_PHANTOM, NETWORK_MAIL))) {
2017-05-11 17:53:04 +02:00
Cache::set("probe_url:".$network.":".$uri, $data, CACHE_DAY);
2016-07-04 00:14:08 +02:00
/// @todo temporary fix - we need a real contact update function that updates only changing fields
/// The biggest problem is the avatar picture that could have a reduced image size.
/// It should only be updated if the existing picture isn't existing anymore.
2017-05-08 22:01:18 +02:00
/// We only update the contact when it is no probing for a specific network.
2017-05-08 17:36:04 +02:00
if (($data['network'] != NETWORK_FEED)
&& ($network == "")
&& $data["name"]
&& $data["nick"]
&& $data["url"]
&& $data["addr"]
&& $data["poll"]
2017-05-08 17:36:04 +02:00
) {
2017-08-30 20:05:41 +02:00
$fields = array('name' => $data['name'],
'nick' => $data['nick'],
'url' => $data['url'],
'addr' => $data['addr'],
'photo' => $data['photo'],
'keywords' => $data['keywords'],
'location' => $data['location'],
'about' => $data['about'],
'notify' => $data['notify'],
'network' => $data['network'],
'server_url' => $data['baseurl'],
'updated' => dbm::date());
foreach ($fields AS $key => $val) {
if (empty($val)) {
unset($fields[$key]);
}
}
dba::update('gcontact', $fields, array('nurl' => normalise_link($data["url"])));
$fields = array('name' => $data['name'],
'nick' => $data['nick'],
'url' => $data['url'],
'addr' => $data['addr'],
'alias' => $data['alias'],
'keywords' => $data['keywords'],
'location' => $data['location'],
'about' => $data['about'],
'batch' => $data['batch'],
'notify' => $data['notify'],
'poll' => $data['poll'],
'request' => $data['request'],
'confirm' => $data['confirm'],
'poco' => $data['poco'],
'network' => $data['network'],
'success_update' => dbm::date());
foreach ($fields AS $key => $val) {
if (empty($val)) {
unset($fields[$key]);
}
}
$condition = array('nurl' => normalise_link($data["url"]), 'self' => false, 'uid' => 0);
dba::update('contact', $fields, $condition);
2017-05-08 17:06:57 +02:00
}
2016-07-04 00:14:08 +02:00
}
2016-07-03 22:27:16 +02:00
return $data;
}
/**
* @brief Switch the scheme of an url between http and https
*
* @param string $url URL
*
* @return string switched URL
*/
private static function switchScheme($url) {
$parts = parse_url($url);
if (!isset($parts['scheme'])) {
return $url;
}
if ($parts['scheme'] == 'http') {
$url = str_replace('http://', 'https://', $url);
} elseif ($parts['scheme'] == 'https') {
$url = str_replace('https://', 'http://', $url);
}
return $url;
}
/**
* @brief Checks if a profile url should be OStatus but only provides partial information
*
* @param array $webfinger Webfinger data
* @param string $lrdd Path template for webfinger request
*
* @return array fixed webfinger data
*/
private static function fixOstatus($webfinger, $lrdd) {
if (empty($webfinger['links']) || empty($webfinger['subject'])) {
return $webfinger;
}
$is_ostatus = false;
$has_key = false;
foreach ($webfinger['links'] as $link) {
if ($link['rel'] == NAMESPACE_OSTATUSSUB) {
$is_ostatus = true;
}
if ($link['rel'] == 'magic-public-key') {
$has_key = true;
}
}
if (!$is_ostatus || $has_key) {
return $webfinger;
}
$url = self::switchScheme($webfinger['subject']);
$path = str_replace('{uri}', urlencode($url), $lrdd);
$webfinger2 = self::webfinger($path);
// Is the new webfinger detectable as OStatus?
if (self::ostatus($webfinger2, true)) {
$webfinger = $webfinger2;
}
return $webfinger;
}
2016-07-07 23:04:30 +02:00
/**
2016-07-10 18:44:48 +02:00
* @brief 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.
2016-07-07 23:04:30 +02:00
*
* @param string $uri Address that should be probed
* @param string $network Test for this specific network
* @param integer $uid User ID for the probe (only used for mails)
*
* @return array uri data
*/
private static function detect($uri, $network, $uid) {
$parts = parse_url($uri);
if (!empty($parts["scheme"]) && !empty($parts["host"]) && !empty($parts["path"])) {
$host = $parts["host"];
if (!empty($parts["port"])) {
2017-05-07 15:11:11 +02:00
$host .= ':'.$parts["port"];
}
2017-03-27 23:02:56 +02:00
if ($host == 'twitter.com') {
return array("network" => NETWORK_TWITTER);
2017-03-27 23:02:56 +02:00
}
$lrdd = self::xrd($host);
if (is_bool($lrdd)) {
return array();
}
$path_parts = explode("/", trim($parts["path"], "/"));
while (!$lrdd && (sizeof($path_parts) > 1)) {
$host .= "/".array_shift($path_parts);
$lrdd = self::xrd($host);
}
2017-03-27 23:02:56 +02:00
if (!$lrdd) {
2017-05-28 07:38:12 +02:00
logger('No XRD data was found for '.$uri, LOGGER_DEBUG);
return self::feed($uri);
2017-03-27 23:02:56 +02:00
}
$nick = array_pop($path_parts);
// Mastodon uses a "@" as prefix for usernames in their url format
$nick = ltrim($nick, '@');
$addr = $nick."@".$host;
2017-05-08 17:06:57 +02:00
} elseif (strstr($uri, '@')) {
2016-07-10 09:29:43 +02:00
// If the URI starts with "mailto:" then jump directly to the mail detection
if (strpos($uri, 'mailto:') !== false) {
$uri = str_replace('mailto:', '', $uri);
return self::mail($uri, $uid);
2016-07-03 22:27:16 +02:00
}
if ($network == NETWORK_MAIL) {
return self::mail($uri, $uid);
}
2016-07-03 22:27:16 +02:00
// Remove "acct:" from the URI
$uri = str_replace('acct:', '', $uri);
2017-05-08 17:06:57 +02:00
$host = substr($uri, strpos($uri, '@') + 1);
$nick = substr($uri, 0, strpos($uri, '@'));
2016-07-03 22:27:16 +02:00
2017-03-27 23:02:56 +02:00
if (strpos($uri, '@twitter.com')) {
2016-07-08 21:15:20 +02:00
return array("network" => NETWORK_TWITTER);
2017-03-27 23:02:56 +02:00
}
2016-07-03 22:27:16 +02:00
$lrdd = self::xrd($host);
if (is_bool($lrdd)) {
return array();
}
2017-03-27 23:02:56 +02:00
if (!$lrdd) {
2017-05-28 07:38:12 +02:00
logger('No XRD data was found for '.$uri, LOGGER_DEBUG);
return self::mail($uri, $uid);
2017-03-27 23:02:56 +02:00
}
2016-07-03 22:27:16 +02:00
$addr = $uri;
2017-05-08 17:06:57 +02:00
2016-07-03 22:27:16 +02:00
} else {
2017-05-23 22:19:39 +02:00
logger("Uri ".$uri." was not detectable", LOGGER_DEBUG);
return false;
2016-07-03 22:27:16 +02:00
}
2016-07-03 22:27:16 +02:00
$webfinger = false;
/// @todo Do we need the prefix "acct:" or "acct://"?
