<!DOCTYPE HTML PUBLIC "-//W3C//DTD HTML 3.2 Final//EN">
<HTML>
<HEAD>
<META NAME="GENERATOR" CONTENT="LinuxDoc-Tools 0.9.21">
<TITLE>The Mutt E-Mail Client: Mutt's MIME Support</TITLE>
<LINK HREF="manual-6.html" REL=next>
<LINK HREF="manual-4.html" REL=previous>
<LINK HREF="manual.html#toc5" REL=contents>
</HEAD>
<BODY>
<A HREF="manual-6.html">Next</A>
<A HREF="manual-4.html">Previous</A>
<A HREF="manual.html#toc5">Contents</A>
<HR>
<H2><A NAME="s5">5.</A> <A HREF="manual.html#toc5">Mutt's MIME Support</A></H2>
<P>Quite a bit of effort has been made to make Mutt the premier text-mode
MIME MUA. Every effort has been made to provide the functionality that
the discerning MIME user requires, and the conformance to the standards
wherever possible. When configuring Mutt for MIME, there are two extra
types of configuration files which Mutt uses. One is the
<CODE>mime.types</CODE> file, which contains the mapping of file extensions to
IANA MIME types. The other is the <CODE>mailcap</CODE> file, which specifies
the external commands to use for handling specific MIME types.</P>
<H2><A NAME="ss5.1">5.1</A> <A HREF="manual.html#toc5.1">Using MIME in Mutt</A>
</H2>
<P>There are three areas/menus in Mutt which deal with MIME, they are the
pager (while viewing a message), the attachment menu and the compose
menu.</P>
<H3>Viewing MIME messages in the pager</H3>
<P>When you select a message from the index and view it in the pager, Mutt
decodes the message to a text representation. Mutt internally supports
a number of MIME types, including <CODE>text/plain, text/enriched,
message/rfc822, and message/news</CODE>. In addition, the export
controlled version of Mutt recognizes a variety of PGP MIME types,
including PGP/MIME and application/pgp.</P>
<P>Mutt will denote attachments with a couple lines describing them.
These lines are of the form:
<BLOCKQUOTE><CODE>
<PRE>
[-- Attachment #1: Description --]
[-- Type: text/plain, Encoding: 7bit, Size: 10000 --]
</PRE>
</CODE></BLOCKQUOTE>
Where the <CODE>Description</CODE> is the description or filename given for the
attachment, and the <CODE>Encoding</CODE> is one of
<CODE>7bit/8bit/quoted-printable/base64/binary</CODE>.</P>
<P>If Mutt cannot deal with a MIME type, it will display a message like:
<BLOCKQUOTE><CODE>
<PRE>
[-- image/gif is unsupported (use 'v' to view this part) --]
</PRE>
</CODE></BLOCKQUOTE>
</P>
<H3><A NAME="attach_menu"></A> The Attachment Menu</H3>
<P>The default binding for <CODE>view-attachments</CODE> is `v', which displays the
attachment menu for a message. The attachment menu displays a list of
the attachments in a message. From the attachment menu, you can save,
print, pipe, delete, and view attachments. You can apply these
operations to a group of attachments at once, by tagging the attachments
and by using the ``tag-prefix'' operator. You can also reply to the
current message from this menu, and only the current attachment (or the
attachments tagged) will be quoted in your reply. You can view
attachments as text, or view them using the mailcap viewer definition.</P>
<P>Finally, you can apply the usual message-related functions (like
<A HREF="manual-2.html#resend-message">resend-message</A>, and the reply
and forward functions) to attachments of type <CODE>message/rfc822</CODE>.</P>
<P>See the help on the attachment menu for more information.</P>
<H3><A NAME="compose_menu"></A> The Compose Menu</H3>
<P>The compose menu is the menu you see before you send a message. It
allows you to edit the recipient list, the subject, and other aspects
of your message. It also contains a list of the attachments of your
message, including the main body. From this menu, you can print, copy,
filter, pipe, edit, compose, review, and rename an attachment or a
list of tagged attachments. You can also modifying the attachment
information, notably the type, encoding and description.</P>
<P>Attachments appear as follows:
<PRE>
- 1 [text/plain, 7bit, 1K] /tmp/mutt-euler-8082-0 <no description>
2 [applica/x-gunzip, base64, 422K] ~/src/mutt-0.85.tar.gz <no description>
</PRE>
</P>
<P>The '-' denotes that Mutt will delete the file after sending (or
postponing, or cancelling) the message. It can be toggled with the
<CODE>toggle-unlink</CODE> command (default: u). The next field is the MIME
content-type, and can be changed with the <CODE>edit-type</CODE> command
(default: ^T). The next field is the encoding for the attachment,
which allows a binary message to be encoded for transmission on 7bit
links. It can be changed with the <CODE>edit-encoding</CODE> command
(default: ^E). The next field is the size of the attachment,
rounded to kilobytes or megabytes. The next field is the filename,
which can be changed with the <CODE>rename-file</CODE> command (default: R).
