lots of work on federated channel discovery

8 files changed, 303 insertions, 70 deletions

diff --git a/doc/hook/discover_by_webbie.bb b/doc/hook/discover_by_webbie.bb
deleted file mode 100644
index f9228932e..000000000
--- a/doc/hook/discover_by_webbie.bb
+++ /dev/null
@@ -1 +0,0 @@
-[h2]discover_by_webbie[/h2]
diff --git a/doc/hook/discover_channel_webfinger.bb b/doc/hook/discover_channel_webfinger.bb
new file mode 100644
index 000000000..b0eb5f2c4
--- /dev/null
+++ b/doc/hook/discover_channel_webfinger.bb
@@ -0,0 +1,14 @@
+[h2]discover_channel_webfinger[/h2]
+
+Called after performing channel discovery using RFC7033 webfinger and where the channel is not recognised as zot. 
+
+Passed an array:
+
+	address: URL or address that is being discovered
+	success: set to true if the plugin discovers something
+	webfinger: array of webfinger links (output of webfinger_rfc7033())
+
+
+	if your plugin indicates success you are expected to generate and populate an xchan (and hubloc) record prior to returning. 
+
+	
\ No newline at end of file
diff --git a/doc/hooklist.bb b/doc/hooklist.bb
index 9172628a0..bae641585 100644
--- a/doc/hooklist.bb
+++ b/doc/hooklist.bb
@@ -127,7 +127,7 @@ Hooks allow plugins/addons to "hook into" the code at many points and alter the
 [zrl=[baseurl]/help/hook/directory_item]directory_item[/zrl]
 	Called when generating a directory listing for display
 
-[zrl=[baseurl]/help/hook/discover_by_webbie]discover_by_webbie[/zrl]
+[zrl=[baseurl]/help/hook/discover_channel_webfinger]discover_channel_webfinger[/zrl]
 	Called when performing a webfinger lookup
 
 [zrl=[baseurl]/help/hook/display_item]display_item[/zrl]
diff --git a/include/follow.php b/include/follow.php
index 5e1146657..319e9e41b 100644
--- a/include/follow.php
+++ b/include/follow.php
@@ -175,7 +175,9 @@ function new_contact($uid,$url,$channel,$interactive = false, $confirm = false)
 		return $result;
 	}
 
-	$x = array('channel_id' => $uid, 'follow_address' => $url, 'xchan' => $r[0], 'allowed' => 1, 'singleton' => 0);
+	$allowed = ($r[0]['xchan_network'] === 'zot' || $r[0]['xchan_network'] === 'rss') ? 1 : 0);
+
+	$x = array('channel_id' => $uid, 'follow_address' => $url, 'xchan' => $r[0], 'allowed' => $allowed, 'singleton' => 0);
 
 	call_hooks('follow_allow',$x);
 
diff --git a/include/items.php b/include/items.php
index 7f5932cf3..fbcf20636 100755
--- a/include/items.php
+++ b/include/items.php
@@ -4069,29 +4069,65 @@ function process_salmon_feed($xml, $importer) {
 }
 
 /*
- * Given an xml (atom) feed, find any links with rel="hub" and return an array of href links or false
+ * Given an xml (atom) feed, find author and hub links
  */
 
 
-function find_hubs($xml) {
+function feed_meta($xml) {
 	require_once('library/simplepie/simplepie.inc');
 
+	$ret = array();
+
     if(! strlen($xml)) {
         logger('empty input');
-        return false;
+        return $ret;
     }
 
     $feed = new SimplePie();
     $feed->set_raw_data($xml);
     $feed->init();
 
