Ensure that there isn't any html in feed text

2020-12-12 15:44:43 +00:00 · 2020-12-12 15:44:43 +00:00 · fd07e67f3b
parent 8c3ecf2e94
commit fd07e67f3b
1 changed files with 5 additions and 1 deletions
--- 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):