@@ -119,6 +119,8 @@ class HTMLTextExtractor(HTMLParser):
121def html_to_text(html):
121def html_to_text(html):
122+ html = html.replace('\n', ' ')
123+ html = ' '.join(html.split())
122 s = HTMLTextExtractor()
124 s = HTMLTextExtractor()
123 s.feed(html)
125 s.feed(html)
124 return s.get_text()
126 return s.get_text()