2017-05-08 17:06:57 +02:00
foreach ($lrdd as $key => $link) {
if ($webfinger) {
2016-07-03 22:27:16 +02:00
continue;
}
if (!in_array($key, array("lrdd", "lrdd-xml", "lrdd-json"))) {
2016-07-03 22:27:16 +02:00
continue;
}
// At first try it with the given uri
$path = str_replace('{uri}', urlencode($uri), $link);
2016-07-03 22:27:16 +02:00
$webfinger = self::webfinger($path);
// Fix possible problems with GNU Social probing to wrong scheme
$webfinger = self::fixOstatus($webfinger, $link);
// We cannot be sure that the detected address was correct, so we don't use the values
if ($webfinger && ($uri != $addr)) {
$nick = "";
$addr = "";
}
// Try webfinger with the address (user@domain.tld)
if (!$webfinger) {
$path = str_replace('{uri}', urlencode($addr), $link);
$webfinger = self::webfinger($path);
}
// Mastodon needs to have it with "acct:"
if (!$webfinger) {
$path = str_replace('{uri}', urlencode("acct:".$addr), $link);
$webfinger = self::webfinger($path);
}
2016-07-03 22:27:16 +02:00
}
if (!$webfinger) {
2016-07-03 22:27:16 +02:00
return self::feed($uri);
}
2016-07-03 22:27:16 +02:00
$result = false;
2016-07-04 00:49:38 +02:00
logger("Probing ".$uri, LOGGER_DEBUG);
2017-05-08 17:06:57 +02:00
if (in_array($network, array("", NETWORK_DFRN))) {
2016-07-03 22:27:16 +02:00
$result = self::dfrn($webfinger);
2017-05-08 17:06:57 +02:00
}
if ((!$result && ($network == "")) || ($network == NETWORK_DIASPORA)) {
2016-07-03 22:27:16 +02:00
$result = self::diaspora($webfinger);
2017-05-08 17:06:57 +02:00
}
if ((!$result && ($network == "")) || ($network == NETWORK_OSTATUS)) {
2016-07-03 22:27:16 +02:00
$result = self::ostatus($webfinger);
2017-05-08 17:06:57 +02:00
}
if ((!$result && ($network == "")) || ($network == NETWORK_PUMPIO)) {
2016-07-03 22:27:16 +02:00
$result = self::pumpio($webfinger);
2017-05-08 17:06:57 +02:00
}
if ((!$result && ($network == "")) || ($network == NETWORK_FEED)) {
2016-07-03 22:27:16 +02:00
$result = self::feed($uri);
2017-05-08 17:06:57 +02:00
} else {
2016-07-03 22:27:16 +02:00
// We overwrite the detected nick with our try if the previois routines hadn't detected it.
// Additionally it is overwritten when the nickname doesn't make sense (contains spaces).
if ((empty($result["nick"]) || (strstr($result["nick"], " "))) && ($nick != "")) {
2016-07-03 22:27:16 +02:00
$result["nick"] = $nick;
2017-05-08 17:06:57 +02:00
}
2016-07-03 22:27:16 +02:00
if (empty($result["addr"]) && ($addr != "")) {
2016-07-03 22:27:16 +02:00
$result["addr"] = $addr;
2017-05-08 17:06:57 +02:00
}
2016-07-03 22:27:16 +02:00
}
2016-07-04 00:49:38 +02:00
logger($uri." is ".$result["network"], LOGGER_DEBUG);
if (empty($result["baseurl"])) {
2016-07-03 22:27:16 +02:00
$pos = strpos($result["url"], $host);
2017-05-08 17:06:57 +02:00
if ($pos) {
2016-07-03 22:27:16 +02:00
$result["baseurl"] = substr($result["url"], 0, $pos).$host;
2017-05-08 17:06:57 +02:00
}
2016-07-03 22:27:16 +02:00
}
return $result;
}
2016-07-07 23:04:30 +02:00
/**
2016-07-11 19:48:37 +02:00
* @brief Perform a webfinger request.
*
* For details see RFC 7033: <https://tools.ietf.org/html/rfc7033>
2016-07-07 23:04:30 +02:00
*
* @param string $url Address that should be probed
*
* @return array webfinger data
*/
private static function webfinger($url) {
2016-07-03 22:27:16 +02:00
2017-05-08 17:06:57 +02:00
$xrd_timeout = Config::get('system', 'xrd_timeout', 20);
2016-07-03 22:27:16 +02:00
$redirects = 0;
2017-02-01 22:35:01 +01:00
$ret = z_fetch_url($url, false, $redirects, array('timeout' => $xrd_timeout, 'accept_content' => 'application/xrd+xml'));
if ($ret['errno'] == CURLE_OPERATION_TIMEDOUT) {
return false;
}
$data = $ret['body'];
2017-08-30 20:05:41 +02:00
// This is a bugfix for this issue: https://github.com/redmatrix/hubzilla/issues/851
// $data = str_replace('&url=', '&amp;url=', $data);
// we have to decide if we want to create a workaround - or we wait for an update
2016-07-03 22:27:16 +02:00
$xrd = parse_xml_string($data, false);
if (!is_object($xrd)) {
// If it is not XML, maybe it is JSON
$webfinger = json_decode($data, true);
2017-05-08 17:06:57 +02:00
if (!isset($webfinger["links"])) {
2017-05-23 22:19:39 +02:00
logger("No json webfinger links for ".$url, LOGGER_DEBUG);
2016-07-03 22:27:16 +02:00
return false;
2017-05-08 17:06:57 +02:00
}
2016-07-03 22:27:16 +02:00
return $webfinger;
}
2017-05-11 17:53:04 +02:00
$xrd_arr = xml::element_to_array($xrd);
2017-05-08 17:06:57 +02:00
if (!isset($xrd_arr["xrd"]["link"])) {
2017-05-23 22:19:39 +02:00
logger("No XML webfinger links for ".$url, LOGGER_DEBUG);
2016-07-03 22:27:16 +02:00
return false;
2017-05-08 17:06:57 +02:00
}
2016-07-03 22:27:16 +02:00
$webfinger = array();
if (!empty($xrd_arr["xrd"]["subject"])) {
2016-07-03 22:27:16 +02:00
$webfinger["subject"] = $xrd_arr["xrd"]["subject"];
2017-05-08 17:06:57 +02:00
}
2016-07-03 22:27:16 +02:00
if (!empty($xrd_arr["xrd"]["alias"])) {
2016-07-03 22:27:16 +02:00
$webfinger["aliases"] = $xrd_arr["xrd"]["alias"];
2017-05-08 17:06:57 +02:00
}
2016-07-03 22:27:16 +02:00
$webfinger["links"] = array();
2017-05-08 17:06:57 +02:00
foreach ($xrd_arr["xrd"]["link"] as $value => $data) {
if (!empty($data["@attributes"])) {
2016-07-03 22:27:16 +02:00
$attributes = $data["@attributes"];
2017-05-08 17:06:57 +02:00
} elseif ($value == "@attributes") {
2016-07-03 22:27:16 +02:00
$attributes = $data;
2017-05-08 17:06:57 +02:00
} else {
2016-07-03 22:27:16 +02:00
continue;
2017-05-08 17:06:57 +02:00
}
2016-07-03 22:27:16 +02:00
$webfinger["links"][] = $attributes;
}
return $webfinger;
}
2016-07-07 23:04:30 +02:00
/**
2016-07-11 19:48:37 +02:00
* @brief Poll the Friendica specific noscrape page.