The final field is the description of the attachment, and can be
changed with the <CODE>edit-description</CODE> command (default: d).</P>
<H2><A NAME="ss5.2">5.2</A> <A HREF="manual.html#toc5.2">MIME Type configuration with <CODE>mime.types</CODE></A>
</H2>
<P>When you add an attachment to your mail message, Mutt searches your
personal mime.types file at <CODE>${HOME}/.mime.types</CODE>, and then
the system mime.types file at <CODE>/usr/local/share/mutt/mime.types</CODE> or
<CODE>/etc/mime.types</CODE></P>
<P>The mime.types file consist of lines containing a MIME type and a space
separated list of extensions. For example:
<BLOCKQUOTE><CODE>
<PRE>
application/postscript ps eps
application/pgp pgp
audio/x-aiff aif aifc aiff
</PRE>
</CODE></BLOCKQUOTE>
A sample <CODE>mime.types</CODE> file comes with the Mutt distribution, and
should contain most of the MIME types you are likely to use. </P>
<P>If Mutt can not determine the mime type by the extension of the file you
attach, it will look at the file. If the file is free of binary
information, Mutt will assume that the file is plain text, and mark it
as <CODE>text/plain</CODE>. If the file contains binary information, then Mutt will
mark it as <CODE>application/octet-stream</CODE>. You can change the MIME
type that Mutt assigns to an attachment by using the <CODE>edit-type</CODE>
command from the compose menu (default: ^T). The MIME type is actually a
major mime type followed by the sub-type, separated by a '/'. 6 major
types: application, text, image, video, audio, and model have been approved
after various internet discussions. Mutt recognises all of these if the
appropriate entry is found in the mime.types file. It also recognises other
major mime types, such as the chemical type that is widely used in the
molecular modelling community to pass molecular data in various forms to
various molecular viewers. Non-recognised mime types should only be used
if the recipient of the message is likely to be expecting such attachments.</P>
<H2><A NAME="ss5.3">5.3</A> <A HREF="manual.html#toc5.3">MIME Viewer configuration with <CODE>mailcap</CODE></A>
</H2>
<P>Mutt supports RFC 1524 MIME Configuration, in particular the Unix
specific format specified in Appendix A of RFC 1524. This file format
is commonly referred to as the mailcap format. Many MIME compliant
programs utilize the mailcap format, allowing you to specify handling
for all MIME types in one place for all programs. Programs known to
use this format include Netscape, XMosaic, lynx and metamail.</P>
<P>In order to handle various MIME types that Mutt can not handle
internally, Mutt parses a series of external configuration files to
find an external handler. The default search string for these files
is a colon delimited list set to
<BLOCKQUOTE><CODE>
<PRE>
${HOME}/.mailcap:/usr/local/share/mutt/mailcap:/etc/mailcap:/etc/mailcap:/usr/etc/mailcap:/usr/local/etc/mailcap
</PRE>
</CODE></BLOCKQUOTE>
where <CODE>$HOME</CODE> is your home directory.</P>
<P>In particular, the metamail distribution will install a mailcap file,
usually as <CODE>/usr/local/etc/mailcap</CODE>, which contains some baseline
entries.</P>
<H3>The Basics of the mailcap file</H3>
<P>A mailcap file consists of a series of lines which are comments, blank,
or definitions.</P>
<P>A comment line consists of a # character followed by anything you want.</P>
<P>A blank line is blank.</P>
<P>A definition line consists of a content type, a view command, and any
number of optional fields. Each field of a definition line is divided
by a semicolon ';' character.</P>
<P>The content type is specified in the MIME standard type/subtype method.
