Can someone look at it @mvz pls? I've been using you package since version 0.903, but when it was changed from utf8 to raw binary in version 0.919 (by Andreas Pflug) it is useless for our latin characters. It is the same for multiple code pages.

I still have to use PERL version v5.26.1 from Ubuntu 18 and the old version 0.903 with perl msgconvert.pl <msgfile>

I'm guessing this means the codepage is set in a property that I'm currently skipping.

@czende if you have a message that you can share with me that has this problem, it would really help in debugging if you can send it to me.

Hi, I'm a co-worker of @czende , unfortunately we can't share the problematic email because it contains sensitive data and we couldn't reproduce it any other way.

I tested your PR locally on our problematic email but the result was the same, the output .eml had a broken message body, so your modification did not help in our case.

But I found that in our case, "latin extended" characters like "ěščřž...." in the headers are probably responsible for the problem.
If the Subject or other headers like From, To, .. contain these characters, it causes "Value for 'Subject' header with wide characters at .../perl5/lib/perl5/Email/Outlook/Message.pm line 331."

Based on this, I changed the header_set method to header_str_set on the mentioned line and the wide characters problem disappeared and the message body in output .eml file is also fixed.

I'm not a Perl developer, so I'm not sure about this modification, but even according to the documentation, the header_str_set method seems to be preferable and also seems to solve the broken message body problem.