2016-07-07 23:04:30 +02:00
*
2016-07-11 19:48:37 +02:00
* "noscrape" is a faster alternative to fetch the data from the hcard.
* This functionality was originally created for the directory.
2016-07-10 18:44:48 +02:00
*
* @param string $noscrape_url Link to the noscrape page
2016-07-07 23:04:30 +02:00
* @param array $data The already fetched data
*
* @return array noscrape data
*/
private static function pollNoscrape($noscrape_url, $data) {
$ret = z_fetch_url($noscrape_url);
2017-02-01 22:35:01 +01:00
if ($ret['errno'] == CURLE_OPERATION_TIMEDOUT) {
return false;
}
$content = $ret['body'];
if (!$content) {
2017-05-23 22:19:39 +02:00
logger("Empty body for ".$noscrape_url, LOGGER_DEBUG);
2016-07-04 00:14:08 +02:00
return false;
2017-02-01 22:35:01 +01:00
}
2016-07-03 23:11:21 +02:00
$json = json_decode($content, true);
2017-05-08 17:06:57 +02:00
if (!is_array($json)) {
2017-05-23 22:19:39 +02:00
logger("No json data for ".$noscrape_url, LOGGER_DEBUG);
2016-07-04 00:14:08 +02:00
return false;
2017-05-08 17:06:57 +02:00
}
2016-07-03 23:11:21 +02:00
if (!empty($json["fn"])) {
2016-07-03 23:11:21 +02:00
$data["name"] = $json["fn"];
2017-05-08 17:06:57 +02:00
}
2016-07-03 23:11:21 +02:00
if (!empty($json["addr"])) {
2016-07-03 23:11:21 +02:00
$data["addr"] = $json["addr"];
2017-05-08 17:06:57 +02:00
}
2016-07-03 23:11:21 +02:00
if (!empty($json["nick"])) {
2016-07-03 23:11:21 +02:00
$data["nick"] = $json["nick"];
2017-05-08 17:06:57 +02:00
}
2016-07-03 23:11:21 +02:00
if (!empty($json["comm"])) {
2016-07-03 23:11:21 +02:00
$data["community"] = $json["comm"];
2017-05-08 17:06:57 +02:00
}
2016-07-03 23:11:21 +02:00
if (!empty($json["tags"])) {
2016-07-03 23:11:21 +02:00
$keywords = implode(" ", $json["tags"]);
2017-05-08 17:06:57 +02:00
if ($keywords != "") {
2016-07-03 23:11:21 +02:00
$data["keywords"] = $keywords;
2017-05-08 17:06:57 +02:00
}
2016-07-03 23:11:21 +02:00
}
$location = formatted_location($json);
2017-05-08 17:06:57 +02:00
if ($location) {
2016-07-03 23:11:21 +02:00
$data["location"] = $location;
2017-05-08 17:06:57 +02:00
}
2016-07-03 23:11:21 +02:00
if (!empty($json["about"])) {
2016-07-03 23:11:21 +02:00
$data["about"] = $json["about"];
2017-05-08 17:06:57 +02:00
}
2016-07-03 23:11:21 +02:00
if (!empty($json["key"])) {
2016-07-03 23:11:21 +02:00
$data["pubkey"] = $json["key"];
2017-05-08 17:06:57 +02:00
}
2016-07-03 23:11:21 +02:00
if (!empty($json["photo"])) {
2016-07-03 23:11:21 +02:00
$data["photo"] = $json["photo"];
2017-05-08 17:06:57 +02:00
}
2016-07-03 23:11:21 +02:00
if (!empty($json["dfrn-request"])) {
2016-07-03 23:11:21 +02:00
$data["request"] = $json["dfrn-request"];
2017-05-08 17:06:57 +02:00
}
2016-07-03 23:11:21 +02:00
if (!empty($json["dfrn-confirm"])) {
2016-07-03 23:11:21 +02:00
$data["confirm"] = $json["dfrn-confirm"];
2017-05-08 17:06:57 +02:00
}
2016-07-03 23:11:21 +02:00
if (!empty($json["dfrn-notify"])) {
2016-07-03 23:11:21 +02:00
$data["notify"] = $json["dfrn-notify"];
2017-05-08 17:06:57 +02:00
}
2016-07-03 23:11:21 +02:00
if (!empty($json["dfrn-poll"])) {
2016-07-03 23:11:21 +02:00
$data["poll"] = $json["dfrn-poll"];
2017-05-08 17:06:57 +02:00
}
2016-07-03 23:11:21 +02:00
return $data;
}
2016-07-07 23:04:30 +02:00
/**
* @brief Check for valid DFRN data
*
* @param array $data DFRN data
*
* @return int Number of errors
*/
public static function validDfrn($data) {
$errors = 0;
2017-05-08 17:06:57 +02:00
if (!isset($data['key'])) {
$errors ++;
2017-05-08 17:06:57 +02:00
}
if (!isset($data['dfrn-request'])) {
$errors ++;
2017-05-08 17:06:57 +02:00
}
if (!isset($data['dfrn-confirm'])) {
$errors ++;
2017-05-08 17:06:57 +02:00
}
if (!isset($data['dfrn-notify'])) {
$errors ++;
2017-05-08 17:06:57 +02:00
}
if (!isset($data['dfrn-poll'])) {
$errors ++;
2017-05-08 17:06:57 +02:00
}
return $errors;
}
2016-07-07 23:04:30 +02:00
/**
2016-07-10 18:44:48 +02:00
* @brief Fetch data from a DFRN profile page and via "noscrape"
2016-07-07 23:04:30 +02:00
*
* @param string $profile_link Link to the profile page
2016-07-07 23:04:30 +02:00
*
* @return array profile data
*/
public static function profile($profile_link) {
$data = array();
logger("Check profile ".$profile_link, LOGGER_DEBUG);
// Fetch data via noscrape - this is faster
$noscrape_url = str_replace(array("/hcard/", "/profile/"), "/noscrape/", $profile_link);
$data = self::pollNoscrape($noscrape_url, $data);
2017-05-08 17:36:04 +02:00
if (!isset($data["notify"])
|| !isset($data["confirm"])
|| !isset($data["request"])
|| !isset($data["poll"])
|| !isset($data["poco"])
|| !isset($data["name"])
|| !isset($data["photo"])
2017-05-08 17:36:04 +02:00
) {
$data = self::pollHcard($profile_link, $data, true);
2017-05-08 17:06:57 +02:00
}
$prof_data = array();
2017-05-08 18:28:30 +02:00
