python 3.15 email.header Header() encoding name change in returned string.

[rouilj]

Bug report

Bug description:

My test suite broke when it executed the equivalent of following:

from email.header import Header
print('%s' % Header(u'café', "iso8859-1").encode())

When this is run under python 3.10->3.14 it returns:

=?iso8859-1?q?caf=E9?=

but when run under python 3.15 it returns:

=?iso-8859-1?q?caf=E9?=

based on the CI output. I don't have a 3.15 local environment. I am interpreting the output as normalizing the encoding.

I have searched the release notes for 3.15 and didn't see anything that indicated Header would not use the supplied name for the charset. I also didn't see anything that indicated the iso8859-1 form was deprecated.

I can change the test case to use iso-8859-1 in place of iso8859-1, which returns the =?iso-8859-1?q?caf=E9?= output in all the python versions I support, but...

Is my search fu just bad or is this a bug/intended change?

Thanks.

-- rouilj

CPython versions tested on:

3.15

Operating systems tested on:

Linux

[StanFromIreland]

Changed by c195a04, CC @serhiy-storchaka.

[serhiy-storchaka]

This is intentional change (see #53144). iso-8859-1 is valid canonical name.

Some charsets were already corrected earlier, for example latin-1 was transformed to iso-8859-1. Now this has been extended to all charsets supported in the stdlib.

[rouilj]

Fair enough. I'll change my test.

Thanks for the quick response.

-- rouilj