>>> re.match(r'\s', '\n')
<_sre.SRE_Match object; span=(0, 1), match='\n'>
>>> re.match(r'\s', r'\n')
>>>

This result came as a surprise to me, but explains why widont has this behavior with newlines. But is this the desired behavior? That is, is the text passed in supposed to be escaped already?

I believe this would be the easiest way to get the correct behavior in the above example:

text = 'blah<br>\nblah'
text = text.encode('unicode-escape')  # b'blah<br>\\nblah'
text = text.decode('utf-8')  # 'blah<br>\\nblah'
text = widont(text)  # 'blah<br>\\nblah'
text = text.encode('utf-8')  # b'blah<br>\\nblah'
text = text.decode('unicode-escape')  # 'blah<br>\nblah'

It seems like it would be preferable for typogrify to deal with this, and I think it can be done without any encoding/decoding.