$prof_data["addr"] = $data["addr"];
$prof_data["nick"] = $data["nick"];
$prof_data["dfrn-request"] = $data["request"];
$prof_data["dfrn-confirm"] = $data["confirm"];
2017-05-08 18:28:30 +02:00
$prof_data["dfrn-notify"] = $data["notify"];
$prof_data["dfrn-poll"] = $data["poll"];
$prof_data["dfrn-poco"] = $data["poco"];
$prof_data["photo"] = $data["photo"];
$prof_data["fn"] = $data["name"];
$prof_data["key"] = $data["pubkey"];
logger("Result for profile ".$profile_link.": ".print_r($prof_data, true), LOGGER_DEBUG);
return $prof_data;
}
2016-07-07 23:04:30 +02:00
/**
* @brief Check for DFRN contact
*
* @param array $webfinger Webfinger data
*
* @return array DFRN data
*/
private static function dfrn($webfinger) {
2016-07-03 22:27:16 +02:00
$hcard_url = "";
2016-07-03 22:27:16 +02:00
$data = array();
2017-05-08 17:06:57 +02:00
foreach ($webfinger["links"] as $link) {
if (($link["rel"] == NAMESPACE_DFRN) && ($link["href"] != "")) {
2016-07-03 22:27:16 +02:00
$data["network"] = NETWORK_DFRN;
} elseif (($link["rel"] == NAMESPACE_FEED) && ($link["href"] != "")) {
2016-07-03 22:27:16 +02:00
$data["poll"] = $link["href"];
} elseif (($link["rel"] == "http://webfinger.net/rel/profile-page") && ($link["type"] == "text/html") && ($link["href"] != "")) {
2016-07-03 22:27:16 +02:00
$data["url"] = $link["href"];
} elseif (($link["rel"] == "http://microformats.org/profile/hcard") && ($link["href"] != "")) {
$hcard_url = $link["href"];
} elseif (($link["rel"] == NAMESPACE_POCO) && ($link["href"] != "")) {
2016-07-03 22:27:16 +02:00
$data["poco"] = $link["href"];
} elseif (($link["rel"] == "http://webfinger.net/rel/avatar") && ($link["href"] != "")) {
2016-07-03 22:27:16 +02:00
$data["photo"] = $link["href"];
} elseif (($link["rel"] == "http://joindiaspora.com/seed_location") && ($link["href"] != "")) {
2016-07-03 22:27:16 +02:00
$data["baseurl"] = trim($link["href"], '/');
} elseif (($link["rel"] == "http://joindiaspora.com/guid") && ($link["href"] != "")) {
2016-07-03 22:27:16 +02:00
$data["guid"] = $link["href"];
} elseif (($link["rel"] == "diaspora-public-key") && ($link["href"] != "")) {
2016-07-03 22:27:16 +02:00
$data["pubkey"] = base64_decode($link["href"]);
//if (strstr($data["pubkey"], 'RSA ') || ($link["type"] == "RSA"))
2017-05-08 17:06:57 +02:00
if (strstr($data["pubkey"], 'RSA ')) {
2016-07-03 22:27:16 +02:00
$data["pubkey"] = rsatopem($data["pubkey"]);
2017-05-08 17:06:57 +02:00
}
2016-07-03 22:27:16 +02:00
}
}
if (is_array($webfinger["aliases"])) {
foreach ($webfinger["aliases"] as $alias) {
if (substr($alias, 0, 5) == 'acct:') {
$data["addr"] = substr($alias, 5);
}
2017-07-11 00:35:32 +02:00
}
}
if (!isset($data["network"]) || ($hcard_url == "")) {
2016-07-03 22:27:16 +02:00
return false;
2017-05-08 17:06:57 +02:00
}
2016-07-03 22:27:16 +02:00
2016-07-03 23:11:21 +02:00
// Fetch data via noscrape - this is faster
$noscrape_url = str_replace("/hcard/", "/noscrape/", $hcard_url);
$data = self::pollNoscrape($noscrape_url, $data);
2016-07-03 23:11:21 +02:00
2017-05-08 17:36:04 +02:00
if (isset($data["notify"])
&& isset($data["confirm"])
&& isset($data["request"])
&& isset($data["poll"])
&& isset($data["name"])
&& isset($data["photo"])
2017-05-08 17:36:04 +02:00
) {
2016-07-03 23:11:21 +02:00
return $data;
2017-05-08 17:06:57 +02:00
}
2016-07-03 23:11:21 +02:00
$data = self::pollHcard($hcard_url, $data, true);
2016-07-03 22:27:16 +02:00
return $data;
}
2016-07-07 23:04:30 +02:00
/**
* @brief Poll the hcard page (Diaspora and Friendica specific)
*
* @param string $hcard_url Link to the hcard page
2016-07-07 23:04:30 +02:00
* @param array $data The already fetched data
* @param boolean $dfrn Poll DFRN specific data
*
* @return array hcard data
*/
private static function pollHcard($hcard_url, $data, $dfrn = false) {
$ret = z_fetch_url($hcard_url);
2017-02-01 22:35:01 +01:00
if ($ret['errno'] == CURLE_OPERATION_TIMEDOUT) {
return false;
}
$content = $ret['body'];
if (!$content) {
return false;
2017-02-01 22:35:01 +01:00
}
2017-05-11 17:53:04 +02:00
$doc = new DOMDocument();
2017-05-08 17:06:57 +02:00
if (!@$doc->loadHTML($content)) {
2016-07-03 22:27:16 +02:00
return false;
2017-05-08 17:06:57 +02:00
}
2016-07-03 22:27:16 +02:00
2017-05-11 17:53:04 +02:00
$xpath = new DomXPath($doc);
2016-07-03 22:27:16 +02:00
$vcards = $xpath->query("//div[contains(concat(' ', @class, ' '), ' vcard ')]");
2017-05-08 17:06:57 +02:00
if (!is_object($vcards)) {
2016-07-03 22:27:16 +02:00
return false;
2017-05-08 17:06:57 +02:00
}
2016-07-03 22:27:16 +02:00
if ($vcards->length > 0) {
$vcard = $vcards->item(0);
2016-07-03 22:27:16 +02:00
// We have to discard the guid from the hcard in favour of the guid from lrdd
// Reason: Hubzilla doesn't use the value "uid" in the hcard like Diaspora does.
