OK, so how do I *use* MIME?

OK, so how do I *use* MIME?

[Hrvoje Niksic]

So now we have MIME handling in Gnus, and it should be possible to
program handling of new data types.  To take an example, I am
receiving patches in an ugly `application/x-gzip64' attachment, which
I want decoded by Gnus (please don't explain how broken that format is 
-- it's only an example.)

So I tried this in .gnus:

    (eval-after-load "mm-decode"
      '(progn
	 (push '("application/x-gunzip" inline-gunzip t)
	       mm-inline-media-tests)
	 (push '("application/x-gunzip" . inline) mm-user-display-methods)))

    (defun inline-gunzip (handle)
      (let ((b (point))
	    (text (mm-get-part handle)))
	(insert text)
	(save-restriction
	  (narrow-to-region b (point))
	  (call-process-region (point-min) (point-max) "gunzip" t t nil)
	  (mm-handle-set-undisplayer
	   handle
	   `(lambda ()
	      (let (buffer-read-only)
		(delete-region ,(point-min-marker) ,(point-max-marker))))))))

And indeed, it is fine, with two exceptions:

1) It does allow me to view the article in the buffer, but saving it
   or piping it to a command still saves the original junk, instead of 
   the decoded stuff.  Is there a way to hook into the way *contents*
   are handled, in order to allow easy saving of the junk?

2) Hooking into `mm-inline-media-tests' and `mm-user-display-methods'
   seems unclean, given that these variables are all but undocumented
   (hi Lloyd!).  How about having a variable for "user" handles, which 
   would be easy to access and modify?

Re: OK, so how do I *use* MIME?

[Lars Magne Ingebrigtsen]

Hrvoje Niksic <hniksic@srce.hr> writes:

> 1) It does allow me to view the article in the buffer, but saving it
>    or piping it to a command still saves the original junk, instead of 
>    the decoded stuff.  Is there a way to hook into the way *contents*
>    are handled, in order to allow easy saving of the junk?

I don't think so.  The CTE is decoded before saving/piping, but other
than that, you deal with the untampered-with contents.  Which seems
logical to me -- when saving a GIF, one doesn't want it fiddled with
before saving, for instance.  

> 2) Hooking into `mm-inline-media-tests' and `mm-user-display-methods'
>    seems unclean, given that these variables are all but undocumented
>    (hi Lloyd!).  How about having a variable for "user" handles, which 
>    would be easy to access and modify?

Yes.  The variable situation in this area is rather confusing.  I'll
try going over it and rename/add/remove variables to create a more
sensible variable set.

-- 
(domestic pets only, the antidote for overdose, milk.)
  larsi@gnus.org * Lars Magne Ingebrigtsen

Re: OK, so how do I *use* MIME?

[Hrvoje Niksic]

Lars Magne Ingebrigtsen <larsi@gnus.org> writes:

> Hrvoje Niksic <hniksic@srce.hr> writes:
> 
> > 1) It does allow me to view the article in the buffer, but saving it
> >    or piping it to a command still saves the original junk, instead of 
> >    the decoded stuff.  Is there a way to hook into the way *contents*
> >    are handled, in order to allow easy saving of the junk?
> 
> I don't think so.  The CTE is decoded before saving/piping, but
> other than that, you deal with the untampered-with contents.  Which
> seems logical to me -- when saving a GIF, one doesn't want it
> fiddled with before saving, for instance.

For GIF, I agree with you.  But in this case, I would *like* to have a 
hook to tamper with the contents at an early stage, similar to what
CTE processing does.

Re: OK, so how do I *use* MIME?

[Lars Magne Ingebrigtsen]

Hrvoje Niksic <hniksic@srce.hr> writes:

> For GIF, I agree with you.  But in this case, I would *like* to have a 
> hook to tamper with the contents at an early stage, similar to what
> CTE processing does.

Well, aren't you kinda asking for double Content-Types, in a way?  The 
thing you got (application/x-gzip, or something) can first be
unpacked, and then the contents of that can be something else (say,
text/plain or whatever).  But there is no framework for doing that in
MIME, and I think it would be a mistake to try to guess (or "guess")
what the second-level (or third-level, etc) type of a part is.

-- 
(domestic pets only, the antidote for overdose, milk.)
  larsi@gnus.org * Lars Magne Ingebrigtsen

Re: OK, so how do I *use* MIME?