-    if($feed->error())
+    if($feed->error()) {
         logger('Error parsing XML: ' . $feed->error());
+		return $ret;
+	}
+
+    $ret['hubs'] = $feed->get_links('hub');
+
+//     logger('consume_feed: hubs: ' . print_r($hubs,true), LOGGER_DATA);
+	
+	$author = array();
+
+    $found_author = $feed->get_author();
+    if($found_author) {
+        $author['author_name'] = unxmlify($found_author->get_name());
+        $author['author_link'] = unxmlify($found_author->get_link());
+
+    	$rawauthor = $feed->get_feed_tags(SIMPLEPIE_NAMESPACE_ATOM_10,'author');
+		logger('rawauthor: ' . print_r($rawauthor,true));
+
+	    if($rawauthor && $rawauthor[0]['child'][SIMPLEPIE_NAMESPACE_ATOM_10]['link']) {
+    	    $base = $rawauthor[0]['child'][SIMPLEPIE_NAMESPACE_ATOM_10]['link'];
+        	foreach($base as $link) {
+            	if(!x($author, 'author_photo') || ! $author['author_photo']) {
+                	if($link['attribs']['']['rel'] === 'photo' || $link['attribs']['']['rel'] === 'avatar') {
+                    	$author['author_photo'] = unxmlify($link['attribs']['']['href']);
+						break;
+					}
+            	}
+        	}
+			if($rawauthor[0]['child'][NAMESPACE_POCO]['displayName'][0]['data'])
+				$author['full_name'] = unxmlify($rawauthor[0]['child'][NAMESPACE_POCO]['displayName'][0]['data']);
+		}
+    }
 
-    $hubs = $feed->get_links('hub');
-    logger('consume_feed: hubs: ' . print_r($hubs,true), LOGGER_DATA);
 
-	return $hubs;
+    if(substr($author['author_link'],-1,1) == '/')
+        $author['author_link'] = substr($author['author_link'],0,-1);
+
+	$ret['author'] = $author; 
+
+	return $ret;
 }
 
 
diff --git a/include/network.php b/include/network.php
index 7399b4526..7d41a7eb7 100644
--- a/include/network.php
+++ b/include/network.php
@@ -1043,75 +1043,184 @@ function discover_by_url($url,$arr = null) {
 
 }
 
+
+function convert_salmon_key($key) {
+
+	if(strstr($key,','))
+		$rawkey = substr($key,strpos($key,',')+1);
+	else
+		$rawkey = substr($key,5);
+
+	$key_info = explode('.',$rawkey);
+
+	$m = base64url_decode($key_info[1]);
+	$e = base64url_decode($key_info[2]);
+
+	logger('key details: ' . print_r($key_info,true), LOGGER_DEBUG);
+	$salmon_key = metopem($m,$e);
+	return $salmon_key;
+
+}
+
+
 function discover_by_webbie($webbie) {
 	require_once('library/HTML5/Parser.php');
 
+	$result = array();
+	$network = null;
+	$diaspora = false;
+	$gnusoc = false;
+	
+	$has_salmon = false;
+	$salmon_key = false;
+	$atom_feed = false;
+	$diaspora_base = '';
+	$diaspora_guid = '';
+	$diaspora_key = '';
+	$dfrn = false;
+
 	$webbie = strtolower($webbie);
 
 	$x = webfinger_rfc7033($webbie,true);
 	if($x && array_key_exists('links',$x) && $x['links']) {
 		foreach($x['links'] as $link) {
-			if(array_key_exists('rel',$link) && $link['rel'] == 'http://purl.org/zot/protocol') {
-				logger('discover_by_webbie: zot found for ' . $webbie, LOGGER_DEBUG);
-				if(array_key_exists('zot',$x) && $x['zot']['success'])
-					$i = import_xchan($x['zot']);
-				else {
-					$z = z_fetch_url($link['href']);
-					if($z['success']) {
-						$j = json_decode($z['body'],true);
-						$i = import_xchan($j);
-						return true;
+			if(array_key_exists('rel',$link)) {
+				if($link['rel'] == 'http://purl.org/zot/protocol') {
+					logger('discover_by_webbie: zot found for ' . $webbie, LOGGER_DEBUG);
+					if(array_key_exists('zot',$x) && $x['zot']['success'])
+						$i = import_xchan($x['zot']);
+					else {
+						$z = z_fetch_url($link['href']);
+						if($z['success']) {
+							$j = json_decode($z['body'],true);
+							$i = import_xchan($j);
+							return true;
+						}
+					}
+				}
+				if($link['rel'] == 'magic-public-key') {
+        			if(substr($link['href'],0,5) === 'data:') {
+						$salmon_key = convert_salmon_key($link['href']);
 					}
 				}
+				if($link['rel'] == 'salmon') {
+					$has_salmon = true;
+				}
+				if($link['rel'] == 'http://schemas.google.com/g/2010#updates-from') {
+					$atom_feed = $link['href'];
+				}
 			}
 		}
 	}
 