$search = $xpath->query("//*[contains(concat(' ', @class, ' '), ' uid ')]", $vcard); // */
if (($search->length > 0) && ($data["guid"] == "")) {
$data["guid"] = $search->item(0)->nodeValue;
2017-05-08 17:06:57 +02:00
}
2016-07-03 22:27:16 +02:00
$search = $xpath->query("//*[contains(concat(' ', @class, ' '), ' nickname ')]", $vcard); // */
2017-05-08 17:06:57 +02:00
if ($search->length > 0) {
$data["nick"] = $search->item(0)->nodeValue;
2017-05-08 17:06:57 +02:00
}
2016-07-03 22:27:16 +02:00
$search = $xpath->query("//*[contains(concat(' ', @class, ' '), ' fn ')]", $vcard); // */
2017-05-08 17:06:57 +02:00
if ($search->length > 0) {
$data["name"] = $search->item(0)->nodeValue;
2017-05-08 17:06:57 +02:00
}
2016-07-03 22:27:16 +02:00
$search = $xpath->query("//*[contains(concat(' ', @class, ' '), ' searchable ')]", $vcard); // */
2017-05-08 17:06:57 +02:00
if ($search->length > 0) {
$data["searchable"] = $search->item(0)->nodeValue;
2017-05-08 17:06:57 +02:00
}
2016-07-03 22:27:16 +02:00
$search = $xpath->query("//*[contains(concat(' ', @class, ' '), ' key ')]", $vcard); // */
if ($search->length > 0) {
$data["pubkey"] = $search->item(0)->nodeValue;
2017-05-08 17:06:57 +02:00
if (strstr($data["pubkey"], 'RSA ')) {
$data["pubkey"] = rsatopem($data["pubkey"]);
2017-05-08 17:06:57 +02:00
}
}
2016-07-03 22:27:16 +02:00
$search = $xpath->query("//*[@id='pod_location']", $vcard); // */
2017-05-08 17:06:57 +02:00
if ($search->length > 0) {
$data["baseurl"] = trim($search->item(0)->nodeValue, "/");
2017-05-08 17:06:57 +02:00
}
2016-07-03 22:27:16 +02:00
}
$avatar = array();
$photos = $xpath->query("//*[contains(concat(' ', @class, ' '), ' photo ') or contains(concat(' ', @class, ' '), ' avatar ')]", $vcard); // */
2017-05-08 17:06:57 +02:00
foreach ($photos as $photo) {
2016-07-03 22:27:16 +02:00
$attr = array();
foreach ($photo->attributes as $attribute) {
2016-07-03 22:27:16 +02:00
$attr[$attribute->name] = trim($attribute->value);
}
2016-07-03 22:27:16 +02:00
if (isset($attr["src"]) && isset($attr["width"])) {
2016-07-03 22:27:16 +02:00
$avatar[$attr["width"]] = $attr["src"];
}
// We don't have a width. So we just take everything that we got.
// This is a Hubzilla workaround which doesn't send a width.
if ((sizeof($avatar) == 0) && !empty($attr["src"])) {
$avatar[] = $attr["src"];
}
2016-07-03 22:27:16 +02:00
}
if (sizeof($avatar)) {
ksort($avatar);
$data["photo"] = self::fixAvatar(array_pop($avatar), $data["baseurl"]);
2016-07-03 22:27:16 +02:00
}
if ($dfrn) {
// Poll DFRN specific data
$search = $xpath->query("//link[contains(concat(' ', @rel), ' dfrn-')]");
if ($search->length > 0) {
2017-05-08 17:06:57 +02:00
foreach ($search as $link) {
2016-07-03 22:27:16 +02:00
//$data["request"] = $search->item(0)->nodeValue;
$attr = array();
2017-05-08 17:06:57 +02:00
foreach ($link->attributes as $attribute) {
2016-07-03 22:27:16 +02:00
$attr[$attribute->name] = trim($attribute->value);
2017-05-08 17:06:57 +02:00
}
2016-07-03 22:27:16 +02:00
$data[substr($attr["rel"], 5)] = $attr["href"];
}
}
// Older Friendica versions had used the "uid" field differently than newer versions
2017-05-08 17:06:57 +02:00
if ($data["nick"] == $data["guid"]) {
2016-07-03 22:27:16 +02:00
unset($data["guid"]);
2017-05-08 17:06:57 +02:00
}
2016-07-03 22:27:16 +02:00
}
return $data;
}
2016-07-07 23:04:30 +02:00
/**
* @brief Check for Diaspora contact
*
* @param array $webfinger Webfinger data
*
* @return array Diaspora data
*/
private static function diaspora($webfinger) {
2016-07-03 22:27:16 +02:00
$hcard_url = "";
2016-07-03 22:27:16 +02:00
$data = array();
2017-05-08 17:06:57 +02:00
foreach ($webfinger["links"] as $link) {
if (($link["rel"] == "http://microformats.org/profile/hcard") && ($link["href"] != "")) {
$hcard_url = $link["href"];
} elseif (($link["rel"] == "http://joindiaspora.com/seed_location") && ($link["href"] != "")) {
2016-07-03 22:27:16 +02:00
$data["baseurl"] = trim($link["href"], '/');
} elseif (($link["rel"] == "http://joindiaspora.com/guid") && ($link["href"] != "")) {
2016-07-03 22:27:16 +02:00
$data["guid"] = $link["href"];
} elseif (($link["rel"] == "http://webfinger.net/rel/profile-page") && ($link["type"] == "text/html") && ($link["href"] != "")) {
2016-07-03 22:27:16 +02:00
$data["url"] = $link["href"];
} elseif (($link["rel"] == NAMESPACE_FEED) && ($link["href"] != "")) {
2016-07-03 22:27:16 +02:00
$data["poll"] = $link["href"];
} elseif (($link["rel"] == NAMESPACE_POCO) && ($link["href"] != "")) {
2016-07-03 22:27:16 +02:00
$data["poco"] = $link["href"];
} elseif (($link["rel"] == "salmon") && ($link["href"] != "")) {
2016-07-03 22:27:16 +02:00
$data["notify"] = $link["href"];
} elseif (($link["rel"] == "diaspora-public-key") && ($link["href"] != "")) {
2016-07-03 22:27:16 +02:00
$data["pubkey"] = base64_decode($link["href"]);
//if (strstr($data["pubkey"], 'RSA ') || ($link["type"] == "RSA"))
2017-05-08 17:06:57 +02:00
if (strstr($data["pubkey"], 'RSA ')) {
2016-07-03 22:27:16 +02:00
$data["pubkey"] = rsatopem($data["pubkey"]);
2017-05-08 17:06:57 +02:00
}
2016-07-03 22:27:16 +02:00
}
}
if (!isset($data["url"]) || ($hcard_url == "")) {
2016-07-03 22:27:16 +02:00
return false;
2017-05-08 17:06:57 +02:00
