mac/dos/unix newline conversion without specify from

mac/dos/unix newline conversion without specify from

[Xah Lee]

i was surprised today, to find that Carbon Emacs 22 and also Aquamacs,
does not open a file with classic Mac OS line endings properly? (i.e.
EOL shows up as ^M)

i realized this when trying to write a elisp program to do the
conversion given a file name.

originally, i thought set-buffer-file-coding-system does it. But not
so if the file opened is a classic mac os file with CR as newline.
After some elisp doc reading, i find that i have to call (coding-
system-for-read 'mac) first. But then, it is not better if i actually
just do the newline replacement myself like this:

; code untested
(defun xx ()
  "temp... Convert list of files to unix/mac/dos's newline char."
  (interactive)
  (let (fpath mybuffer)
    (setq fpath "~/web/emacs/x2/x1mac")
    (setq mybuffer (find-file fpath))
    ; note: Mac here denote Mac OS up to 9.
    ; In Mac OS X, both unix newline and old mac newline are accepted,
but unix newline is recommended.
    (replace-string "\n" "\r" nil 1 (1+ (buffer-size))) ; unix to mac
    (replace-string "\n" "\r\n" nil 1 (1+ (buffer-size))) ; unix to
dos
    (replace-string "\r" "\n" nil 1 (1+ (buffer-size))) ; mac to unix
    (replace-string "\r" "\r\n" nil 1 (1+ (buffer-size))) ; mac to dos
    (replace-string "\r\n" "\n" nil 1 (1+ (buffer-size))) ; dos to
unix
    (replace-string "\r\n" "\r" nil 1 (1+ (buffer-size))) ; dos to mac
;    (set-buffer-file-coding-system 'unix) ; or 'mac or 'dos
    (save-buffer)
    (kill-buffer mybuffer)
   )
)

Basically, i was trying to avoid having both to specify the from and
to. I was hoping to just specifiy the to.

any suggestions? Thanks.

  Xah
  xah@xahlee.org
\xAD\xF4 http://xahlee.org/

Re: mac/dos/unix newline conversion without specify from

[David Reitter]

Xah Lee wrote:

> i was surprised today, to find that Carbon Emacs 22 and also Aquamacs,
> does not open a file with classic Mac OS line endings properly? (i.e.
> EOL shows up as ^M)

I just tried it and it works for me (Aquamacs from CVS). The coding
shows up as "Mac" in the mode-line.

Re: mac/dos/unix newline conversion without specify from

[Xah Lee]

Xah Lee wrote:
<<Carbon Emacs 22 and also Aquamacs (launched with -q), does not open a
file with classic Mac OS line endings properly? (i.e.  EOL shows up as
^M)>>

David Reitter wrote:
<<I just tried it and it works for me (Aquamacs from CVS). The coding
shows up as "Mac" in the mode-line.>>

I was able to reproduce this 100% on Carbon emacs and aquamacs without
customization (emacs -q).

GNU Emacs 22.1.50.1 (powerpc-apple-darwin8.10.0, Carbon Version 1.6.0)
of 2007-11-06 on cf-ppc-macosx.usr.sourceforge.jp

GNU Emacs 22.1.50.1 (powerpc-apple-darwin7.9.0, Carbon Version 1.6.0)
of 2007-10-01 on applecore.inf.ed.ac.uk - Aquamacs Distribution 1.2a

Could it be just a version 22 thing?

  Xah
  xah@xahlee.org
\xAD\xF4 http://xahlee.org/

Re: mac/dos/unix newline conversion without specify from

[David Kastrup]

Xah Lee <xah@xahlee.org> writes:

> Xah Lee wrote:
> <<Carbon Emacs 22 and also Aquamacs (launched with -q), does not open a
> file with classic Mac OS line endings properly? (i.e.  EOL shows up as
> ^M)>>
>
> David Reitter wrote:
> <<I just tried it and it works for me (Aquamacs from CVS). The coding
> shows up as "Mac" in the mode-line.>>
>
> I was able to reproduce this 100% on Carbon emacs and aquamacs without
> customization (emacs -q).

More likely than not, you have a file with inconsistent line end
characters.  In that case, Emacs has to choose "-unix" line encoding or
it would change the file when writing it back to disk.