For example,
<CODE>text/plain, text/html, image/gif, </CODE>
etc. In addition, the mailcap format includes two formats for
wildcards, one using the special '*' subtype, the other is the implicit
wild, where you only include the major type. For example, <CODE>image/*</CODE>, or
<CODE>video,</CODE> will match all image types and video types,
respectively.</P>
<P>The view command is a Unix command for viewing the type specified. There
are two different types of commands supported. The default is to send
the body of the MIME message to the command on stdin. You can change
this behaviour by using %s as a parameter to your view command.
This will cause Mutt to save the body of the MIME message to a temporary
file, and then call the view command with the %s replaced by
the name of the temporary file. In both cases, Mutt will turn over the
terminal to the view program until the program quits, at which time Mutt
will remove the temporary file if it exists.</P>
<P>So, in the simplest form, you can send a text/plain message to the
external pager more on stdin:
<BLOCKQUOTE><CODE>
<PRE>
text/plain; more
</PRE>
</CODE></BLOCKQUOTE>
Or, you could send the message as a file:
<BLOCKQUOTE><CODE>
<PRE>
text/plain; more %s
</PRE>
</CODE></BLOCKQUOTE>
Perhaps you would like to use lynx to interactively view a text/html
message:
<BLOCKQUOTE><CODE>
<PRE>
text/html; lynx %s
</PRE>
</CODE></BLOCKQUOTE>
In this case, lynx does not support viewing a file from stdin, so you
must use the %s syntax.
<B>Note:</B> <EM>Some older versions of lynx contain a bug where they
will check the mailcap file for a viewer for text/html. They will find
the line which calls lynx, and run it. This causes lynx to continuously
spawn itself to view the object.</EM></P>
<P>On the other hand, maybe you don't want to use lynx interactively, you
just want to have it convert the text/html to text/plain, then you can
use:
<BLOCKQUOTE><CODE>
<PRE>
text/html; lynx -dump %s | more
</PRE>
</CODE></BLOCKQUOTE>
</P>
<P>Perhaps you wish to use lynx to view text/html files, and a pager on
all other text formats, then you would use the following:
<BLOCKQUOTE><CODE>
<PRE>
text/html; lynx %s
text/*; more
</PRE>
</CODE></BLOCKQUOTE>
This is the simplest form of a mailcap file.</P>
<H3>Secure use of mailcap</H3>
<P>The interpretion of shell meta-characters embedded in MIME parameters
can lead to security problems in general. Mutt tries to quote parameters
in expansion of %s syntaxes properly, and avoids risky characters by
substituting them, see the
<A HREF="manual-6.html#mailcap_sanitize">mailcap_sanitize</A> variable.</P>
<P>Although mutt's procedures to invoke programs with mailcap seem to be
safe, there are other applications parsing mailcap, maybe taking less care
of it. Therefore you should pay attention to the following rules:</P>
<P><EM>Keep the %-expandos away from shell quoting.</EM>
Don't quote them with single or double quotes. Mutt does this for
you, the right way, as should any other program which interprets
mailcap. Don't put them into backtick expansions. Be highly careful
with eval statements, and avoid them if possible at all. Trying to fix
broken behaviour with quotes introduces new leaks - there is no
alternative to correct quoting in the first place.</P>
<P>If you have to use the %-expandos' values in context where you need
quoting or backtick expansions, put that value into a shell variable
and reference the shell variable where necessary, as in the following
example (using <CODE>$charset</CODE> inside the backtick expansion is safe,
since it is not itself subject to any further expansion):</P>
<P>
<BLOCKQUOTE><CODE>
<PRE>
text/test-mailcap-bug; cat %s; copiousoutput; test=charset=%{charset} \
&& test "`echo $charset | tr '[A-Z]' '[a-z]'`" != iso-8859-1
</PRE>
</CODE></BLOCKQUOTE>
</P>
<H3>Advanced mailcap Usage</H3>
<H3>Optional Fields</H3>
<P>In addition to the required content-type and view command fields, you
can add semi-colon ';' separated fields to set flags and other options.
