X-Git-Url: https://git.mxchange.org/?a=blobdiff_plain;f=include%2FScrape.php;h=a6bb5f728c6d659be732ca65b7ab66439b552671;hb=b43ed54967e133ca250ee7f42a90d699781d904f;hp=bb42c3bdd3474946db9325ad4fb6846141c5571e;hpb=8c5c07b2654f5ff28cea1a7012198f74c5b1af03;p=friendica.git

diff --git a/include/Scrape.php b/include/Scrape.php
index bb42c3bdd3..a6bb5f728c 100644
--- a/include/Scrape.php
+++ b/include/Scrape.php
@@ -8,12 +8,18 @@ function scrape_dfrn($url) {
 	$a = get_app();
 
 	$ret = array();
+
+	logger('scrape_dfrn: url=' . $url);
+
 	$s = fetch_url($url);
 
 	if(! $s) 
 		return $ret;
 
 	$headers = $a->get_curl_headers();
+	logger('scrape_dfrn: headers=' . $headers, LOGGER_DEBUG);
+
+
 	$lines = explode("\n",$headers);
 	if(count($lines)) {
 		foreach($lines as $line) {				
@@ -93,12 +99,17 @@ function scrape_meta($url) {
 	$a = get_app();
 
 	$ret = array();
+
+	logger('scrape_meta: url=' . $url);
+
 	$s = fetch_url($url);
 
 	if(! $s) 
 		return $ret;
 
 	$headers = $a->get_curl_headers();
+	logger('scrape_meta: headers=' . $headers, LOGGER_DEBUG);
+
 	$lines = explode("\n",$headers);
 	if(count($lines)) {
 		foreach($lines as $line) {				
@@ -135,6 +146,9 @@ function scrape_vcard($url) {
 	$a = get_app();
 
 	$ret = array();
+
+	logger('scrape_vcard: url=' . $url);
+
 	$s = fetch_url($url);
 
 	if(! $s) 
@@ -190,17 +204,19 @@ function scrape_feed($url) {
 		return $ret;
 
 	$headers = $a->get_curl_headers();
+	logger('scrape_feed: headers=' . $headers, LOGGER_DEBUG);
+
 	$lines = explode("\n",$headers);
 	if(count($lines)) {
 		foreach($lines as $line) {				
 			if(stristr($line,'content-type:')) {
-				if(stristr($line,'application/atom+xml')) {
+				if(stristr($line,'application/atom+xml') || stristr($s,'<feed')) {
 					$ret['feed_atom'] = $url;
 					return $ret;
 				}
- 				if(stristr($line,'application/rss+xml')) {
+ 				if(stristr($line,'application/rss+xml') || stristr($s,'<rss')) {
 					$ret['feed_rss'] = $url;
-					return ret;
+					return $ret;
 				}
 			}
 		}
@@ -211,17 +227,265 @@ function scrape_feed($url) {
 	if(! $dom)
 		return $ret;
 
+
+	$items = $dom->getElementsByTagName('img');
+
+	// get img elements (twitter)
+
+	if($items) {
+		foreach($items as $item) {
+			$x = $item->getAttribute('id');
+			if($x === 'profile-image') {
+				$ret['photo'] = $item->getAttribute('src');
+			}
+		}
+	}
+
 	$items = $dom->getElementsByTagName('link');
 
-	// get Atom link elements
+	// get Atom/RSS link elements, take the first one of either.
 
-	foreach($items as $item) {
-		$x = $item->getAttribute('rel');
-		if(($x === 'alternate') && ($item->getAttribute('type') === 'application/atom+xml'))
-			$ret['feed_atom'] = $item->getAttribute('href');
-		if(($x === 'alternate') && ($item->getAttribute('type') === 'application/rss+xml'))
-			$ret['feed_rss'] = $item->getAttribute('href');
+	if($items) {
+		foreach($items as $item) {
+			$x = $item->getAttribute('rel');
+			if(($x === 'alternate') && ($item->getAttribute('type') === 'application/atom+xml')) {
+				if(! x($ret,'feed_atom'))
+					$ret['feed_atom'] = $item->getAttribute('href');
+			}
+			if(($x === 'alternate') && ($item->getAttribute('type') === 'application/rss+xml')) {
+				if(! x($ret,'feed_rss'))
+					$ret['feed_rss'] = $item->getAttribute('href');
+			}
+		}	
 	}
 
