From 793967a1d3c23fcf1f3b00a2832f51e6f473f4bd Mon Sep 17 00:00:00 2001
From: Friendika <info@friendika.com>
Date: Mon, 4 Apr 2011 19:36:18 -0700
Subject: [PATCH] better handling of troublesome feeds.

---
 boot.php             | 42 +++++++++++++++++++++++++++++++-----------
 include/Scrape.php   |  2 +-
 include/items.php    | 23 +++++++++++++++--------
 include/poller.php   |  2 +-
 mod/dfrn_confirm.php |  2 +-
 mod/dfrn_poll.php    |  4 ++--
 mod/follow.php       | 14 +++++++++++---
 7 files changed, 62 insertions(+), 27 deletions(-)

diff --git a/boot.php b/boot.php
index 3b86d0dbe..f5c0e6f92 100644
--- a/boot.php
+++ b/boot.php
@@ -1478,7 +1478,9 @@ function lrdd($uri) {
 		return array();
 
 	logger('lrdd: host_meta: ' . $xml, LOGGER_DATA);
-	$h = simplexml_load_string($xml);
+
+	$h = parse_xml_string($xml);
+
 	$arr = convert_xml_element_to_array($h);
 
 	if(isset($arr['xrd']['property'])) {
@@ -1550,16 +1552,19 @@ function lrdd($uri) {
 	$headers = $a->get_curl_headers();
 	logger('lrdd: headers=' . $headers, LOGGER_DEBUG);
 
-	require_once('library/HTML5/Parser.php');
-	$dom = @HTML5_Parser::parse($html);
+	// don't try and parse raw xml as html
+	if(! strstr($html,'<?xml')) {
+		require_once('library/HTML5/Parser.php');
+		$dom = @HTML5_Parser::parse($html);
 
-	if($dom) {
-		$items = $dom->getElementsByTagName('link');
-		foreach($items as $item) {
-			$x = $item->getAttribute('rel');
-			if($x == "lrdd") {
-				$pagelink = $item->getAttribute('href');
-				break;
+		if($dom) {
+			$items = $dom->getElementsByTagName('link');
+			foreach($items as $item) {
+				$x = $item->getAttribute('rel');
+				if($x == "lrdd") {
+					$pagelink = $item->getAttribute('href');
+					break;
+				}
 			}
 		}
 	}
@@ -1638,7 +1643,7 @@ function fetch_xrd_links($url) {
 		return array();
 
 	logger('fetch_xrd_links: ' . $xml, LOGGER_DATA);
-	$h = simplexml_load_string($xml);
+	$h = parse_xml_string($xml);
 	$arr = convert_xml_element_to_array($h);
 
 	$links = array();
@@ -2759,3 +2764,18 @@ function lang_selector() {
 	$o .= '</select></form></div>';
 	return $o;
 }}
+
+
+if(! function_exists('parse_xml_string')) {
+function parse_xml_string($s) {
+	if(! strstr($s,'<?xml'))
+		return false;
+	$s2 = substr($s,strpos($s,'<?xml'));
+	libxml_use_internal_errors(true);
+	$x = @simplexml_load_string($s2);
+	if(count(libxml_get_errors()))
+		foreach(libxml_get_errors() as $err)
+			logger('libxml: parse: ' . $err, LOGGER_DATA);
+	libxml_clear_errors();
+	return $x;
+}}
diff --git a/include/Scrape.php b/include/Scrape.php
index ff9899252..21820ddaf 100644
--- a/include/Scrape.php
+++ b/include/Scrape.php
@@ -216,7 +216,7 @@ function scrape_feed($url) {
 				}
  				if(stristr($line,'application/rss+xml') || stristr($s,'<rss')) {
 					$ret['feed_rss'] = $url;
-					return ret;
+					return $ret;
 				}
 			}
 		}
diff --git a/include/items.php b/include/items.php
index 1dd39d2ba..a9ac85969 100644
--- a/include/items.php
+++ b/include/items.php
@@ -180,7 +180,7 @@ function construct_activity_object($item) {
 
 	if($item['object']) {
 		$o = '<as:object>' . "\r\n";
-		$r = @simplexml_load_string($item['object']);
+		$r = parse_xml_string($item['object']);
 		if($r->type)
 			$o .= '<as:object-type>' . xmlify($r->type) . '</as:object-type>' . "\r\n";
 		if($r->id)
@@ -206,7 +206,7 @@ function construct_activity_target($item) {
 
 	if($item['target']) {
 		$o = '<as:target>' . "\r\n";
-		$r = @simplexml_load_string($item['target']);
+		$r = parse_xml_string($item['target']);
 		if($r->type)
 			$o .= '<as:object-type>' . xmlify($r->type) . '</as:object-type>' . "\r\n";
 		if($r->id)
@@ -241,8 +241,14 @@ function get_atom_elements($feed,$item) {
 	$res = array();
 
 	$author = $item->get_author();
-	$res['author-name'] = unxmlify($author->get_name());
-	$res['author-link'] = unxmlify($author->get_link());
+	if($author) { 
+		$res['author-name'] = unxmlify($author->get_name());
+		$res['author-link'] = unxmlify($author->get_link());
+	}
+	else {
+		$res['author-name'] = unxmlify($feed->get_title());
+		$res['author-link'] = unxmlify($feed->get_permalink());
+	}
 	$res['uri'] = unxmlify($item->get_id());
 	$res['title'] = unxmlify($item->get_title());
 	$res['body'] = unxmlify($item->get_content());
@@ -343,7 +349,6 @@ function get_atom_elements($feed,$item) {
 	// the wild, by sanitising it and converting supported tags to bbcode before we rip out any remaining 
 	// html.
 
