email.policy.SMTP TypeError: 'Header' object is not subscriptable

[barskykd]

Bug report

Code snippet is taken directly from documentation https://docs.python.org/3/library/email.header.html The only change - policy changed to email.policy.SMTP. Documentation (https://docs.python.org/3/library/email.policy.html#email.policy.SMTP) states that the only difference between email.policy.default and email.policy.SMTP - linesep used.

from email.message import Message
from email.header import Header
from email.policy import SMTP

msg = Message()
h = Header('p\xf6stal', 'iso-8859-1')
msg['Subject'] = h
print(msg.as_string())

msg = Message(policy=SMTP)
h = Header('p\xf6stal', 'iso-8859-1')
msg['Subject'] = h
print(msg.as_string())

Output is:

Subject: =?iso-8859-1?q?p=F6stal?=

Traceback (most recent call last):
  File "C:\Users\barsk\AppData\Roaming\JetBrains\PyCharm2022.3\scratches\scratch_3.py", line 12, in <module>
    msg['Subject'] = h
  File "C:\Program Files (x86)\Python37-32\lib\email\message.py", line 409, in __setitem__
    self._headers.append(self.policy.header_store_parse(name, val))
  File "C:\Program Files (x86)\Python37-32\lib\email\policy.py", line 148, in header_store_parse
    return (name, self.header_factory(name, value))
  File "C:\Program Files (x86)\Python37-32\lib\email\headerregistry.py", line 589, in __call__
    return self[name](name, value)
  File "C:\Program Files (x86)\Python37-32\lib\email\headerregistry.py", line 197, in __new__
    cls.parse(value, kwds)
  File "C:\Program Files (x86)\Python37-32\lib\email\headerregistry.py", line 272, in parse
    kwds['parse_tree'] = cls.value_parser(value)
  File "C:\Program Files (x86)\Python37-32\lib\email\_header_value_parser.py", line 1105, in get_unstructured
    if value[0] in WSP:
TypeError: 'Header' object is not subscriptable

Your environment

Reproduced on windows on

• Python 3.11 (64-bit)
• Python 3.10 (64-bit)
• Python 3.7 (32-bit)

[codzy-ds]

I have the same bug but for the policy.default.

Is there a workaround? It is quite a blocker.

[barskykd]

For this particular bug there is workaround. Convert Header to string before assignment

msg['Subject'] = str(h)

[codzy-ds]

Oh, simple enough, thank you!

[alex-pobeditel-2004]

I've found this bug while fighting with problem described here: https://github.com/python/cpython/issues/44510

When I create an email with Subject as str, spaces that are located near email body line breaks are lost during decode/encode. For example:

msg['Subject'] = 'Приглашение в закрытую программу Перенос удалится'

gets encoded as:

Subject: =?utf-8?b?0J/RgNC40LPQu9Cw0YjQtdC90LjQtSDQsiDQt9Cw0LrRgNGL0YLRg9GO?=
  =?utf-8?b?0L/RgNC+0LPRgNCw0LzQvNGDINCf0LXRgNC10L3QvtGBINGD0LTQsNC70LjRgtGB?=
 =?utf-8?q?=D1=8F?=

and later is decoded by email client in this way:

Subject: Приглашение в закрытуюпрограмму Перенос удалится

There are 2 spaces in the beginning of line 2 of MIME representation, but on decode the "significant" space is lost.

If it's impossible now to declare header in this way (what was recommended in the bug mentioned above):

msg['Subject'] = Header('Приглашение в закрытую программу Перенос удалится', 'utf-8')

Is there a workaround for such cases?

I figured out that it's possible to use longer "raw" lines via email policy (max_line_length=998, for example), but I'm not sure this is the most correct way to overcome this problem.

UPD: No, setting high max_line_length does not affect the Subject header.

UPD2: Looks like my problem can't be solved with Header either, I found the exact same bug here: https://github.com/python/cpython/issues/92081