Mutt recognizes the following optional fields:
<DL>
<DT><B>copiousoutput</B><DD><P>This flag tells Mutt that the command passes possibly large amounts of
text on stdout. This causes Mutt to invoke a pager (either the internal
pager or the external pager defined by the pager variable) on the output
of the view command. Without this flag, Mutt assumes that the command
is interactive. One could use this to replace the pipe to <CODE>more</CODE>
in the <CODE>lynx -dump</CODE> example in the Basic section:
<BLOCKQUOTE><CODE>
<PRE>
text/html; lynx -dump %s ; copiousoutput
</PRE>
</CODE></BLOCKQUOTE>
This will cause lynx to format the text/html output as text/plain
and Mutt will use your standard pager to display the results.</P>
<DT><B>needsterminal</B><DD><P>Mutt uses this flag when viewing attachments with
<A HREF="#auto_view">autoview</A>, in order to decide whether it should honor the setting
of the
<A HREF="manual-6.html#wait_key">$wait_key</A> variable or
not. When an attachment is viewed using an interactive program, and the
corresponding mailcap entry has a <EM>needsterminal</EM> flag, Mutt will use
<A HREF="manual-6.html#wait_key">$wait_key</A> and the exit status
of the program to decide if it will ask you to press a key after the
external program has exited. In all other situations it will not prompt
you for a key.</P>
<DT><B>compose=<command></B><DD><P>This flag specifies the command to use to create a new attachment of a
specific MIME type. Mutt supports this from the compose menu.</P>
<DT><B>composetyped=<command></B><DD><P>This flag specifies the command to use to create a new attachment of a
specific MIME type. This command differs from the compose command in
that mutt will expect standard MIME headers on the data. This can be
used to specify parameters, filename, description, etc. for a new
attachment. Mutt supports this from the compose menu.</P>
<DT><B>print=<command></B><DD><P>This flag specifies the command to use to print a specific MIME type.
Mutt supports this from the attachment and compose menus.</P>
<DT><B>edit=<command></B><DD><P>This flag specifies the command to use to edit a specific MIME type.
Mutt supports this from the compose menu, and also uses it to compose
new attachments. Mutt will default to the defined editor for text
attachments.</P>
<DT><B>nametemplate=<template></B><DD><P>This field specifies the format for the file denoted by %s in the
command fields. Certain programs will require a certain file extension,
for instance, to correctly view a file. For instance, lynx will only
interpret a file as <CODE>text/html</CODE> if the file ends in <CODE>.html</CODE>.
So, you would specify lynx as a <CODE>text/html</CODE> viewer with a line in
the mailcap file like:
<BLOCKQUOTE><CODE>
<PRE>
text/html; lynx %s; nametemplate=%s.html
</PRE>
</CODE></BLOCKQUOTE>
</P>
<DT><B>test=<command></B><DD><P>This field specifies a command to run to test whether this mailcap
entry should be used. The command is defined with the command expansion
rules defined in the next section. If the command returns 0, then the
test passed, and Mutt uses this entry. If the command returns non-zero,
then the test failed, and Mutt continues searching for the right entry.
<B>Note:</B> <EM>the content-type must match before Mutt performs the test.</EM>
For example:
<BLOCKQUOTE><CODE>
<PRE>
text/html; netscape -remote 'openURL(%s)' ; test=RunningX
text/html; lynx %s
</PRE>
</CODE></BLOCKQUOTE>
In this example, Mutt will run the program RunningX which will return 0
if the X Window manager is running, and non-zero if it isn't. If
RunningX returns 0, then Mutt will call netscape to display the
text/html object. If RunningX doesn't return 0, then Mutt will go on
to the next entry and use lynx to display the text/html object.</P>
</DL>
</P>
<H3>Search Order</H3>
<P>When searching for an entry in the mailcap file, Mutt will search for
the most useful entry for its purpose. For instance, if you are
attempting to print an <CODE>image/gif</CODE>, and you have the following
entries in your mailcap file, Mutt will search for an entry with the
print command:
<BLOCKQUOTE><CODE>
<PRE>
image/*; xv %s
image/gif; ; print= anytopnm %s | pnmtops | lpr; \
nametemplate=%s.gif
</PRE>
</CODE></BLOCKQUOTE>
Mutt will skip the <CODE>image/*</CODE> entry and use the <CODE>image/gif</CODE>
entry with the print command.</P>
<P>In addition, you can use this with
<A HREF="#auto_view">Autoview</A>
to denote two commands for viewing an attachment, one to be viewed
automatically, the other to be viewed interactively from the attachment
menu. In addition, you can then use the test feature to determine which
viewer to use interactively depending on your environment.
