Home / server / Questions / 264472
In Process
Dan
Asked: 2011-04-29 12:45:54 +0800 CST

Postfix doesn't seem to treat non-English characters properly (probably character set issue)

  • 772

I have set up Postfix on my CentOS server.

If the subject of the email contains some non-English characters (i.e.: accented letters), it gets ruined.

Here is an example.

Original subject:

this is a testà è ciao

Subject, once delivered to the server:
=?ISO-8859-1?Q?this_is_a_test=E0_=E8_ciao?=

I know it is now an issue with MUAs because: _ A lot of users have reported the issue - thus the MUA of the sender is out of the equation
_ I check the raw email on the server (I access the actual file with VI)

I guess it is a problem with charset, but I can't find anywhere in the Postfix config files how to handle the charset.

Thanks,
Dan

postfix

1 Answers

  1. Actually, Postfix seems to do what it should do, according to RFC2047: Message Header Extensions for Non-ASCII Text. In the body you can have 8-bit data, but I believe that headers are universally agreed to use only 7-bit data.

    Testing the conversion with some sample python suggests that the header is transformed correctly:

    >>> import email.header as eh
>>> r= eh.decode_header('=?ISO-8859-1?Q?this_is_a_test=E0_=E8_ciao?=')
>>> print r
[('this is a test\xe0 \xe8 ciao', 'iso-8859-1')]
>>> print r[0][0].decode(r[0][1])
this is a testà è ciao

    As Mark Henderson gently hinted in a comment, I didn't try to be thorough about the reasons for your problem. So: it is possible that the end user mail client does not transform back the subject line. Can you provide us with the raw header from the message? You can see that by viewing the “message source” in Thunderbird (typically Ctrl+U), as an example.

    • 2