}
2016-07-03 22:27:16 +02:00
2017-05-08 17:06:57 +02:00
if (is_array($webfinger["aliases"])) {
foreach ($webfinger["aliases"] as $alias) {
if (normalise_link($alias) != normalise_link($data["url"]) && ! strstr($alias, "@")) {
2016-07-03 22:27:16 +02:00
$data["alias"] = $alias;
2017-05-08 17:06:57 +02:00
}
}
}
2016-07-03 22:27:16 +02:00
// Fetch further information from the hcard
$data = self::pollHcard($hcard_url, $data);
2016-07-03 22:27:16 +02:00
2017-05-08 17:06:57 +02:00
if (!$data) {
2016-07-03 22:27:16 +02:00
return false;
2017-05-08 17:06:57 +02:00
}
2016-07-03 22:27:16 +02:00
2017-05-08 17:36:04 +02:00
if (isset($data["url"])
&& isset($data["guid"])
&& isset($data["baseurl"])
&& isset($data["pubkey"])
&& ($hcard_url != "")
2017-05-08 17:36:04 +02:00
) {
2016-07-03 22:27:16 +02:00
$data["network"] = NETWORK_DIASPORA;
// The Diaspora handle must always be lowercase
$data["addr"] = strtolower($data["addr"]);
2016-07-03 22:27:16 +02:00
// We have to overwrite the detected value for "notify" since Hubzilla doesn't send it
2017-05-08 17:06:57 +02:00
$data["notify"] = $data["baseurl"] . "/receive/users/" . $data["guid"];
2017-05-08 18:28:30 +02:00
$data["batch"] = $data["baseurl"] . "/receive/public";
2017-05-08 17:06:57 +02:00
} else {
2016-07-03 22:27:16 +02:00
return false;
2017-05-08 17:06:57 +02:00
}
2016-07-03 22:27:16 +02:00
return $data;
}
2016-07-07 23:04:30 +02:00
/**
* @brief Check for OStatus contact
*
* @param array $webfinger Webfinger data
* @param bool $short Short detection mode
2016-07-07 23:04:30 +02:00
*
* @return array|bool OStatus data or "false" on error or "true" on short mode
2016-07-07 23:04:30 +02:00
*/
private static function ostatus($webfinger, $short = false) {
2016-07-03 22:27:16 +02:00
$data = array();
if (is_array($webfinger["aliases"])) {
2017-05-08 17:06:57 +02:00
foreach ($webfinger["aliases"] as $alias) {
if (strstr($alias, "@")) {
$data["addr"] = str_replace('acct:', '', $alias);
}
}
}
if (is_string($webfinger["subject"]) && strstr($webfinger["subject"], "@")) {
$data["addr"] = str_replace('acct:', '', $webfinger["subject"]);
}
$pubkey = "";
2017-08-29 06:51:02 +02:00
if (is_array($webfinger["links"])) {
foreach ($webfinger["links"] as $link) {
if (($link["rel"] == "http://webfinger.net/rel/profile-page")
&& ($link["type"] == "text/html")
&& ($link["href"] != "")
) {
$data["url"] = $link["href"];
} elseif (($link["rel"] == "salmon") && ($link["href"] != "")) {
$data["notify"] = $link["href"];
} elseif (($link["rel"] == NAMESPACE_FEED) && ($link["href"] != "")) {
$data["poll"] = $link["href"];
} elseif (($link["rel"] == "magic-public-key") && ($link["href"] != "")) {
$pubkey = $link["href"];
if (substr($pubkey, 0, 5) === 'data:') {
if (strstr($pubkey, ',')) {
$pubkey = substr($pubkey, strpos($pubkey, ',') + 1);
} else {
$pubkey = substr($pubkey, 5);
}
} elseif (normalise_link($pubkey) == 'http://') {
$ret = z_fetch_url($pubkey);
if ($ret['errno'] == CURLE_OPERATION_TIMEDOUT) {
return false;
}
$pubkey = $ret['body'];
}
2016-07-03 22:27:16 +02:00
2017-08-29 06:51:02 +02:00
$key = explode(".", $pubkey);
2016-07-03 22:27:16 +02:00
2017-08-29 06:51:02 +02:00
if (sizeof($key) >= 3) {
$m = base64url_decode($key[1]);
$e = base64url_decode($key[2]);
$data["pubkey"] = metopem($m, $e);
}
2016-07-03 22:27:16 +02:00
}
}
}
if (isset($data["notify"]) && isset($data["pubkey"])
&& isset($data["poll"])
&& isset($data["url"])
2017-05-08 17:36:04 +02:00
) {
2016-07-03 22:27:16 +02:00
$data["network"] = NETWORK_OSTATUS;
} else {
2016-07-03 22:27:16 +02:00
return false;
}
if ($short) {
return true;
}
2016-07-03 22:27:16 +02:00
// Fetch all additional data from the feed
$ret = z_fetch_url($data["poll"]);
if ($ret['errno'] == CURLE_OPERATION_TIMEDOUT) {
return false;
}
$feed = $ret['body'];
2017-05-08 17:06:57 +02:00
$feed_data = feed_import($feed, $dummy1, $dummy2, $dummy3, true);
if (!$feed_data) {
2016-07-03 22:27:16 +02:00
return false;
}
2017-08-21 22:21:04 +02:00
if ($feed_data["header"]["author-name"] != "") {
2016-07-03 22:27:16 +02:00
$data["name"] = $feed_data["header"]["author-name"];
}
if ($feed_data["header"]["author-nick"] != "") {
2016-07-03 22:27:16 +02:00
$data["nick"] = $feed_data["header"]["author-nick"];
}
if ($feed_data["header"]["author-avatar"] != "") {
$data["photo"] = self::fixAvatar($feed_data["header"]["author-avatar"], $data["url"]);
}
if ($feed_data["header"]["author-id"] != "") {
2016-07-03 22:27:16 +02:00
$data["alias"] = $feed_data["header"]["author-id"];
}
if ($feed_data["header"]["author-location"] != "") {
$data["location"] = $feed_data["header"]["author-location"];
}
if ($feed_data["header"]["author-about"] != "") {
$data["about"] = $feed_data["header"]["author-about"];
}
2016-07-03 22:27:16 +02:00
// OStatus has serious issues when the the url doesn't fit (ssl vs. non ssl)
// So we take the value that we just fetched, although the other one worked as well
if ($feed_data["header"]["author-link"] != "") {
2016-07-03 22:27:16 +02:00
$data["url"] = $feed_data["header"]["author-link"];
}
2017-08-21 22:21:04 +02:00
if (($data['poll'] == $data['url']) && ($data["alias"] != '')) {
$data['url'] = $data["alias"];
$data["alias"] = '';
}
2016-07-03 22:27:16 +02:00
