<?php
/**
- * @brief This class contain functions for probing URL
+ * @file include/Probe.php
+ * @brief Functions for probing URL
*
*/
require_once('include/email.php');
require_once('include/network.php');
+/**
+ * @brief This class contain functions for probing URL
+ *
+ */
class Probe {
+ /**
+ * @brief Rearrange the array so that it always has the same order
+ *
+ * @param array $data Unordered data
+ *
+ * @return array Ordered data
+ */
private function rearrange_data($data) {
$fields = array("name", "nick", "guid", "url", "addr", "alias",
"photo", "community", "keywords", "location", "about",
return false;
$xrd_data = array();
+
foreach ($links["xrd"]["link"] AS $value => $link) {
if (isset($link["@attributes"]))
$attributes = $link["@attributes"];
return $xrd_data;
}
- public static function uri($uri, $network = "", $cache = true) {
+ /**
+ * @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 $hcard Link to the hcard - is returned by reference
+ *
+ * @return string profile link
+ */
+
+ public static function webfinger_dfrn($webbie, &$hcard) {
+
+ $profile_link = '';
+
+ $links = self::lrdd($webbie);
+ logger('webfinger_dfrn: '.$webbie.':'.print_r($links,true), LOGGER_DATA);
+ if (count($links)) {
+ foreach ($links as $link) {
+ if ($link['@attributes']['rel'] === NAMESPACE_DFRN)
+ $profile_link = $link['@attributes']['href'];
+ if (($link['@attributes']['rel'] === NAMESPACE_OSTATUSSUB) AND ($profile_link == ""))
+ $profile_link = 'stat:'.$link['@attributes']['template'];
+ if ($link['@attributes']['rel'] === 'http://microformats.org/profile/hcard')
+ $hcard = $link['@attributes']['href'];
+ }
+ }
+ 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);
+
+ if (!$lrdd) {
+ $parts = @parse_url($uri);
+ if (!$parts)
+ return array();
+
+ $host = $parts["host"];
+
+ $path_parts = explode("/", trim($parts["path"], "/"));
+
+ do {
+ $lrdd = self::xrd($host);
+ $host .= "/".array_shift($path_parts);
+ } while (!$lrdd AND (sizeof($path_parts) > 0));
+ }
+
+ if (!$lrdd)
+ return array();
+
+ foreach ($lrdd AS $key => $link) {
+ if ($webfinger)
+ continue;
+
+ if (!in_array($key, array("lrdd", "lrdd-xml", "lrdd-json")))
+ continue;
+
+ $path = str_replace('{uri}', urlencode($uri), $link);
+ $webfinger = self::webfinger($path);
+
+ if (!$webfinger AND (strstr($uri, "@"))) {
+ $path = str_replace('{uri}', urlencode("acct:".$uri), $link);
+ $webfinger = self::webfinger($path);
+ }
+ }
+
+ if (!is_array($webfinger["links"]))
+ return false;
+
+ $data = array();
+
+ foreach ($webfinger["links"] AS $link)
+ $data[] = array("@attributes" => $link);
+
+ if (is_array($webfinger["aliases"]))
+ foreach ($webfinger["aliases"] AS $alias)
+ $data[] = array("@attributes" =>
+ array("rel" => "alias",
+ "href" => $alias));
+
+ return $data;
+ }
+
+ /**
+ * @brief 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
+ * @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, $cache = true) {
if ($cache) {
$result = Cache::get("probe_url:".$network.":".$uri);
if (!is_null($result)) {
- $result = unserialize($result);
return $result;
}
}
- $data = self::detect($uri, $network);
+ if ($uid == 0)
+ $uid = local_user();
+
+ $data = self::detect($uri, $network, $uid);
if (!isset($data["url"]))
$data["url"] = $uri;
else
$data["photo"] = App::get_baseurl().'/images/person-175.jpg';
- if (!isset($data["name"]))
- $data["name"] = $data["url"];
+ if (!isset($data["name"]) OR ($data["name"] == "")) {
+ if (isset($data["nick"]))
+ $data["name"] = $data["nick"];
- if (!isset($data["nick"]))
+ if ($data["name"] == "")
+ $data["name"] = $data["url"];
+ }
+
+ if (!isset($data["nick"]) OR ($data["nick"] == "")) {
$data["nick"] = strtolower($data["name"]);
+ if (strpos($data['nick'], ' '))
+ $data['nick'] = trim(substr($data['nick'], 0, strpos($data['nick'], ' ')));
+ }
+
if (!isset($data["network"]))
$data["network"] = NETWORK_PHANTOM;
$data = self::rearrange_data($data);
// Only store into the cache if the value seems to be valid
- if ($data['network'] != NETWORK_PHANTOM) {
- Cache::set("probe_url:".$network.":".$uri,serialize($data), CACHE_DAY);
+ if (!in_array($data['network'], array(NETWORK_PHANTOM, NETWORK_MAIL))) {
+ Cache::set("probe_url:".$network.":".$uri, $data, CACHE_DAY);
/// @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.
