[fix] whitespace duplication fix in nonhtml outputs

This commit is contained in:
asciimoo 2014-02-04 19:42:32 +01:00
parent 474f6a861d
commit 5695ec58df

View file

@ -190,7 +190,9 @@ def index():
else: else:
if 'content' in result: if 'content' in result:
result['content'] = html_to_text(result['content']).strip() result['content'] = html_to_text(result['content']).strip()
result['title'] = html_to_text(result['title']).strip() # removing html content and whitespace duplications
result['title'] = ' '.join(html_to_text(result['title'])\
.strip().split())
if len(result['url']) > 74: if len(result['url']) > 74:
url_parts = result['url'][:35], result['url'][-35:] url_parts = result['url'][:35], result['url'][-35:]
result['pretty_url'] = '{0}[...]{1}'.format(*url_parts) result['pretty_url'] = '{0}[...]{1}'.format(*url_parts)