When using email.charset.Charset to encode MIME bodie as quoted-printable, some characters that are encodable with the quopri_codec cause a KeyError in email.quoprimime:

Python 3.3.0 (default, Oct  7 2012, 14:43:21) 
[GCC 4.6.3] on linux
Type "help", "copyright", "credits" or "license" for more information.
>>> text_encoded = b'mo \xe2\x82\xac'
>>> text = text_encoded.decode('utf-8')
>>> print(text)
mo €
>>> import codecs
>>> codecs.getencoder('quopri_codec')(text_encoded)[0]
b'mo=20=E2=82=AC'
>>> from email import charset
>>> c = charset.Charset('utf-8')
>>> c.header_encoding = charset.QP
>>> c.body_encoding = charset.QP
>>> c.header_encode(text)
'=?utf-8?q?mo_=E2=82=AC?='
>>> c.body_encode(text)
Traceback (most recent call last):
  File "<stdin>", line 1, in <module>
  File "/usr/lib/python3.3/email/charset.py", line 395, in body_encode
    return email.quoprimime.body_encode(string)
  File "/usr/lib/python3.3/email/quoprimime.py", line 240, in body_encode
    if body_check(ord(c)):
  File "/usr/lib/python3.3/email/quoprimime.py", line 81, in body_check
    return chr(octet) != _QUOPRI_BODY_MAP[octet]
KeyError: 8364

What is the reason that email.quoprimime doesn't use codecs.getencoder('quopri_codec') to do the actual character encoding?

There is a long history behind quoprime in the email package, and I don't know all of it.  It is on my list of things to sort out (there is also potential duplication with the binascii module), but I haven't gotten to it yet.

The issue you report, though is a duplicate of issue 16948 and is already fixed in the repo.

You checked off 2.7.  Did you actually see a failure in 2.7?  It shouldn't be broken there.

I thought I had tested it under 2.7, but I'm not entirely sure.

OK, if you find it is a bug on 2.7 you can reopen.