return $data;
}
- private function detect($uri, $network) {
+ /**
+ * @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.
+ *
+ * @param string $uri Address that should be probed
+ * @param string $network Test for this specific network
+ * @param integer $uid User ID for the probe (only used for mails)
+ *
+ * @return array uri data
+ */
+ private function detect($uri, $network, $uid) {
if (strstr($uri, '@')) {
- // If the URI starts with "mailto:" then jum directly to the mail detection
+ // If the URI starts with "mailto:" then jump directly to the mail detection
if (strpos($url,'mailto:') !== false) {
$uri = str_replace('mailto:', '', $url);
- return self::mail($uri);
+ return self::mail($uri, $uid);
}
if ($network == NETWORK_MAIL)
- return self::mail($uri);
+ return self::mail($uri, $uid);
// Remove "acct:" from the URI
$uri = str_replace('acct:', '', $uri);
$host = substr($uri,strpos($uri, '@') + 1);
$nick = substr($uri,0, strpos($uri, '@'));
+ if (strpos($uri, '@twitter.com'))
+ return array("network" => NETWORK_TWITTER);
+
$lrdd = self::xrd($host);
+
if (!$lrdd)
- return self::mail($uri);
+ return self::mail($uri, $uid);
$addr = $uri;
} else {
!isset($parts["path"]))
return false;
- // todo: Ports?
+ /// @todo: Ports?
$host = $parts["host"];
+
+ if ($host == 'twitter.com')
+ return array("network" => NETWORK_TWITTER);
+
$lrdd = self::xrd($host);
$path_parts = explode("/", trim($parts["path"], "/"));
$nick = array_pop($path_parts);
$addr = $nick."@".$host;
}
-
$webfinger = false;
/// @todo Do we need the prefix "acct:" or "acct://"?
if (!in_array($key, array("lrdd", "lrdd-xml", "lrdd-json")))
continue;
+ // Try webfinger with the address (user@domain.tld)
$path = str_replace('{uri}', urlencode($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);
+ }
+
+ // If webfinger wasn't successful then try it with the URL - possibly in the format https://...
+ if (!$webfinger AND ($uri != $addr)) {
+ $path = str_replace('{uri}', urlencode($uri), $link);
+ $webfinger = self::webfinger($path);
+
+ // Since the detection with the address wasn't successful, we delete it.
+ if ($webfinger) {
+ $nick = "";
+ $addr = "";
+ }
+ }
+
}
if (!$webfinger)
return self::feed($uri);
else {
// We overwrite the detected nick with our try if the previois routines hadn't detected it.
// Additionally it is overwritten when the nickname doesn't make sense (contains spaces).
- if (!isset($result["nick"]) OR ($result["nick"] == "") OR (strstr($result["nick"], " ")))
+ if ((!isset($result["nick"]) OR ($result["nick"] == "") OR (strstr($result["nick"], " "))) AND ($nick != ""))
$result["nick"] = $nick;
- if (!isset($result["addr"]) OR ($result["addr"] == ""))
+ if ((!isset($result["addr"]) OR ($result["addr"] == "")) AND ($addr != ""))
$result["addr"] = $addr;
}
return $result;
}
+ /**
+ * @brief Perform a webfinger request.
