Hi,
I sent this mail to mutt-users some days ago. In the past couple of days
I've been hit by this behaviour (using mutt 1.5.21 compiled from
scratch). Suppose that I receive a mail with several attachments:
I 1 <no description> [multipa/alternativ, 7bit, 7.2K]
I 2 ├─><no description> [text/plain, 7bit, iso-8859-1, 0.7K]
I 3 └─><no description> [text/html, 7bit, iso-8859-1, 6.3K]
A 4 file.txt [text/plain, 8bit, us-ascii, 2.3K]
A 5 file.xml [text/xml, 8bit, us-ascii, 0.1K]
A 6 file.pdf [text/x-unknown-, base64, us-ascii, 66K]
The message has 3 attachments, one text file, one xml document and one
pdf document (this is a real example,the sender MUA being Thuderbird
3.1.18). It turns out that the sender MUA got it wrong, and the
character encoding is wrongly set (us-ascii), but:
- file.txt is utf-8 encoded
- file.xml is also utf-8 encoded.
- file.pdf is a pdf document.
The problem is that there is no way to save these attachment
properly. For instance, when saving file.xml all the non ascii
characters are replaced with "?". So if file.xml originally was:
<?xml version="1.0" encoding="utf-8"?>
<a>©Áéñio</a>
After saving to disk it becomes:
<?xml version="1.0" encoding="utf-8"?>
<a>????????io</a>
The same happens with the text file. Even the pdf document is wrongly
saved. The only way to actually get the original content of the
attchments is to manually set the mime type (using ctrl-e) to
"application/octet-stream", and then saving them to disk.
I locally modified mutt the following below, because I don't want to
perform any character conversions when saving attachments:
diff -r 25a7f8f7d50d attach.c
--- a/attach.c Wed Sep 15 10:10:39 2010 -0700
+++ b/attach.c Thu Mar 01 07:12:54 2012 +0100
@@ -802,7 +802,7 @@
STATE s;
memset (&s, 0, sizeof (s));
- s.flags |= M_CHARCONV;
+ s.flags &= ~M_CHARCONV;
if ((s.fpout = mutt_save_attachment_open (path, flags)) == NULL)
{
It would be better to parametrize this behaviour with a global
option. Even better would be to first try to translate the attachment,
and failback to a "raw" save if a conversion error occured (a character
could not be translated). But I'm afraid I don't have the capabilities
to code this :-(
thanks,
aitor
