Email headers do not properly decode to unicode.

[TimRawlinson]

BPO	18044
Nosy	@warsaw, @ezio-melotti, @bitdancer

^{Note: these values reflect the state of the issue at the time it was migrated and might not reflect the current state.}

Show more details

GitHub fields:

assignee = None
closed_at = <Date 2013-07-11.20:06:03.514>
created_at = <Date 2013-05-23.11:57:36.183>
labels = ['type-bug', 'expert-email', 'docs']
title = 'Email headers do not properly decode to unicode.'
updated_at = <Date 2013-07-11.20:11:20.868>
user = 'https://bugs.python.org/TimRawlinson'

bugs.python.org fields:

activity = <Date 2013-07-11.20:11:20.868>
actor = 'r.david.murray'
assignee = 'docs@python'
closed = True
closed_date = <Date 2013-07-11.20:06:03.514>
closer = 'r.david.murray'
components = ['Documentation', 'email']
creation = <Date 2013-05-23.11:57:36.183>
creator = 'Tim.Rawlinson'
dependencies = []
files = []
hgrepos = []
issue_num = 18044
keywords = []
message_count = 4.0
messages = ['189862', '192894', '192895', '192899']
nosy_count = 6.0
nosy_names = ['barry', 'ezio.melotti', 'r.david.murray', 'docs@python', 'python-dev', 'Tim.Rawlinson']
pr_nums = []
priority = 'normal'
resolution = 'fixed'
stage = 'resolved'
status = 'closed'
superseder = None
type = 'behavior'
url = 'https://bugs.python.org/issue18044'
versions = ['Python 3.3', 'Python 3.4']

[TimRawlinson]

In Python 3.3 decoding of headers to unicode is supposed to be automatic but fails in several cases, including one shown as successful in the documentation:

   >>> msg = message_from_string('Subject: =?utf-8?q?=C3=89ric?=\n\n', policy=default)
   >>> msg['Subject']
   '=?utf-8?q?=C3=89ric?='

   >>> msg = message_from_string('To: =?utf-8?q?=C3=89ric <foo@example.com>\n\n', policy=default)
   >>> msg['To']
   '=?utf-8?q?=C3=89ric?= <foo@example.com>'

Although the following works:

   >>> msg = message_from_string('Subject: =?utf-8?q?Eric?=\n\n', policy=default)
   >>> msg['Subject']
   'Eric'

Though this does not:

   >>> msg = message_from_string('To: =?utf-8?q?Eric?= <foo@example.com>\n\n', policy=default)
   >>> msg['To']
   '=?utf-8?q?Eric?= <foo@example.com>'

And just to prove some things are working as they should:

    >>> msg = message_from_string("Subject: =?gb2312?b?1eLKx9bQzsSy4srUo6E=?=\n\n", policy=default)
    >>> msg['Subject']
    '这是中文测试！'

[python-dev]

New changeset 4acb822f4c43 by R David Murray in branch '3.3':
bpo-18044: Fix parsing of encoded words of the form =?utf8?q?=XX...?=
http://hg.python.org/cpython/rev/4acb822f4c43

New changeset 32c6cfffbddd by R David Murray in branch 'default':
Merge: bpo-18044: Fix parsing of encoded words of the form =?utf8?q?=XX...?=
http://hg.python.org/cpython/rev/32c6cfffbddd

[bitdancer]

This is actually two separate bugs, both a bit embarrassing. The first one (that I just fixed) is that when parsing an encoded word I was only checking for decimal digits after an '=' (instead of the correct hex digits) when trying to do robust detection of encoded word limits.

The second (the address one) is due to the fact that somewhere between stopping my full time work on the email project and actually committing the code, I lost track of the fact that I never implemented encoded word parsing for anything other than unstructured headers. The infrastructure is there, I just need to write tests and hook it up.

I'm going to open a separate issue for that.

[bitdancer]

The issue for the second bug is bpo-18431.