+ *
+ * For details see RFC 7033: <https://tools.ietf.org/html/rfc7033>
+ *
+ * @param string $url Address that should be probed
+ *
+ * @return array webfinger data
+ */
private function webfinger($url) {
$xrd_timeout = Config::get('system','xrd_timeout', 20);
return $webfinger;
}
+ /**
+ * @brief 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.
+ *
+ * @param string $noscrape Link to the noscrape page
+ * @param array $data The already fetched data
+ *
+ * @return array noscrape data
+ */
private function poll_noscrape($noscrape, $data) {
$content = fetch_url($noscrape);
if (!$content)
return $data;
}
+ /**
+ * @brief Check for valid DFRN data
+ *
+ * @param array $data DFRN data
+ *
+ * @return int Number of errors
+ */
public static function valid_dfrn($data) {
$errors = 0;
if(!isset($data['key']))
return $errors;
}
+ /**
+ * @brief Fetch data from a DFRN profile page and via "noscrape"
+ *
+ * @param string $profile Link to the profile page
+ *
+ * @return array profile data
+ */
public static function profile($profile) {
$data = array();
+ logger("Check profile ".$profile, LOGGER_DEBUG);
+
// Fetch data via noscrape - this is faster
$noscrape = str_replace(array("/hcard/", "/profile/"), "/noscrape/", $profile);
$data = self::poll_noscrape($noscrape, $data);
$prof_data["fn"] = $data["name"];
$prof_data["key"] = $data["pubkey"];
+ logger("Result for profile ".$profile.": ".print_r($prof_data, true), LOGGER_DEBUG);
+
return $prof_data;
}
+ /**
+ * @brief Check for DFRN contact
+ *
+ * @param array $webfinger Webfinger data
+ *
+ * @return array DFRN data
+ */
private function dfrn($webfinger) {
$hcard = "";
return $data;
}
+ /**
+ * @brief Poll the hcard page (Diaspora and Friendica specific)
+ *
+ * @param string $hcard Link to the hcard page
+ * @param array $data The already fetched data
+ * @param boolean $dfrn Poll DFRN specific data
+ *
+ * @return array hcard data
+ */
private function poll_hcard($hcard, $data, $dfrn = false) {
+ $content = fetch_url($hcard);
+ if (!$content)
+ return false;
+
$doc = new DOMDocument();
- if (!@$doc->loadHTMLFile($hcard))
+ if (!@$doc->loadHTML($content))
return false;
$xpath = new DomXPath($doc);
if (!is_object($vcards))
return false;
- if ($vcards->length == 0)
- return false;
+ if ($vcards->length > 0) {
+ $vcard = $vcards->item(0);
- $vcard = $vcards->item(0);
+ // 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) AND ($data["guid"] == ""))
+ $data["guid"] = $search->item(0)->nodeValue;
- // 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) AND ($data["guid"] == ""))
- $data["guid"] = $search->item(0)->nodeValue;
+ $search = $xpath->query("//*[contains(concat(' ', @class, ' '), ' nickname ')]", $vcard); // */
+ if ($search->length > 0)
+ $data["nick"] = $search->item(0)->nodeValue;
- $search = $xpath->query("//*[contains(concat(' ', @class, ' '), ' nickname ')]", $vcard); // */
- if ($search->length > 0)
- $data["nick"] = $search->item(0)->nodeValue;
+ $search = $xpath->query("//*[contains(concat(' ', @class, ' '), ' fn ')]", $vcard); // */
+ if ($search->length > 0)
+ $data["name"] = $search->item(0)->nodeValue;
- $search = $xpath->query("//*[contains(concat(' ', @class, ' '), ' fn ')]", $vcard); // */
- if ($search->length > 0)
- $data["name"] = $search->item(0)->nodeValue;
+ $search = $xpath->query("//*[contains(concat(' ', @class, ' '), ' searchable ')]", $vcard); // */
+ if ($search->length > 0)
+ $data["searchable"] = $search->item(0)->nodeValue;
- $search = $xpath->query("//*[contains(concat(' ', @class, ' '), ' searchable ')]", $vcard); // */