-	$arr = array('address' => $webbie, 'success' => false);
-	call_hooks('discover_by_webbie', $arr);
+
+	logger('webfing: ' . print_r($x,true));
+
+	$arr = array('address' => $webbie, 'success' => false, 'webfinger' => $x);
+	call_hooks('discover_channel_webfinger', $arr);
 	if($arr['success'])
 		return true;
 
-	$result = array();
-	$network = null;
-	$diaspora = false;
+	if($salmon_key && $has_salmon && $atom_feed) {
+		
+		$gnusoc = true;
+		$addr = $x['address'];
 
-	$diaspora_base = '';
-	$diaspora_guid = '';
-	$diaspora_key = '';
-	$dfrn = false;
+		$m = parse_url($x['location']);
+
+		$k = z_fetch_url($atom_feed);
+		if($k['success'])
+			$feed_meta = feed_meta($k['body']);
+		if($feed_meta && $feed_meta['author']) {
+			$r = q("select * from xchan where xchan_hash = '%s' limit 1",
+				dbesc($addr)
+			);
+			if($r) {
+				$r = q("update xchan set xchan_name = '%s', xchan_network = '%s', xchan_name_date = '%s' where xchan_hash = '%s' limit 1",
+					dbesc(($feed_meta['author']['author_name']) ? $feed_meta['author']['author_name'] : $x['nickname']),
+					dbesc('gnusoc'),
+					dbesc(datetime_convert()),
+					dbesc($addr)
+				);
+			}
+			else {
 
-	$x = old_webfinger($webbie);			
-	if($x) {
-		logger('old_webfinger: ' . print_r($x,true));
-		foreach($x as $link) {
-			if($link['@attributes']['rel'] === NAMESPACE_DFRN)
-				$dfrn = unamp($link['@attributes']['href']);				
-			if($link['@attributes']['rel'] === 'salmon')
-				$notify = unamp($link['@attributes']['href']);
- 			if($link['@attributes']['rel'] === NAMESPACE_FEED)
-				$poll = unamp($link['@attributes']['href']);
-			if($link['@attributes']['rel'] === 'http://microformats.org/profile/hcard')
-				$hcard = unamp($link['@attributes']['href']);
-			if($link['@attributes']['rel'] === 'http://webfinger.net/rel/profile-page')
-				$profile = unamp($link['@attributes']['href']);
-			if($link['@attributes']['rel'] === 'http://portablecontacts.net/spec/1.0')
-				$poco = unamp($link['@attributes']['href']);
-			if($link['@attributes']['rel'] === 'http://joindiaspora.com/seed_location') {
-				$diaspora_base = unamp($link['@attributes']['href']);
-				$diaspora = true;
+				$r = q("insert into xchan ( xchan_hash, xchan_guid, xchan_pubkey, xchan_addr, xchan_url, xchan_name, xchan_network, xchan_name_date ) values ('%s', '%s', '%s', '%s', '%s', '%s', '%s', '%s') ",
+					dbesc($addr),
+					dbesc($x['location']),
+					dbesc($salmon_key),
+					dbesc($addr),
+					dbesc($x['location']),
+					dbesc(($feed_meta['author']['author_name']) ? $feed_meta['author']['author_name'] : $x['nickname']),
+					dbesc('gnusoc'),
+					dbescdate(datetime_convert())
+				);
 			}
-			if($link['@attributes']['rel'] === 'http://joindiaspora.com/guid') {
-				$diaspora_guid = unamp($link['@attributes']['href']);
-				$diaspora = true;
+
+			$r = q("select * from hubloc where hubloc_hash = '%s' limit 1",
+				dbesc($addr)
+			);
+
+			if(! $r) {
+
+				$r = q("insert into hubloc ( hubloc_guid, hubloc_hash, hubloc_addr, hubloc_network, hubloc_url, hubloc_host, hubloc_callback, hubloc_updated, hubloc_primary ) values ('%s','%s','%s','%s','%s','%s','%s','%s', 1)",
+					dbesc($x['location']),
+					dbesc($addr),
+					dbesc($addr),
+					dbesc('gnusoc'),
+					dbesc($m['scheme'] . '://' . $m['host']),
+					dbesc($m['host']),
+					dbesc($salmon),
+					dbescdate(datetime_convert())
+				);
 			}
-			if($link['@attributes']['rel'] === 'diaspora-public-key') {
-				$diaspora_key = base64_decode(unamp($link['@attributes']['href']));
-				if(strstr($diaspora_key,'RSA '))
-					$pubkey = rsatopem($diaspora_key);
-				else
-					$pubkey = $diaspora_key;
-				$diaspora = true;
+			$photos = import_xchan_photo($feed_meta['author']['author_photo'],$addr);
+			$r = q("update xchan set xchan_photo_date = '%s', xchan_photo_l = '%s', xchan_photo_m = '%s', xchan_photo_s = '%s', xchan_photo_mimetype = '%s' where xchan_hash = '%s'",
+				dbescdate(datetime_convert()),
+				dbesc($photos[0]),
+				dbesc($photos[1]),
+				dbesc($photos[2]),
+				dbesc($photos[3]),
+				dbesc($addr)
+			);
+			return true;
+
+		}
+	}
+	else {
+
+		$x = old_webfinger($webbie);			
+		if($x) {
+			logger('old_webfinger: ' . print_r($x,true));
+			foreach($x as $link) {
+				if($link['@attributes']['rel'] === NAMESPACE_DFRN)
+					$dfrn = unamp($link['@attributes']['href']);				
+				if($link['@attributes']['rel'] === 'salmon')
+					$notify = unamp($link['@attributes']['href']);
+	 			if($link['@attributes']['rel'] === NAMESPACE_FEED)
+					$poll = unamp($link['@attributes']['href']);
+				if($link['@attributes']['rel'] === 'http://microformats.org/profile/hcard')
+					$hcard = unamp($link['@attributes']['href']);
+				if($link['@attributes']['rel'] === 'http://webfinger.net/rel/profile-page')
+					$profile = unamp($link['@attributes']['href']);
+				if($link['@attributes']['rel'] === 'http://portablecontacts.net/spec/1.0')
+					$poco = unamp($link['@attributes']['href']);
+				if($link['@attributes']['rel'] === 'http://joindiaspora.com/seed_location') {
+					$diaspora_base = unamp($link['@attributes']['href']);
+					$diaspora = true;
+				}
+				if($link['@attributes']['rel'] === 'http://joindiaspora.com/guid') {
+					$diaspora_guid = unamp($link['@attributes']['href']);
+					$diaspora = true;
+				}
+				if($link['@attributes']['rel'] === 'diaspora-public-key') {
+					$diaspora_key = base64_decode(unamp($link['@attributes']['href']));
+					if(strstr($diaspora_key,'RSA '))
+						$pubkey = rsatopem($diaspora_key);
+					else
+						$pubkey = $diaspora_key;
+					$diaspora = true;
+				}
 			}
 		}
 
