From a1475943a7484cb732e0be0a8dc1391acfe643a3 Mon Sep 17 00:00:00 2001
From: 0x80 <0x80@negativezero.link>
Date: Sun, 24 Apr 2022 00:00:00 +0000
Subject: [PATCH] [generic] Implement basic check that downloaded page is an
 RSS or Atom feed.

---
 src/extractors/generic.py | 11 ++++++-----
 1 file changed, 6 insertions(+), 5 deletions(-)

diff --git a/src/extractors/generic.py b/src/extractors/generic.py
index c4e6a88..03e42d8 100644
--- a/src/extractors/generic.py
+++ b/src/extractors/generic.py
@@ -36,11 +36,12 @@ def try_common_paths (verbosity, url, args):
 	for path in common_paths:
 		page,response_code = download (platform, url + '/' + path, args, verbosity, True)
 		if response_code == 200:
-			# TODO: verify it is a valid RSS feed
-			# Some pages serve response 200 for invalid pages
-
-			# assume we found a feed
-			return url + '/' + path
+			# basic check that it looks like an RSS or Atom file
+			if "<rss" in page or "<feed" in page:
+				# assume we found a feed
+				return url + '/' + path
+			else:
+				debug ("Server returned 200, but it does not appear to be an RSS/Atom feed.", verbosity, platform)
 
 	# failed to find
 	return None