[Kai.Grossjohann]

Lars Magne Ingebrigtsen <larsi@gnus.org> writes:

  > Well, aren't you kinda asking for double Content-Types, in a way?  The 
  > thing you got (application/x-gzip, or something) can first be
  > unpacked, and then the contents of that can be something else (say,
  > text/plain or whatever).  But there is no framework for doing that in
  > MIME, and I think it would be a mistake to try to guess (or "guess")
  > what the second-level (or third-level, etc) type of a part is.

Right.  But what if people send you application/x-gzip stuff?  It
should at least be painlessly possible to gunzip it and save it to
disk.

kai
-- 
Just because I'm paranoid doesn't mean they're not out there to get me!

Re: OK, so how do I *use* MIME?

[William M. Perry]

Kai.Grossjohann@CS.Uni-Dortmund.DE writes:

> Lars Magne Ingebrigtsen <larsi@gnus.org> writes:
> 
>   > Well, aren't you kinda asking for double Content-Types, in a way?  The 
>   > thing you got (application/x-gzip, or something) can first be
>   > unpacked, and then the contents of that can be something else (say,
>   > text/plain or whatever).  But there is no framework for doing that in
>   > MIME, and I think it would be a mistake to try to guess (or "guess")
>   > what the second-level (or third-level, etc) type of a part is.
> 
> Right.  But what if people send you application/x-gzip stuff?  It
> should at least be painlessly possible to gunzip it and save it to
> disk.

I would argue that it should just be saved with a '.gz' extension by
default.  That's what they _mailed_ you and explicitly marked the
attachment as being a gzip file.  You could | it to an external command if
you really want to save it raw.

Perhaps we need a way for per-content-type items in the popup menu?  That'd 
be kind of handy in general.  Could do the browse-url thing for text/html,
etc.

-Bill P.

Re: OK, so how do I *use* MIME?

[Hrvoje Niksic]

Lars Magne Ingebrigtsen <larsi@gnus.org> writes:

> Hrvoje Niksic <hniksic@srce.hr> writes:
> 
> > For GIF, I agree with you.  But in this case, I would *like* to have a 
> > hook to tamper with the contents at an early stage, similar to what
> > CTE processing does.
> 
> Well, aren't you kinda asking for double Content-Types, in a way?

If you must say it that way, "double CTE" would be a closer
description.  Anyway, I would like a way to hook into the contents of
the MIME part.

Re: OK, so how do I *use* MIME?

[Lars Magne Ingebrigtsen]

Hrvoje Niksic <hniksic@srce.hr> writes:

> Anyway, I would like a way to hook into the contents of the MIME
> part.

Well, you could just pipe it to zcat, and from there to whatever
displayer you want (with the `|' command).

-- 
(domestic pets only, the antidote for overdose, milk.)
  larsi@gnus.org * Lars Magne Ingebrigtsen

Re: OK, so how do I *use* MIME?

[Hrvoje Niksic]

Lars Magne Ingebrigtsen <larsi@gnus.org> writes:

> Hrvoje Niksic <hniksic@srce.hr> writes:
> 
> > Anyway, I would like a way to hook into the contents of the MIME
> > part.
> 
> Well, you could just pipe it to zcat, and from there to whatever
> displayer you want (with the `|' command).

That doesn't really "hook into" its contents.  With that solution,
viewing inline won't work right, and saving won't work right.

Re: OK, so how do I *use* MIME?

[Lars Magne Ingebrigtsen]

Hrvoje Niksic <hniksic@srce.hr> writes:

> That doesn't really "hook into" its contents.  With that solution,
> viewing inline won't work right, and saving won't work right.

Well, that depends on how you define ",".  I mean, "right".  Or is
that "define"?

Anyway.  I think if someone sends you something that's gzipped, then
that gzipped thing is what you're meant to receive.  So I think the
right thing is to save/view what you got.

I mean, ftp clients don't automatically ungzip files, do they?  They
do?  Er.

-- 
(domestic pets only, the antidote for overdose, milk.)
  larsi@gnus.org * Lars Magne Ingebrigtsen

Re: OK, so how do I *use* MIME?

[Hrvoje Niksic]

Lars Magne Ingebrigtsen <larsi@gnus.org> writes:

> I mean, ftp clients don't automatically ungzip files, do they?  They
> do?  Er.