-
 	if((strpos($res['body'],'<') !== false) || (strpos($res['body'],'>') !== false)) {
 
 		$res['body'] = preg_replace('#<object[^>]+>.+?' . 'http://www.youtube.com/((?:v|cp)/[A-Za-z0-9\-_=]+).+?</object>#s',
@@ -783,7 +788,7 @@ function dfrn_deliver($owner,$contact,$atom, $dissolve = false) {
 		return 3;
 	}
 
-	$res = simplexml_load_string($xml);
+	$res = parse_xml_string($xml);
 
 	if((intval($res->status) != 0) || (! strlen($res->challenge)) || (! strlen($res->dfrn_id)))
 		return (($res->status) ? $res->status : 3);
@@ -878,7 +883,7 @@ function dfrn_deliver($owner,$contact,$atom, $dissolve = false) {
 		return 3;
 	}
 
-	$res = simplexml_load_string($xml);
+	$res = parse_xml_string($xml);
 
 	return $res->status;
  
@@ -916,6 +921,7 @@ function consume_feed($xml,$importer,&$contact, &$hub, $datedir = 0, $secure_fee
 	if($feed->error())
 		logger('consume_feed: Error parsing XML: ' . $feed->error());
 
+	$permalink = $feed->get_permalink();
 
 	// Check at the feed level for updated contact name and/or photo
 
@@ -1230,6 +1236,7 @@ function consume_feed($xml,$importer,&$contact, &$hub, $datedir = 0, $secure_fee
 				// Head post of a conversation. Have we seen it? If not, import it.
 
 				$item_id  = $item->get_id();
+
 				$datarray = get_atom_elements($feed,$item);
 
 				$r = q("SELECT `uid`, `last-child`, `edited`, `body` FROM `item` WHERE `uri` = '%s' AND `uid` = %d LIMIT 1",
@@ -1275,7 +1282,7 @@ function consume_feed($xml,$importer,&$contact, &$hub, $datedir = 0, $secure_fee
 				if(! is_array($contact))
 					return;
 
-				if($contact['network'] === 'stat') {
+				if($contact['network'] === 'stat' || stristr($permalink,'twitter.com')) {
 					if(strlen($datarray['title']))
 						unset($datarray['title']);
 					$datarray['last-child'] = 1;
diff --git a/include/poller.php b/include/poller.php
index 3b80c1c04..9362c28b3 100644
--- a/include/poller.php
+++ b/include/poller.php
@@ -203,7 +203,7 @@ function poller_run($argv, $argc){
 				}
 
 
-				$res = simplexml_load_string($xml);
+				$res = parse_xml_string($xml);
 	
 				if(intval($res->status) == 1) {
 					logger("poller: $url replied status 1 - marking for death ");
diff --git a/mod/dfrn_confirm.php b/mod/dfrn_confirm.php
index 1bf1ba954..2db745d25 100644
--- a/mod/dfrn_confirm.php
+++ b/mod/dfrn_confirm.php
@@ -240,7 +240,7 @@ function dfrn_confirm_post(&$a,$handsfree = null) {
 				notice( t('Unexpected response from remote site: ') . EOL . $leading_junk . EOL );
 			}
 
-			$xml = simplexml_load_string($res);
+			$xml = parse_xml_string($res);
 			$status = (int) $xml->status;
 			$message = unxmlify($xml->message);   // human readable text of what may have gone wrong.
 			switch($status) {
diff --git a/mod/dfrn_poll.php b/mod/dfrn_poll.php
index 5149dc3b2..2ccfadd03 100644
--- a/mod/dfrn_poll.php
+++ b/mod/dfrn_poll.php
@@ -69,7 +69,7 @@ function dfrn_poll_init(&$a) {
 
 			if(strlen($s)) {
 
-				$xml = simplexml_load_string($s);
+				$xml = parse_xml_string($s);
 
 				if((int) $xml->status == 1) {
 					$_SESSION['authenticated'] = 1;
@@ -468,7 +468,7 @@ function dfrn_poll_content(&$a) {
 
 			if(strlen($s) && strstr($s,'<?xml')) {
 
-				$xml = simplexml_load_string($s);
+				$xml = parse_xml_string($s);
 
 				logger('dfrn_poll: profile: parsed xml: ' . print_r($xml,true), LOGGER_DATA);
 
diff --git a/mod/follow.php b/mod/follow.php
index 4ce3ccb82..06e81ceed 100644
--- a/mod/follow.php
+++ b/mod/follow.php
@@ -15,7 +15,8 @@ function follow_post(&$a) {
 	$email_conversant = false;
 
 	if($url) {
-		$links = @lrdd($url);
+		$links = lrdd($url);
+
 		if(count($links)) {
 			foreach($links as $link) {
 				if($link['@attributes']['rel'] === NAMESPACE_DFRN)
@@ -107,7 +108,7 @@ function follow_post(&$a) {
 	if((! isset($vcard)) && (! $poll)) {
 
 		$ret = scrape_feed($url);
-
+		logger('mod_follow: 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']));
 			$vcard = array();
@@ -156,7 +157,14 @@ function follow_post(&$a) {
 			}
 			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;
 		}