/// @todo Fetch location and "about" from the feed as well
return $data;
}
2016-07-07 23:04:30 +02:00
/**
* @brief Fetch data from a pump.io profile page
*
* @param string $profile_link Link to the profile page
2016-07-07 23:04:30 +02:00
*
* @return array profile data
*/
private static function pumpioProfileData($profile_link) {
2016-07-03 22:27:16 +02:00
2017-05-11 17:53:04 +02:00
$doc = new DOMDocument();
if (!@$doc->loadHTMLFile($profile_link)) {
2016-07-03 22:27:16 +02:00
return false;
2017-05-08 17:06:57 +02:00
}
2016-07-03 22:27:16 +02:00
2017-05-11 17:53:04 +02:00
$xpath = new DomXPath($doc);
2016-07-03 22:27:16 +02:00
$data = array();
// This is ugly - but pump.io doesn't seem to know a better way for it
$data["name"] = trim($xpath->query("//h1[@class='media-header']")->item(0)->nodeValue);
$pos = strpos($data["name"], chr(10));
2017-05-08 17:06:57 +02:00
if ($pos) {
2016-07-03 22:27:16 +02:00
$data["name"] = trim(substr($data["name"], 0, $pos));
2017-05-08 17:06:57 +02:00
}
2016-07-03 22:27:16 +02:00
$avatar = $xpath->query("//img[@class='img-rounded media-object']")->item(0);
2017-05-08 17:06:57 +02:00
if ($avatar) {
foreach ($avatar->attributes as $attribute) {
if ($attribute->name == "src") {
2016-07-03 22:27:16 +02:00
$data["photo"] = trim($attribute->value);
2017-05-08 17:06:57 +02:00
}
}
}
2016-07-03 22:27:16 +02:00
$data["location"] = $xpath->query("//p[@class='location']")->item(0)->nodeValue;
$data["about"] = $xpath->query("//p[@class='summary']")->item(0)->nodeValue;
return $data;
}
2016-07-07 23:04:30 +02:00
/**
* @brief Check for pump.io contact
*
* @param array $webfinger Webfinger data
*
* @return array pump.io data
*/
private static function pumpio($webfinger) {
2016-07-03 22:27:16 +02:00
$data = array();
2017-05-08 17:06:57 +02:00
foreach ($webfinger["links"] as $link) {
2017-05-08 17:36:04 +02:00
if (($link["rel"] == "http://webfinger.net/rel/profile-page")
&& ($link["type"] == "text/html")
&& ($link["href"] != "")
2017-05-08 17:36:04 +02:00
) {
2016-07-03 22:27:16 +02:00
$data["url"] = $link["href"];
} elseif (($link["rel"] == "activity-inbox") && ($link["href"] != "")) {
$data["notify"] = $link["href"];
} elseif (($link["rel"] == "activity-outbox") && ($link["href"] != "")) {
$data["poll"] = $link["href"];
} elseif (($link["rel"] == "dialback") && ($link["href"] != "")) {
2016-07-03 22:27:16 +02:00
$data["dialback"] = $link["href"];
2017-05-08 17:06:57 +02:00
}
2016-07-03 22:27:16 +02:00
}
if (isset($data["poll"]) && isset($data["notify"])
&& isset($data["dialback"])
&& isset($data["url"])
2017-05-08 17:36:04 +02:00
) {
2016-07-03 22:27:16 +02:00
// by now we use these fields only for the network type detection
// So we unset all data that isn't used at the moment
unset($data["dialback"]);
$data["network"] = NETWORK_PUMPIO;
2017-05-08 17:06:57 +02:00
} else {
2016-07-03 22:27:16 +02:00
return false;
2017-05-08 17:06:57 +02:00
}
2016-07-03 22:27:16 +02:00
$profile_data = self::pumpioProfileData($data["url"]);
2016-07-03 22:27:16 +02:00
2017-05-08 17:06:57 +02:00
if (!$profile_data) {
2016-07-03 22:27:16 +02:00
return false;
2017-05-08 17:06:57 +02:00
}
2016-07-03 22:27:16 +02:00
$data = array_merge($data, $profile_data);
return $data;
}
2016-07-07 23:04:30 +02:00
/**
* @brief Check page for feed link
*
* @param string $url Page link
*
* @return string feed link
*/
private static function getFeedLink($url) {
2017-05-11 17:53:04 +02:00
$doc = new DOMDocument();
2017-05-08 17:06:57 +02:00
if (!@$doc->loadHTMLFile($url)) {
return false;
2017-05-08 17:06:57 +02:00
}
2017-05-11 17:53:04 +02:00
$xpath = new DomXPath($doc);
//$feeds = $xpath->query("/html/head/link[@type='application/rss+xml']");
$feeds = $xpath->query("/html/head/link[@type='application/rss+xml' and @rel='alternate']");
2017-05-08 17:06:57 +02:00
if (!is_object($feeds)) {
return false;
2017-05-08 17:06:57 +02:00
}
2017-05-08 17:06:57 +02:00
if ($feeds->length == 0) {
return false;
2017-05-08 17:06:57 +02:00
}
$feed_url = "";
2017-05-08 17:06:57 +02:00
foreach ($feeds as $feed) {
$attr = array();
2017-05-08 17:06:57 +02:00
foreach ($feed->attributes as $attribute) {
$attr[$attribute->name] = trim($attribute->value);
}
2017-05-08 17:06:57 +02:00
if ($feed_url == "") {
$feed_url = $attr["href"];
2017-05-08 17:06:57 +02:00
}
}
return $feed_url;
}
2016-07-07 23:04:30 +02:00
/**
* @brief Check for feed contact
*
* @param string $url Profile link
* @param boolean $probe Do a probe if the page contains a feed link
*
* @return array feed data
*/
private static function feed($url, $probe = true) {
2017-02-01 22:35:01 +01:00
$ret = z_fetch_url($url);
if ($ret['errno'] == CURLE_OPERATION_TIMEDOUT) {
return false;
}
$feed = $ret['body'];
2016-07-03 22:27:16 +02:00
$feed_data = feed_import($feed, $dummy1, $dummy2, $dummy3, true);
if (!$feed_data) {
2017-05-08 17:06:57 +02:00
if (!$probe) {
return false;
2017-05-08 17:06:57 +02:00
}
$feed_url = self::getFeedLink($url);
2017-05-08 17:06:57 +02:00
if (!$feed_url) {
return false;
2017-05-08 17:06:57 +02:00
}
return self::feed($feed_url, false);
}
2016-07-03 22:27:16 +02:00
2017-05-08 17:06:57 +02:00
if ($feed_data["header"]["author-name"] != "") {
2016-07-03 22:27:16 +02:00
