This snippet (for #11684, but it's simply BytesParser with
headersonly=True in the end)


        with openfile('msg_46.txt', 'rb') as fp:
            msgdata = fp.read()
        parser = email.parser.BytesHeaderParser()
        msg = parser.parsebytes(msgdata)
        out = BytesIO()
        gen = email.generator.BytesGenerator(out)
        gen.flatten(msg)
        self.assertEqual(out.getvalue(), msgdata)


causes this error:


ERROR: test_byte_message_rfc822_only (test_email.TestMessageAPI)
----------------------------------------------------------------------
Traceback (most recent call last):
  File "/Users/steffen/usr/opt/py3k/lib/python3.3/test/test_email/test_email.py", line 200, in test_byte_message_rfc822_only
    gen.flatten(msg)
  File "/Users/steffen/usr/opt/py3k/lib/python3.3/email/generator.py", line 91, in flatten
    self._write(msg)
  File "/Users/steffen/usr/opt/py3k/lib/python3.3/email/generator.py", line 137, in _write
    self._dispatch(msg)
  File "/Users/steffen/usr/opt/py3k/lib/python3.3/email/generator.py", line 163, in _dispatch
    meth(msg)
  File "/Users/steffen/usr/opt/py3k/lib/python3.3/email/generator.py", line 304, in _handle_message
    self._fp.write(payload)
TypeError: 'str' does not support the buffer interface

The error message suggests that somehow a str string has gotten mixed in with the bytes. Something like '\n' or ' '?

Steffen, if you want to help track this down:
1. What is a minimal msgdata that gives the same error; post it.
2. Add 'print(payload)' before line 304: self._fp.write(payload)
and perhaps further back to see what the offending string is and where it comes from.

Terry, the test is in the other issue, so this time Steffen has provided the test :).

I'll take a look at both issues, probably next week.

On Fri, Apr 08, 2011 at 11:39:36PM +0000, Terry J. Reedy wrote:
> 1. What is a minimal msgdata that gives the same error; post it.

Stepping a bit..  Remove 'Content-Type' header field and this does 
not crash.  Thus the real problem may again lie in get_payload() 
trying to convert data instead of leaving it alone. 
I'll try to instrument the path a bit ..

import email, email.parser, email.generator, io, textwrap
msgdata = textwrap.dedent("""\
            Date: Mon, 01 Feb 2010 12:21:16 +0100
            From: "three" <four@example.net>
            To: <one@two.com>
            Subject: Content-Type crash
            Content-Type: message/rfc822

            Remove Content-Type header field and no crash happens!
        """).encode('ascii')
msg = email.parser.BytesParser().parsebytes(msgdata, headersonly=True)
email.generator.BytesGenerator(io.BytesIO()).flatten(msg)

On Sat, Apr 09, 2011 at 11:56:16AM +0000, Steffen Daode Nurpmeso wrote:
> I'll try to instrument the path a bit ..

Sorry, no time today.  All the stuff next week.
Nice weekend.

So here is a patch which steps forward the not-yet-fully-completed 
transition to the mixed bytes/str EMail stuff. 
Test coverage is available if you patch in 
http://bugs.python.org/file21549/11684.1.diff from #11684.
(Because i leak the great picture of this module ... say.)

I applied this as part of #11684.  Thanks.