@@ -1167,7 +1276,7 @@ function discover_by_webbie($webbie) {
 			}
 			else {
 
-				$r = q("insert into xchan ( xchan_hash, xchan_guid, xchan_pubkey, xchan_addr, xchan_url, xchan_name, xchan_network, xchan_instance_url, xchan_name_date ) values ('%s', '%s', '%s', '%s', '%s', '%s', '%s', '%s', '%s') ",
+				$r = q("insert into xchan ( xchan_hash, xchan_guid, xchan_pubkey, xchan_addr, xchan_url, xchan_name, xchan_network, xchan_name_date ) values ('%s', '%s', '%s', '%s', '%s', '%s', '%s', '%s') ",
 					dbesc($addr),
 					dbesc($guid),
 					dbesc($pubkey),
@@ -1175,7 +1284,6 @@ function discover_by_webbie($webbie) {
 					dbesc($profile),
 					dbesc($vcard['fn']),
 					dbesc($network),
-					dbesc(z_root()),
 					dbescdate(datetime_convert())
 				);
 			}
@@ -1199,7 +1307,7 @@ function discover_by_webbie($webbie) {
 			}
 			$photos = import_xchan_photo($vcard['photo'],$addr);
 			$r = q("update xchan set xchan_photo_date = '%s', xchan_photo_l = '%s', xchan_photo_m = '%s', xchan_photo_s = '%s', xchan_photo_mimetype = '%s' where xchan_hash = '%s'",
-				dbescdate(datetime_convert('UTC','UTC',$arr['photo_updated'])),
+				dbescdate(datetime_convert()),
 				dbesc($photos[0]),
 				dbesc($photos[1]),
 				dbesc($photos[2]),
@@ -1209,7 +1317,7 @@ function discover_by_webbie($webbie) {
 			return true;
 