<BLOCKQUOTE><CODE>
<PRE>
text/html; netscape -remote 'openURL(%s)' ; test=RunningX
text/html; lynx %s; nametemplate=%s.html
text/html; lynx -dump %s; nametemplate=%s.html; copiousoutput
</PRE>
</CODE></BLOCKQUOTE>
For
<A HREF="#auto_view">Autoview</A>, Mutt will choose the third
entry because of the copiousoutput tag. For interactive viewing, Mutt
will run the program RunningX to determine if it should use the first
entry. If the program returns non-zero, Mutt will use the second entry
for interactive viewing.</P>
<H3>Command Expansion</H3>
<P>The various commands defined in the mailcap files are passed to the
<CODE>/bin/sh</CODE> shell using the system() function. Before the
command is passed to <CODE>/bin/sh -c</CODE>, it is parsed to expand
various special parameters with information from Mutt. The keywords
Mutt expands are:
<DL>
<DT><B>%s</B><DD><P>As seen in the basic mailcap section, this variable is expanded
to a filename specified by the calling program. This file contains
the body of the message to view/print/edit or where the composing
program should place the results of composition. In addition, the
use of this keyword causes Mutt to not pass the body of the message
to the view/print/edit program on stdin.</P>
<DT><B>%t</B><DD><P>Mutt will expand %t to the text representation of the content
type of the message in the same form as the first parameter of the
mailcap definition line, ie <CODE>text/html</CODE> or
<CODE>image/gif</CODE>.</P>
<DT><B>%{<parameter>}</B><DD><P>Mutt will expand this to the value of the specified parameter
from the Content-Type: line of the mail message. For instance, if
Your mail message contains:
<BLOCKQUOTE><CODE>
<PRE>
Content-Type: text/plain; charset=iso-8859-1
</PRE>
</CODE></BLOCKQUOTE>
then Mutt will expand %{charset} to iso-8859-1. The default metamail
mailcap file uses this feature to test the charset to spawn an xterm
using the right charset to view the message.</P>
<DT><B>\%</B><DD><P>This will be replaced by a %</P>
</DL>
Mutt does not currently support the %F and %n keywords
specified in RFC 1524. The main purpose of these parameters is for
multipart messages, which is handled internally by Mutt.</P>
<H3>Example mailcap files</H3>
<P>This mailcap file is fairly simple and standard:
<HR>
<PRE>
# I'm always running X :)
video/*; xanim %s > /dev/null
image/*; xv %s > /dev/null
# I'm always running netscape (if my computer had more memory, maybe)
text/html; netscape -remote 'openURL(%s)'
</PRE>
<HR>
</P>
<P>This mailcap file shows quite a number of examples:</P>
<P>
<HR>
<PRE>
# Use xanim to view all videos Xanim produces a header on startup,
# send that to /dev/null so I don't see it
video/*; xanim %s > /dev/null
# Send html to a running netscape by remote
text/html; netscape -remote 'openURL(%s)'; test=RunningNetscape
# If I'm not running netscape but I am running X, start netscape on the
# object
text/html; netscape %s; test=RunningX
# Else use lynx to view it as text
text/html; lynx %s
# This version would convert the text/html to text/plain
text/html; lynx -dump %s; copiousoutput
# I use enscript to print text in two columns to a page
text/*; more %s; print=enscript -2Gr %s
# Netscape adds a flag to tell itself to view jpegs internally
image/jpeg;xv %s; x-mozilla-flags=internal
# Use xv to view images if I'm running X
# In addition, this uses the \ to extend the line and set my editor
# for images
image/*;xv %s; test=RunningX; \
edit=xpaint %s
# Convert images to text using the netpbm tools
image/*; (anytopnm %s | pnmscale -xysize 80 46 | ppmtopgm | pgmtopbm |
pbmtoascii -1x2 ) 2>&1 ; copiousoutput
# Send excel spreadsheets to my NT box
application/ms-excel; open.pl %s
</PRE>
<HR>
</P>
<H2><A NAME="auto_view"></A> <A NAME="ss5.4">5.4</A> <A HREF="manual.html#toc5.4">MIME Autoview</A>
</H2>
<P>In addition to explicitly telling Mutt to view an attachment with the
MIME viewer defined in the mailcap file, Mutt has support for
automatically viewing MIME attachments while in the pager.</P>
<P>To work, you must define a viewer in the mailcap file which uses the
<CODE>copiousoutput</CODE> option to denote that it is non-interactive.
