From fd07e67f3bab65b3d2b4d4accd0ca612811090e5 Mon Sep 17 00:00:00 2001
From: Bob Mottram <bob@freedombone.net>
Date: Sat, 12 Dec 2020 15:44:43 +0000
Subject: [PATCH] Ensure that there isn't any html in feed text

---
 newswire.py | 6 +++++-
 1 file changed, 5 insertions(+), 1 deletion(-)

diff --git a/newswire.py b/newswire.py
index 1734a3a05..144e7854a 100644
--- a/newswire.py
+++ b/newswire.py
@@ -103,7 +103,11 @@ def addNewswireDictEntry(baseDir: str, domain: str,
                          tags=[], maxTags=32) -> None:
     """Update the newswire dictionary
     """
-    allText = removeHtml(title + ' ' + description)
+    # remove any markup
+    title = removeHtml(title)
+    description = removeHtml(description)
+
+    allText = title + ' ' + description
 
     # check that none of the text is filtered against
     if isFiltered(baseDir, 'news', domain, allText):