From ee45dee9324be48cd87a883405bdd9e11e3f39f5 Mon Sep 17 00:00:00 2001
From: Friendika <info@friendika.com>
Date: Wed, 2 Feb 2011 14:48:27 -0800
Subject: suppress some scraping errors when confronted with hybrid/strange
 feeds that provide insufficient content-type and choke the html parser.

---
 boot.php | 3 +++
 1 file changed, 3 insertions(+)

(limited to 'boot.php')

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)) {
-- 
cgit v1.2.3