Error parsing email headers: AttributeError: 'ValueTerminal' object has no attribute 'fold'

[mgmacias95]

Bug report

Bug description:

The following code breaks with an attribute error:

import email.parser
import email.policy

a = 't'*46

h = f'''\
To: =?utf-8?B?dGVzdC50ZXN0LnRlc3QudGVzdEB0ZXN0LmNvbeKAiw=?= <test@test.com>,\r\n\t"tttest&{a}.t.t.t.t.t.t.t@yahoo.ES" <test@test.tj>,\r\n\t"tttest&{a}.t.t.t.t.t.t@yahoo.ES" <info@test.tj>'''

m = email.parser.HeaderParser(policy=email.policy.default).parsestr(h)
m.as_string()

The problem was introduced on #100885, setting ListSeparator.as_ew_allowed = False to True fixes the problem. Changing any character in the header in the example above also fixes the problem (which makes it harder to understand exactly why it's broken).

CPython versions tested on:

3.12

Operating systems tested on:

macOS

Linked PRs

• gh-118643: Fix AttributeError in the email module #119099
• [3.13] gh-118643: Fix AttributeError in the email module (GH-119099) #119389
• [3.12] gh-118643: Fix AttributeError in the email module (GH-119099) #119390
• [3.11] gh-118643: Fix AttributeError in the email module (GH-119099) #119393

[serhiy-storchaka]

From RFC 822, section 3.11:

        Note:  While the standard  permits  folding  wherever  linear-
               white-space is permitted, it is recommended that struc-
               tured fields, such as those containing addresses, limit
               folding  to higher-level syntactic breaks.  For address
               fields, it  is  recommended  that  such  folding  occur
               between addresses, after the separating comma.

I think it means that we should also set ListSeparator.syntactic_break to False.

[serhiy-storchaka]

But this does not help. It just returns the bug reported in #100884.

[mgmacias95]

But this does not help. It just returns the bug reported in #100884.

I don't understand this comment, I just tested the code mentioned in that issue setting ListSeparator.syntactic_break to
False and it works. What bug is returned then?

[serhiy-storchaka]

It no longer raises an exception, but it encodes the comma as =?utf-8?q?=2C?=. #100885 was not only incorrect, even after fixing the error it is not enough to fix the original issue #100884. I am trying to find a solution which fixes both the original issue and the new error.