Note that here I'm not talking about default behaviour, but about a
function I'd like to write for my own use.  The point is not whether
FTP clients automatically ungzip files, but whether a particular
client allows me to automatically ungzip files I want to see and save
ungzipped.

For example, with jka-compr, Emacs' ange-ftp very much automatically
ungzips files.

Re: OK, so how do I *use* MIME?

[Lars Magne Ingebrigtsen]

Hrvoje Niksic <hniksic@srce.hr> writes:

> Note that here I'm not talking about default behaviour, but about a
> function I'd like to write for my own use.  The point is not whether
> FTP clients automatically ungzip files, but whether a particular
> client allows me to automatically ungzip files I want to see and save
> ungzipped.

I've now made the new mm-insert-part function be the only function
that actually does CTE decoding (based on the handle).  I could add a
hook that would let you do further stuff, or we could add a new
variable of Content-Type/function-to-be-called pairs...  Or
something... 

-- 
(domestic pets only, the antidote for overdose, milk.)
  larsi@gnus.org * Lars Magne Ingebrigtsen

Re: OK, so how do I *use* MIME?

[lconrad]

>>>>> "Lars" == Lars Magne Ingebrigtsen <larsi@gnus.org> writes:

    Lars> I mean, ftp clients don't automatically ungzip files, do
    Lars> they?  They do?  Er.

Some of them do some of the time.  The RedHat mailing list was in a
state of utter confusion for several days when the new wordperfect
distribution came out and some people could use it by following the
instructions and other people couldn't.  

-- 
Laura (mailto:lconrad@world.std.com , http://www.world.std.com/~lconrad/ )
(617) 661-8097	 
233 Broadway, Cambridge, MA 02139

Re: OK, so how do I *use* MIME?

[Lars Magne Ingebrigtsen]

lconrad@world.std.com writes:

>     Lars> I mean, ftp clients don't automatically ungzip files, do
>     Lars> they?  They do?  Er.
> 
> Some of them do some of the time.  The RedHat mailing list was in a
> state of utter confusion for several days when the new wordperfect
> distribution came out and some people could use it by following the
> instructions and other people couldn't.  

Yup.  I think clients doing automatic decoding (beyond what's defined
by the protocol (whether it be ftp or rfc822/MIME) is evil.

-- 
(domestic pets only, the antidote for overdose, milk.)
  larsi@gnus.org * Lars Magne Ingebrigtsen

Re: OK, so how do I *use* MIME?

[Steinar Bang]

>>>>> Lars Magne Ingebrigtsen <larsi@gnus.org>:

> I mean, ftp clients don't automatically ungzip files, do they?  They
> do?  Er.

Netscape 4.5 seems to be removing any .gz extensions, without
gunzip'ing... 

Re: OK, so how do I *use* MIME?

[Lars Magne Ingebrigtsen]

Steinar Bang <sb@metis.no> writes:

> Netscape 4.5 seems to be removing any .gz extensions, without
> gunzip'ing... 

How utterly useful.  :-)

-- 
(domestic pets only, the antidote for overdose, milk.)
  larsi@gnus.org * Lars Magne Ingebrigtsen

Re: OK, so how do I *use* MIME?

[Graham Murray]

Steinar Bang <sb@metis.no> writes:

> Netscape 4.5 seems to be removing any .gz extensions, without
> gunzip'ing... 

It normally seems to be gunzipping for me. When I use it to download
gnus patches, I can use the downloaded file directly in patch without
having to gunzip first.

Re: OK, so how do I *use* MIME?

[Andrew Hobson]

On 14 Jan 1999 20:17:36 +0100, Lars Magne Ingebrigtsen <larsi@gnus.org> said:

> Hrvoje Niksic <hniksic@srce.hr> writes:
>> Anyway, I would like a way to hook into the contents of the MIME
>> part.

> Well, you could just pipe it to zcat, and from there to whatever
> displayer you want (with the `|' command).

What if you receive a quoted-printable message and you want to save
the decoded version?  I don't have an external program that does the
decoding since gnus works so well.  The only way I can think to do it
is to cut and paste and, well, that leaves something to be desired.

Drew

-- 
"Joe, release me from your Kung-Fu grip." -- Stacy Lavelle

Re: OK, so how do I *use* MIME?

[Lars Magne Ingebrigtsen]

Andrew Hobson <ahobson@eng.mindspring.net> writes:

> What if you receive a quoted-printable message and you want to save
> the decoded version?