 	return $ret;
-}}
\ No newline at end of file
+}}
+
+
+function probe_url($url) {
+	require_once('include/email.php');
+
+	$result = array();
+
+	if(! $url)
+		return $result;
+
+	$diaspora = false;	
+	$email_conversant = false;
+
+	if($url) {
+		$links = lrdd($url);
+
+		if(count($links)) {
+			foreach($links 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://joindiaspora.com/seed_location')
+					$diaspora = true;
+			}
+
+			// Status.Net can have more than one profile URL. We need to match the profile URL
+			// to a contact on incoming messages to prevent spam, and we won't know which one
+			// to match. So in case of two, one of them is stored as an alias. Only store URL's
+			// and not webfinger user@host aliases. If they've got more than two non-email style
+			// aliases, let's hope we're lucky and get one that matches the feed author-uri because 
+			// otherwise we're screwed.
+
+			foreach($links as $link) {
+				if($link['@attributes']['rel'] === 'alias') {
+					if(strpos($link['@attributes']['href'],'@') === false) {
+						if(isset($profile)) {
+							if($link['@attributes']['href'] !== $profile)
+								$alias = unamp($link['@attributes']['href']);
+						}
+						else
+							$profile = unamp($link['@attributes']['href']);
+					}
+				}
+			}
+		}
+		else {
+
+			// Check email
+
+			$orig_url = $url;
+			if((strpos($orig_url,'@')) && validate_email($orig_url)) {
+				$x = q("SELECT `prvkey` FROM `user` WHERE `uid` = %d LIMIT 1",
+					intval(local_user())
+				);
+				$r = q("SELECT * FROM `mailacct` WHERE `uid` = %d AND `server` != '' LIMIT 1",
+					intval(local_user())
+				);
+				if(count($x) && count($r)) {
+				    $mailbox = construct_mailbox_name($r[0]);
+					$password = '';
+					openssl_private_decrypt(hex2bin($r[0]['pass']),$password,$x[0]['prvkey']);
+					$mbox = email_connect($mailbox,$r[0]['user'],$password);
+					unset($password);
+				}
+				if($mbox) {
+					$msgs = email_poll($mbox,$orig_url);
+					if(count($msgs)) {
+						$addr = $orig_url;
+						$network = NETWORK_MAIL;
+						$name = substr($url,0,strpos($url,'@'));
+						$profile = 'http://' . substr($url,strpos($url,'@')+1);
+						// fix nick character range
+						$vcard = array('fn' => $name, 'nick' => $name, 'photo' => gravatar_img($url));
+						$notify = 'smtp ' . random_string();
+						$poll = 'email ' . random_string();
+						$priority = 0;
+						$x = email_msg_meta($mbox,$msgs[0]);
+						$adr = imap_rfc822_parse_adrlist($x->from,'');
+						if(strlen($adr[0]->personal))
+							$vcard['fn'] = notags($adr[0]->personal);
+					}
+					imap_close($mbox);
+				}
+			}
+		}
+	}	
+
+	if(strlen($dfrn)) {
+		$ret = scrape_dfrn($dfrn);
+		if(is_array($ret) && x($ret,'dfrn-request')) {
+			$network = NETWORK_DFRN;
+			$request = $ret['dfrn-request'];
+			$confirm = $ret['dfrn-confirm'];
+			$notify  = $ret['dfrn-notify'];
+			$poll    = $ret['dfrn-poll'];
+		}
+	}
+
+	if($network !== NETWORK_DFRN && $network !== NETWORK_MAIL) {
+		$network  = NETWORK_OSTATUS;
+		$priority = 0;
+
+		if($hcard) {
+			$vcard = scrape_vcard($hcard);
+
+			// Google doesn't use absolute url in profile photos
+	
+			if((x($vcard,'photo')) && substr($vcard['photo'],0,1) == '/') {
+				$h = @parse_url($hcard);
+				if($h)
+					$vcard['photo'] = $h['scheme'] . '://' . $h['host'] . $vcard['photo'];
+			}
+		
+			logger('probe_url: scrape_vcard: ' . print_r($vcard,true), LOGGER_DATA);