 		}
-
+	}
 	return false;
 
 /*
@@ -1278,10 +1386,6 @@ LSIeXnd14lQYK/uxW/8cTFjcmddsKxeXysoQxbSa9VdDK+KkpZdgYXYrTTofXs6v+
 )
 */
 
-
-
-
-	}
 }
 
 
@@ -1308,13 +1412,89 @@ function webfinger_rfc7033($webbie,$zot = false) {
 
 	$s = z_fetch_url('https://' . $rhs . '/.well-known/webfinger?f=&resource=' . $resource . (($zot) ? '&zot=1' : ''));
 
-	if($s['success'])
+	if($s['success']) {
 		$j = json_decode($s['body'],true);
+
+		// We could have a number of URL aliases and webbies
+		// make an executive decision about the most likely "best" of each
+		// by comparing against some examples from known networks we're likely to encounter.
+		// Otherwise we have to store every alias that we may ever encounter and 
+		// validate every URL we ever find against every possible alias
+
+		// @fixme pump.io is going to be a real bugger since it doesn't return subject or aliases
+		// or provide lookup by url
+
+		$j['address'] = find_webfinger_address($j,$rhs);
+		$j['location'] = find_webfinger_location($j,$rhs);
+		if($j['address'])
+			$j['nickname'] = substr($j['address'],0,strpos($j['address'],'@'));
+	}
 	else
 		return false;
+
 	return($j);
 }
 
+function find_webfinger_address($j,$rhs) {
+	if(is_array($j) && ($j)) {
+		if(strpos($j['subject'],'acct:') !== false && strpos($j['subject'],'@' . $rhs))
+			return str_replace('acct:','',$j['subject']);
+		if($j['aliases']) {
+			foreach($j['aliases'] as $alias) {
+				if(strpos($alias,'acct:') !== false && strpos($alias,'@' . $rhs)) {
+					return str_replace('acct:','',$alias);
+				}
+			}
+		}
+	}
+	return '';
+}
+
+
+function find_webfinger_location($j,$rhs) {
+	if(is_array($j) && ($j)) {
+		if(strpos($j['subject'],'http') === 0) {
+			$x = match_webfinger_location($j['subject'],$rhs);
+			if($x)
+				return $x;
+		}
+		if($j['aliases']) {
+			foreach($j['aliases'] as $alias) {
+				if(strpos($alias,'http') === 0) {
+					$x = match_webfinger_location($alias,$rhs);
+					if($x)
+						return($x);
+				}
+			}
+		}
+	}
+	return '';
+}
+
+function match_webfinger_location($s,$h) {
+
+	// GNU-social and the older StatusNet
+	if(preg_match('|' . $h . '/user/([0-9]*?)$|',$s))
+		return $s;
+	// Redmatrix / hubzilla
+	if(preg_match('|' . $h . '/channel/|',$s))
+		return $s;
+	// Friendica
+	if(preg_match('|' . $h . '/profile/|',$s))
+		return $s;
+
+	$arr = array('test' => $s, 'host' => $h, 'success' => false);
+	call_hooks('match_webfinger_location',$arr);
+	if($arr['success'])
+		return $s;
+	return '';
+}
+	
+
+
+
+
+
 
 function old_webfinger($webbie) {
 
diff --git a/mod/wfinger.php b/mod/wfinger.php
index 3f9826f9b..5c1a74f10 100644
--- a/mod/wfinger.php
+++ b/mod/wfinger.php
@@ -73,13 +73,15 @@ function wfinger_init(&$a) {
 
 		$result['aliases'] = array();
 
-		$result['properties'] = array('http://webfinger.net/ns/name' => $r[0]['channel_name']);
+		$result['properties'] = array(
+				'http://webfinger.net/ns/name' => $r[0]['channel_name'],
+				'http://xmlns.com/foaf/0.1/name' => $r[0]['channel_name']
+		);
 
 		foreach($aliases as $alias) 
 			if($alias != $resource)
 				$result['aliases'][] = $alias;
 
-
 		$result['links'] = array(
 
 			array(
diff --git a/version.inc b/version.inc
index 5ade25348..7825710e4 100644
--- a/version.inc
+++ b/version.inc
@@ -1 +1 @@
-2016-03-15.1336H
+2016-03-16.1337H