Do you mean Content-Type: application/quoted-printable?  I've never
seen anything like that.

Content-Type-Encoding is always decoded, as per the protocols.

-- 
(domestic pets only, the antidote for overdose, milk.)
  larsi@gnus.org * Lars Magne Ingebrigtsen

Re: OK, so how do I *use* MIME?

[William M. Perry]

Hrvoje Niksic <hniksic@srce.hr> writes:

> Lars Magne Ingebrigtsen <larsi@gnus.org> writes:
> 
> > Hrvoje Niksic <hniksic@srce.hr> writes:
> > 
> > > 1) It does allow me to view the article in the buffer, but saving it
> > >    or piping it to a command still saves the original junk, instead of 
> > >    the decoded stuff.  Is there a way to hook into the way *contents*
> > >    are handled, in order to allow easy saving of the junk?
> > 
> > I don't think so.  The CTE is decoded before saving/piping, but
> > other than that, you deal with the untampered-with contents.  Which
> > seems logical to me -- when saving a GIF, one doesn't want it
> > fiddled with before saving, for instance.
> 
> For GIF, I agree with you.  But in this case, I would *like* to have a
> hook to tamper with the contents at an early stage, similar to what CTE
> processing does.

The problem here is that whoever is sending you such heinous attachments
with a CONTENT-TYPE of application/x-gzip64 should really be sending it as
application/octet-stream (or text/plain or whatever) with a
CONTENT-TRANSFER-ENCODING of x-gzip64

Then you could define just one new handler for the CTE x-gzip64 and all
would be well.  I assume such a beasty would have to base64 decode and then 
gunzip it.  I have no idea if this is easy to do with the mm stuff in pGnus 
or not though.

I think Gnus is doing the right thing by giving you the raw data, since
that is what the content-type displayers _should_ get.  The fact that
application/x-gzip64 is an abomination is a totally different problem. :)

Perhaps there is a way to hook into the article parsing before mimification 
begins and change the CTE header if the CT is application/x-gzip64?  That
might work...

-Bill P.

Re: OK, so how do I *use* MIME?

[Jari Aalto+list.ding]

* 1999-01-13 wmperry@aventail.com (William Perry) list.ding
| ...The fact that
| application/x-gzip64 is an abomination is a totally different problem. :)

What is the proper MIME type for for base64 encoded gzip files ?
I've also seen plain "application/gzip"

jari

Re: OK, so how do I *use* MIME?

[William M. Perry]

jari.aalto@poboxes.com (Jari Aalto+list.ding) writes:

> * 1999-01-13 wmperry@aventail.com (William Perry) list.ding
> | ...The fact that
> | application/x-gzip64 is an abomination is a totally different problem. :)
> 
> What is the proper MIME type for for base64 encoded gzip files ?
> I've also seen plain "application/gzip"

  i assume one would need an appropriate content-transfer-encoding.  Not a
content-type for this.  So you would have:

Content-type: application/x-tar
Content-transfer-encoding: x-base64-gzip

[base64 gobbledygook]

I don't know if such a beast exists or not.  Or just have a content-type of 
application/octet-stream and a content-transfer-encoding of 'base64' and
let the user figure it out.

-Bill P.

Re: OK, so how do I *use* MIME?

[Richard Coleman]

> > | ...The fact that
> > | application/x-gzip64 is an abomination is a totally different problem. :)
> > 
> > What is the proper MIME type for for base64 encoded gzip files ?
> > I've also seen plain "application/gzip"
> 
>   i assume one would need an appropriate content-transfer-encoding.  Not a
> content-type for this.  So you would have:
> 
> Content-type: application/x-tar
> Content-transfer-encoding: x-base64-gzip
> 
> [base64 gobbledygook]
> 
> I don't know if such a beast exists or not.  Or just have a content-type of 
> application/octet-stream and a content-transfer-encoding of 'base64' and
> let the user figure it out.

Defining your own CTE such as x-base64-gzip is strongly discouraged
by the MIME RFC's.  Most e-mail clients would be at a loss on how to
handle these.  Many would just bomb out and not even give you the
opportunity to save it to a file.

Since you would have to make arrangements beforehand on what the
format of these message would be, you might as will just use:

Content-transfer-encoding: base64
Content-type: application/x-gzip

and make arrangements on how to handle the file after the base64 is
removed.

--
Richard Coleman
coleman@math.gatech.edu