I am writing on a script that extracts and saves JPEG-attachements from emails and passes them to imagemagick. However, I am living in Germany and special characters in email text/subject as "ö", "ä", "ü" and "ß" are pretty common.

I am extracting the subject with formail:

    SUBJECT=$(formail -zxSubject: <"$file")

and that results in:

  • =?UTF-8?Q?Meine_G=c3=bcte?=

("Meine Güte") or even worse

  • =?UTF-8?B?U2Now7ZuZSBHcsO8w59lIQ==?=

("Schöne Grüße!").

I try to use part of the subject as a filename and as imagemagick text annotation, which obviously doesn't work.

How do I convert this UTF-8 text to text with special characters in bash?

Thanks in advance! Markus

  • Spaces between words and punctuation are missing as well :( – Markus Sep 12 '15 at 23:44

How do I convert this UTF-8 text to text with special characters in bash?

What you have isn't quite "UTF-8 text". You actually want plain UTF-8 text as output, as it's what Linux uses for "special characters" everywhere.

Your input, instead, is MIME (RFC 2047) encoded UTF-8. The "Q" marks Quoted-Printable mode, and "B" marks Base64 mode. Among others, Perl's Encode::MIME::Header can be used to decode both:

#!/usr/bin/env perl
use open qw(:std :utf8);
use Encode qw(decode);

while (my $line = <STDIN>) {
        print decode("MIME-Header", $line);

Oneliner (see perldoc perlrun for explanation):

perl -CS -MEncode -ne 'print decode("MIME-Header", $_)'

This can take any format as input:

$ echo "Subject: =?UTF-8?Q?Meine_G=c3=bcte?=, \
                 =?UTF-8?B?U2Now7ZuZSBHcsO8w59lIQ==?=" | perl ./decode.pl
Subject: Meine Güte, Schöne Grüße!

A version in Python 3:

#!/usr/bin/env python3
import email.header, sys

words = email.header.decode_header(sys.stdin.read())
words = [s.decode(c or "utf-8") for (s, c) in words]
| improve this answer | |
  • The one-liner didn't work for me. (see @blami ). But the pl code worked fine. When I used the last script, the Base64 part wasn't encoded at all. Not even when I tried only that part. Your explanation helped me as well. – Markus Sep 13 '15 at 1:16

E-mail subject itself is header and headers must contain only ASCII characters. This is why UTF-8 (or any other non-ASCII charset) subject must be encoded.

This way of encoding non-ASCII characters in to ASCII is described in RFC 1342.

Basically, encoded subject has (as you've already listed in your examples) following format:


Based on encoding value is encoded-text decoded either as quoted-printable (Q) or as base64 (B).

To get human readable form you need to pass encoded-text portion of subject header value to program that decode it. I believe there are some standalone commands to do that (uudecode), but I prefer to use Perl one-liners:

For quoted-printable:

perl -pe 'use MIME::QuotedPrint; $_=MIME::QuotedPrint::decode($_);'

and for base64:

perl -pe 'use MIME::Base64; $_=MIME::Base64::decode($_);'

Be sure you pass only encoded-text portion and not whole subject header value.

| improve this answer | |
  • Thank you for your comment! I learned a lot from your detailed explanation. However, i couldn't get the one-liners to run properly (in putty on raspberry pi). But I'm sure this is only because of my poor linux skills. I used echo "=?UT...:" | perl ... and the result was "=?UTF-8?Q?Meine_Güte?". Only the "ü" was translated. @grawity 's provided .pl script has produced good results. – Markus Sep 13 '15 at 1:10

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.