- if ($search->length > 0)
- $data["searchable"] = $search->item(0)->nodeValue;
+ $search = $xpath->query("//*[contains(concat(' ', @class, ' '), ' key ')]", $vcard); // */
+ if ($search->length > 0) {
+ $data["pubkey"] = $search->item(0)->nodeValue;
+ if (strstr($data["pubkey"], 'RSA '))
+ $data["pubkey"] = rsatopem($data["pubkey"]);
+ }
- $search = $xpath->query("//*[contains(concat(' ', @class, ' '), ' key ')]", $vcard); // */
- if ($search->length > 0) {
- $data["pubkey"] = $search->item(0)->nodeValue;
- if (strstr($data["pubkey"], 'RSA '))
- $data["pubkey"] = rsatopem($data["pubkey"]);
+ $search = $xpath->query("//*[@id='pod_location']", $vcard); // */
+ if ($search->length > 0)
+ $data["baseurl"] = trim($search->item(0)->nodeValue, "/");
}
- $search = $xpath->query("//*[@id='pod_location']", $vcard); // */
- if ($search->length > 0)
- $data["baseurl"] = trim($search->item(0)->nodeValue, "/");
-
$avatar = array();
$photos = $xpath->query("//*[contains(concat(' ', @class, ' '), ' photo ') or contains(concat(' ', @class, ' '), ' avatar ')]", $vcard); // */
foreach ($photos AS $photo) {
$attr = array();
- foreach ($photo->attributes as $attribute)
+ foreach ($photo->attributes as $attribute) {
$attr[$attribute->name] = trim($attribute->value);
+ }
- if (isset($attr["src"]) AND isset($attr["width"]))
+ if (isset($attr["src"]) AND isset($attr["width"])) {
$avatar[$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) AND isset($attr["src"])) {
+ $avatar[] = $attr["src"];
+ }
}
if (sizeof($avatar)) {
return $data;
}
+ /**
+ * @brief Check for Diaspora contact
+ *
+ * @param array $webfinger Webfinger data
+ *
+ * @return array Diaspora data
+ */
private function diaspora($webfinger) {
$hcard = "";
isset($data["pubkey"]) AND ($hcard != "")) {
$data["network"] = NETWORK_DIASPORA;
+ // The Diaspora handle must always be lowercase
+ $data["addr"] = strtolower($data["addr"]);
+
// We have to overwrite the detected value for "notify" since Hubzilla doesn't send it
$data["notify"] = $data["baseurl"]."/receive/users/".$data["guid"];
$data["batch"] = $data["baseurl"]."/receive/public";
return $data;
}
+ /**
+ * @brief Check for OStatus contact
+ *
+ * @param array $webfinger Webfinger data
+ *
+ * @return array OStatus data
+ */
private function ostatus($webfinger) {
- $pubkey = "";
$data = array();
+ if (is_array($webfinger["aliases"]))
+ foreach($webfinger["aliases"] AS $alias)
+ if (strstr($alias, "@"))
+ $data["addr"] = str_replace('acct:', '', $alias);
+
+ if (is_string($webfinger["subject"]) AND strstr($webfinger["subject"], "@"))
+ $data["addr"] = str_replace('acct:', '', $webfinger["subject"]);
+
+ $pubkey = "";
foreach ($webfinger["links"] AS $link) {
if (($link["rel"] == "http://webfinger.net/rel/profile-page") AND
($link["type"] == "text/html") AND ($link["href"] != ""))
$pubkey = substr($pubkey, strpos($pubkey, ',') + 1);
else
$pubkey = substr($pubkey, 5);
- } else
+ } elseif (normalise_link($pubkey) == 'http://')
$pubkey = fetch_url($pubkey);
$key = explode(".", $pubkey);
if ($feed_data["header"]["author-id"] != "")
$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"] = $feed_data["header"]["author-about"];
+
// OStatus has serious issues when the the url doesn't fit (ssl vs. non ssl)
// So we take the value that we just fetched, although the other one worked as well
if ($feed_data["header"]["author-link"] != "")
return $data;
}
+ /**
+ * @brief Fetch data from a pump.io profile page
+ *
+ * @param string $profile Link to the profile page
+ *