Usually, you also use the entry to convert the attachment to a text
representation which you can view in the pager.</P>
<P>You then use the <CODE>auto_view</CODE> muttrc command to list the
content-types that you wish to view automatically.</P>
<P>For instance, if you set auto_view to:
<BLOCKQUOTE><CODE>
<PRE>
auto_view text/html application/x-gunzip application/postscript image/gif application/x-tar-gz
</PRE>
</CODE></BLOCKQUOTE>
</P>
<P>Mutt could use the following mailcap entries to automatically view
attachments of these types.
<BLOCKQUOTE><CODE>
<PRE>
text/html; lynx -dump %s; copiousoutput; nametemplate=%s.html
image/*; anytopnm %s | pnmscale -xsize 80 -ysize 50 | ppmtopgm | pgmtopbm | pbmtoascii ; copiousoutput
application/x-gunzip; gzcat; copiousoutput
application/x-tar-gz; gunzip -c %s | tar -tf - ; copiousoutput
application/postscript; ps2ascii %s; copiousoutput
</PRE>
</CODE></BLOCKQUOTE>
</P>
<P>``unauto_view'' can be used to remove previous entries from the autoview list.
This can be used with message-hook to autoview messages based on size, etc.
``unauto_view *'' will remove all previous entries.</P>
<H2><A NAME="alternative_order"></A> <A NAME="ss5.5">5.5</A> <A HREF="manual.html#toc5.5">MIME Multipart/Alternative</A>
</H2>
<P>Mutt has some heuristics for determining which attachment of a
multipart/alternative type to display. First, mutt will check the
alternative_order list to determine if one of the available types
is preferred. The alternative_order list consists of a number of
mimetypes in order, including support for implicit and explicit
wildcards, for example:
<BLOCKQUOTE><CODE>
<PRE>
alternative_order text/enriched text/plain text application/postscript image/*
</PRE>
</CODE></BLOCKQUOTE>
</P>
<P>Next, mutt will check if any of the types have a defined
<A HREF="#auto_view">auto_view</A>, and use that. Failing
that, Mutt will look for any text type. As a last attempt, mutt will
look for any type it knows how to handle.</P>
<P>To remove a MIME type from the <CODE>alternative_order</CODE> list, use the
<CODE>unalternative_order</CODE> command.</P>
<H2><A NAME="mime_lookup"></A> <A NAME="ss5.6">5.6</A> <A HREF="manual.html#toc5.6">MIME Lookup</A>
</H2>
<P>Mutt's mime_lookup list specifies a list of mime-types that should not
be treated according to their mailcap entry. This option is designed to
deal with binary types such as application/octet-stream. When an attachment's
mime-type is listed in mime_lookup, then the extension of the filename will
be compared to the list of extensions in the mime.types file. The mime-type
associated with this extension will then be used to process the attachment
according to the rules in the mailcap file and according to any other configuration
options (such as auto_view) specified. Common usage would be:
<BLOCKQUOTE><CODE>
<PRE>
mime_lookup application/octet-stream application/X-Lotus-Manuscript
</PRE>
</CODE></BLOCKQUOTE>
</P>
<P>In addition, the unmime_lookup command may be used to disable this feature
for any particular mime-type if it had been set, for example, in a global
muttrc.</P>
<HR>
<A HREF="manual-6.html">Next</A>
<A HREF="manual-4.html">Previous</A>
<A HREF="manual.html#toc5">Contents</A>
</BODY>
</HTML>
