diff --git a/searx/utils.py b/searx/utils.py index 59d4b85be..ef221ef8e 100644 --- a/searx/utils.py +++ b/searx/utils.py @@ -119,6 +119,8 @@ class HTMLTextExtractor(HTMLParser): def html_to_text(html): + html = html.replace('\n', ' ') + html = ' '.join(html.split()) s = HTMLTextExtractor() s.feed(html) return s.get_text()