+ * @return array profile data
+ */
private function pumpio_profile_data($profile) {
$doc = new DOMDocument();
return $data;
}
+ /**
+ * @brief Check for pump.io contact
+ *
+ * @param array $webfinger Webfinger data
+ *
+ * @return array pump.io data
+ */
private function pumpio($webfinger) {
+
$data = array();
foreach ($webfinger["links"] AS $link) {
if (($link["rel"] == "http://webfinger.net/rel/profile-page") AND
($link["type"] == "text/html") AND ($link["href"] != ""))
$data["url"] = $link["href"];
elseif (($link["rel"] == "activity-inbox") AND ($link["href"] != ""))
- $data["activity-inbox"] = $link["href"];
+ $data["notify"] = $link["href"];
elseif (($link["rel"] == "activity-outbox") AND ($link["href"] != ""))
- $data["activity-outbox"] = $link["href"];
+ $data["poll"] = $link["href"];
elseif (($link["rel"] == "dialback") AND ($link["href"] != ""))
$data["dialback"] = $link["href"];
}
- if (isset($data["activity-inbox"]) AND isset($data["activity-outbox"]) AND
+ if (isset($data["poll"]) AND isset($data["notify"]) AND
isset($data["dialback"]) AND isset($data["url"])) {
// by now we use these fields only for the network type detection
// So we unset all data that isn't used at the moment
- unset($data["activity-inbox"]);
- unset($data["activity-outbox"]);
unset($data["dialback"]);
$data["network"] = NETWORK_PUMPIO;
return $data;
}
- private function feed($url) {
+ /**
+ * @brief Check page for feed link
+ *
+ * @param string $url Page link
+ *
+ * @return string feed link
+ */
+ private function get_feed_link($url) {
+ $doc = new DOMDocument();
+
+ if (!@$doc->loadHTMLFile($url))
+ return false;
+
+ $xpath = new DomXPath($doc);
+
+ //$feeds = $xpath->query("/html/head/link[@type='application/rss+xml']");
+ $feeds = $xpath->query("/html/head/link[@type='application/rss+xml' and @rel='alternate']");
+ if (!is_object($feeds))
+ return false;
+
+ if ($feeds->length == 0)
+ return false;
+
+ $feed_url = "";
+
+ foreach ($feeds AS $feed) {
+ $attr = array();
+ foreach ($feed->attributes as $attribute)
+ $attr[$attribute->name] = trim($attribute->value);
+
+ if ($feed_url == "")
+ $feed_url = $attr["href"];
+ }
+
+ return $feed_url;
+ }
+
+ /**
+ * @brief Check for feed contact
+ *
+ * @param string $url Profile link
+ * @param boolean $probe Do a probe if the page contains a feed link
+ *
+ * @return array feed data
+ */
+ private function feed($url, $probe = true) {
$feed = fetch_url($url);
$feed_data = feed_import($feed, $dummy1, $dummy2, $dummy3, true);
- if (!$feed_data)
- return false;
+ if (!$feed_data) {
+ if (!$probe)
+ return false;
+
+ $feed_url = self::get_feed_link($url);
+
+ if (!$feed_url)
+ return false;
+
+ return self::feed($feed_url, false);
+ }
if ($feed_data["header"]["author-name"] != "")
$data["name"] = $feed_data["header"]["author-name"];
return $data;
}
- private function mail($uri) {
+ /**
+ * @brief Check for mail contact
+ *
+ * @param string $uri Profile link
+ * @param integer $uid User ID
+ *
+ * @return array mail data
+ */
+ private function mail($uri, $uid) {
if (!validate_email($uri))
return false;
- $uid = local_user();
- $uid = 1;
-
$x = q("SELECT `prvkey` FROM `user` WHERE `uid` = %d LIMIT 1", intval($uid));
$r = q("SELECT * FROM `mailacct` WHERE `uid` = %d AND `server` != '' LIMIT 1", intval($uid));
- if(count($x) && count($r)) {
+ if (dbm::is_result($x) && dbm::is_result($r)) {
$mailbox = construct_mailbox_name($r[0]);
$password = '';
openssl_private_decrypt(hex2bin($r[0]['pass']), $password,$x[0]['prvkey']);
projects / friendica.git / blobdiff