$data["name"] = $feed_data["header"]["author-name"];
2017-05-08 17:06:57 +02:00
}
2016-07-03 22:27:16 +02:00
2017-05-08 17:06:57 +02:00
if ($feed_data["header"]["author-nick"] != "") {
2016-07-03 22:27:16 +02:00
$data["nick"] = $feed_data["header"]["author-nick"];
2017-05-08 17:06:57 +02:00
}
2016-07-03 22:27:16 +02:00
2017-05-08 17:06:57 +02:00
if ($feed_data["header"]["author-avatar"] != "") {
2016-07-03 22:27:16 +02:00
$data["photo"] = $feed_data["header"]["author-avatar"];
2017-05-08 17:06:57 +02:00
}
2016-07-03 22:27:16 +02:00
2017-05-08 17:06:57 +02:00
if ($feed_data["header"]["author-id"] != "") {
2016-07-03 22:27:16 +02:00
$data["alias"] = $feed_data["header"]["author-id"];
2017-05-08 17:06:57 +02:00
}
2016-07-03 22:27:16 +02:00
$data["url"] = $url;
$data["poll"] = $url;
2017-05-08 17:06:57 +02:00
if ($feed_data["header"]["author-link"] != "") {
2016-07-03 22:27:16 +02:00
$data["baseurl"] = $feed_data["header"]["author-link"];
2017-05-08 17:06:57 +02:00
} else {
2016-07-03 22:27:16 +02:00
$data["baseurl"] = $data["url"];
2017-05-08 17:06:57 +02:00
}
2016-07-03 22:27:16 +02:00
$data["network"] = NETWORK_FEED;
return $data;
}
2016-07-07 23:04:30 +02:00
/**
* @brief Check for mail contact
*
* @param string $uri Profile link
* @param integer $uid User ID
*
* @return array mail data
*/
private static function mail($uri, $uid) {
2016-07-03 22:27:16 +02:00
2017-05-08 17:06:57 +02:00
if (!validate_email($uri)) {
2016-07-03 22:27:16 +02:00
return false;
2017-05-08 17:06:57 +02:00
}
2016-07-03 22:27:16 +02:00
$x = q("SELECT `prvkey` FROM `user` WHERE `uid` = %d LIMIT 1", intval($uid));
$r = q("SELECT * FROM `mailacct` WHERE `uid` = %d AND `server` != '' LIMIT 1", intval($uid));
2017-05-11 17:53:04 +02:00
if (dbm::is_result($x) && dbm::is_result($r)) {
2016-07-03 22:27:16 +02:00
$mailbox = construct_mailbox_name($r[0]);
$password = '';
2017-05-08 17:06:57 +02:00
openssl_private_decrypt(hex2bin($r[0]['pass']), $password, $x[0]['prvkey']);
$mbox = email_connect($mailbox, $r[0]['user'], $password);
if (!mbox) {
2016-07-03 22:27:16 +02:00
return false;
2017-05-08 17:06:57 +02:00
}
2016-07-03 22:27:16 +02:00
}
$msgs = email_poll($mbox, $uri);
logger('searching '.$uri.', '.count($msgs).' messages found.', LOGGER_DEBUG);
2017-05-08 17:06:57 +02:00
if (!count($msgs)) {
2016-07-03 22:27:16 +02:00
return false;
2017-05-08 17:06:57 +02:00
}
2016-07-03 22:27:16 +02:00
2017-05-08 18:28:30 +02:00
$phost = substr($uri, strpos($uri, '@') + 1);
2016-07-03 22:27:16 +02:00
2017-05-08 18:28:30 +02:00
$data = array();
$data["addr"] = $uri;
2016-07-03 22:27:16 +02:00
$data["network"] = NETWORK_MAIL;
2017-05-08 18:28:30 +02:00
$data["name"] = substr($uri, 0, strpos($uri, '@'));
$data["nick"] = $data["name"];
$data["photo"] = avatar_img($uri);
$data["url"] = 'http://'.$phost."/".$data["nick"];
$data["notify"] = 'smtp '.random_string();
$data["poll"] = 'email '.random_string();
2016-07-03 22:27:16 +02:00
$x = email_msg_meta($mbox, $msgs[0]);
2017-05-08 17:06:57 +02:00
if (stristr($x[0]->from, $uri)) {
2016-07-03 22:27:16 +02:00
$adr = imap_rfc822_parse_adrlist($x[0]->from, '');
2017-05-08 17:06:57 +02:00
} elseif (stristr($x[0]->to, $uri)) {
2016-07-03 22:27:16 +02:00
$adr = imap_rfc822_parse_adrlist($x[0]->to, '');
2017-05-08 17:06:57 +02:00
}
if (isset($adr)) {
foreach ($adr as $feadr) {
if ((strcasecmp($feadr->mailbox, $data["name"]) == 0)
2016-07-03 22:27:16 +02:00
&&(strcasecmp($feadr->host, $phost) == 0)
2017-05-08 17:36:04 +02:00
&& (strlen($feadr->personal))
) {
2016-07-03 22:27:16 +02:00
$personal = imap_mime_header_decode($feadr->personal);
$data["name"] = "";
2017-05-08 17:06:57 +02:00
foreach ($personal as $perspart) {
if ($perspart->charset != "default") {
2016-07-03 22:27:16 +02:00
$data["name"] .= iconv($perspart->charset, 'UTF-8//IGNORE', $perspart->text);
2017-05-08 17:06:57 +02:00
} else {
2016-07-03 22:27:16 +02:00
$data["name"] .= $perspart->text;
2017-05-08 17:06:57 +02:00
}
}
2016-07-03 22:27:16 +02:00
$data["name"] = notags($data["name"]);
}
}
}
imap_close($mbox);
return $data;
}
/**
* @brief Mix two paths together to possibly fix missing parts
*
* @param string $avatar Path to the avatar
* @param string $base Another path that is hopefully complete
*
* @return string fixed avatar path
*/
public static function fixAvatar($avatar, $base) {
$base_parts = parse_url($base);
// Remove all parts that could create a problem
unset($base_parts['path']);
unset($base_parts['query']);
unset($base_parts['fragment']);
$avatar_parts = parse_url($avatar);
// Now we mix them
$parts = array_merge($base_parts, $avatar_parts);
// And put them together again
2017-05-08 17:36:04 +02:00
$scheme = isset($parts['scheme']) ? $parts['scheme'] . '://' : '';
$host = isset($parts['host']) ? $parts['host'] : '';
$port = isset($parts['port']) ? ':' . $parts['port'] : '';
$path = isset($parts['path']) ? $parts['path'] : '';
$query = isset($parts['query']) ? '?' . $parts['query'] : '';
$fragment = isset($parts['fragment']) ? '#' . $parts['fragment'] : '';
$fixed = $scheme.$host.$port.$path.$query.$fragment;
logger('Base: '.$base.' - Avatar: '.$avatar.' - Fixed: '.$fixed, LOGGER_DATA);
return $fixed;
}
2016-07-03 22:27:16 +02:00
}