suppress some scraping errors when confronted with hybrid/strange

feeds that provide insufficient content-type and choke the html parser.

1 files changed, 3 insertions, 0 deletions

diff --git a/boot.php b/boot.php
index e47f1834d..0dd507168 100644
--- a/boot.php
+++ b/boot.php
@@ -1366,6 +1366,7 @@ function lrdd($uri) {
 	else {
 		$html = fetch_url($uri);
 		$headers = $a->get_curl_headers();
+		logger('lrdd: headers=' . $headers, LOGGER_DEBUG);
 		$lines = explode("\n",$headers);
 		if(count($lines)) {
 			foreach($lines as $line) {				
@@ -1377,6 +1378,8 @@ function lrdd($uri) {
 				// don't try and run feeds through the html5 parser
 				if(stristr($line,'content-type:') && ((stristr($line,'application/atom+xml')) || (stristr($line,'application/rss+xml'))))
 					return array();
+				if(stristr($html,'<rss') || stristr($html,'<feed'))
+					return array();
 			}
 		}
 		if(! isset($link)) {