+		}
+
+		if(! $profile) {
+			if($diaspora)
+				$profile = $hcard;
+			else
+				$profile = $url;
+		}
+
+		if(! x($vcard,'fn'))
+			if(x($vcard,'nick'))
+				$vcard['fn'] = $vcard['nick'];
+
+		if((! isset($vcard)) && (! $poll)) {
+
+			$ret = scrape_feed($url);
+			logger('probe_url: scrape_feed returns: ' . print_r($ret,true), LOGGER_DATA);
+			if(count($ret) && ($ret['feed_atom'] || $ret['feed_rss'])) {
+				$poll = ((x($ret,'feed_atom')) ? unamp($ret['feed_atom']) : unamp($ret['feed_rss']));
+				$vcard = array();
+					if(x($ret,'photo'))
+						$vcard['photo'] = $ret['photo'];
+				require_once('simplepie/simplepie.inc');
+			    $feed = new SimplePie();
+				$xml = fetch_url($poll);
+
+    			$feed->set_raw_data($xml);
+
+			    $feed->init();
+
+				if(! x($vcard,'photo'))
+					$vcard['photo'] = $feed->get_image_url();
+				$author = $feed->get_author();
+				if($author) {			
+					$vcard['fn'] = unxmlify(trim($author->get_name()));
+					if(! $vcard['fn'])
+						$vcard['fn'] = trim(unxmlify($author->get_email()));
+					if(strpos($vcard['fn'],'@') !== false)
+						$vcard['fn'] = substr($vcard['fn'],0,strpos($vcard['fn'],'@'));
+					$vcard['nick'] = strtolower(notags(unxmlify($vcard['fn'])));
+					if(strpos($vcard['nick'],' '))
+						$vcard['nick'] = trim(substr($vcard['nick'],0,strpos($vcard['nick'],' ')));
+					$email = unxmlify($author->get_email());
+				}
+				else {
+					$item = $feed->get_item(0);
+					if($item) {
+						$author = $item->get_author();
+						if($author) {			
+							$vcard['fn'] = trim(unxmlify($author->get_name()));
+							if(! $vcard['fn'])
+								$vcard['fn'] = trim(unxmlify($author->get_email()));
+							if(strpos($vcard['fn'],'@') !== false)
+								$vcard['fn'] = substr($vcard['fn'],0,strpos($vcard['fn'],'@'));
+							$vcard['nick'] = strtolower(unxmlify($vcard['fn']));
+							if(strpos($vcard['nick'],' '))
+								$vcard['nick'] = trim(substr($vcard['nick'],0,strpos($vcard['nick'],' ')));
+							$email = unxmlify($author->get_email());
+						}
+						if(! $vcard['photo']) {
+							$rawmedia = $item->get_item_tags('http://search.yahoo.com/mrss/','thumbnail');
+							if($rawmedia && $rawmedia[0]['attribs']['']['url'])
+								$vcard['photo'] = unxmlify($rawmedia[0]['attribs']['']['url']);
+						}
+					}
+				}
+				if((! $vcard['photo']) && strlen($email))
+					$vcard['photo'] = gravatar_img($email);
+				if($poll === $profile)
+					$lnk = $feed->get_permalink();
+				if(isset($lnk) && strlen($lnk))
+					$profile = $lnk;	
+				if(! (x($vcard,'fn')))
+					$vcard['fn'] = notags($feed->get_title());
+				if(! (x($vcard,'fn')))
+					$vcard['fn'] = notags($feed->get_description());
+				$network = 'feed';
+				$priority = 2;
+			}
+		}
+	}
+
+	if(! x($vcard,'photo')) {
+		$a = get_app();
+		$vcard['photo'] = $a->get_baseurl() . '/images/default-profile.jpg' ; 
+	}
+	$vcard['fn'] = notags($vcard['fn']);
+	$vcard['nick'] = notags($vcard['nick']);
+
+
+	$result['name'] = $vcard['fn'];
+	$result['nick'] = $vcard['nick'];
+	$result['url'] = $profile;
+	$result['addr'] = $addr;
+	$result['notify'] = $notify;
+	$result['poll'] = $poll;
+	$result['request'] = $request;
+	$result['confirm'] = $confirm;
+	$result['photo'] = $vcard['photo'];
+	$result['priority'] = $priority;
+	$result['network'] = $network;
+	$result['alias'] = $alias;
+
+	logger('probe_url: ' . print_r($result,true), LOGGER_DEBUG);
+
+	return $result;
+}