-- 
David Kastrup, Kriemhildstr. 15, 44793 Bochum

Re: mac/dos/unix newline conversion without specify from

[Xah Lee]

Xah Lee wrote:
<<Carbon Emacs 22 and also Aquamacs (launched with -q), does not open a
file with classic Mac OS line endings properly? (i.e.  EOL shows up as
^M)>>

David Kastrup:
<<More likely than not, you have a file with inconsistent line end
characters.  In that case, Emacs has to choose "-unix" line encoding
or it would change the file when writing it back to disk.>>

jackpot :)

Open a unix file, do replace-string from LF to CR. Save the file. Then
start emacs with -q, open the file. It doesn't interprete the file as
a mac os classic file but instead display newline char as ^M.

If i open the file in TextWrangler, TextEdit, Xcode, they open fine
and indicate that mac os line ending are used.

-----------------

given the current situation, how to instruct emacs (as a user) to open
a file with CR as EOL?

(am interested in this answer for my emacs tutorial)

  Xah
  xah@xahlee.org
\xAD\xF4 http://xahlee.org/

On Dec 5, 7:05 am, David Kastrup <d...@gnu.org> wrote:
> XahLee<x...@xahlee.org> writes:
> >XahLeewrote:
>
> > David Reitter wrote:
> > <<I just tried it and it works for me (Aquamacs from CVS). The coding
> > shows up as "Mac" in the mode-line.>>
>
> > I was able to reproduce this 100% on Carbon emacs and aquamacs without
> > customization (emacs -q).
>
> More likely than not, you have a file with inconsistent line end
> characters.  In that case, Emacs has to choose "-unix" line encoding or
> it would change the file when writing it back to disk.
>
> --
> David Kastrup, Kriemhildstr. 15, 44793 Bochum

Re: mac/dos/unix newline conversion without specify from

[David Kastrup]

Xah Lee <xah@xahlee.org> writes:

> Xah Lee wrote:
> <<Carbon Emacs 22 and also Aquamacs (launched with -q), does not open a
> file with classic Mac OS line endings properly? (i.e.  EOL shows up as
> ^M)>>
>
> David Kastrup:
> <<More likely than not, you have a file with inconsistent line end
> characters.  In that case, Emacs has to choose "-unix" line encoding
> or it would change the file when writing it back to disk.>>
>
> jackpot :)

Huh?  Inconsistent with what you write now:

> Open a unix file, do replace-string from LF to CR. Save the file. Then
> start emacs with -q, open the file. It doesn't interprete the file as
> a mac os classic file but instead display newline char as ^M.

It does switch to Mac endings here in this case.  Just tried it.

> given the current situation, how to instruct emacs (as a user) to open
> a file with CR as EOL?

You can always do something like C-x RET c latin-1-mac RET C-x C-f file RET
but again: this should not be necessary at all.

-- 
David Kastrup, Kriemhildstr. 15, 44793 Bochum

Re: mac/dos/unix newline conversion without specify from

[Xah Lee]

Xah Lee wrote:
<<Open a unix file, do replace-string from LF to CR. Save the file.
Then start emacs with -q, open the file. It doesn't interprete the
file as a mac os classic file but instead display newline char as ^M.>>

David Kastrup wrote:
<<
It does switch to Mac endings here in this case.  Just tried it.
>>

mm... interesting. Here's a sample file for what's worth:
http://xahlee.org/emacs/x-unixmacdos-eol/x1mac

Xah wrote:
<<given the current situation, how to instruct emacs (as a user) to
open a file with CR as EOL?>>

David wrote:
<<
You can always do something like C-x RET c latin-1-mac RET C-x C-f
file RET
>>

Thanks.

  Xah
  xah@xahlee.org
\xAD\xF4 http://xahlee.org/

Re: mac/dos/unix newline conversion without specify from

[David Kastrup]

Xah Lee <xah@xahlee.org> writes:

> Xah Lee wrote:
> <<Open a unix file, do replace-string from LF to CR. Save the file.
> Then start emacs with -q, open the file. It doesn't interprete the
> file as a mac os classic file but instead display newline char as ^M.>>
>
> David Kastrup wrote:
> <<
> It does switch to Mac endings here in this case.  Just tried it.
>>>
>
> mm... interesting. Here's a sample file for what's worth:
> http://xahlee.org/emacs/x-unixmacdos-eol/x1mac

That's because html-mode sets require-final-newline to t.  If you reset
it to nil and then save, you get the file saved without a final
(non-Macish-looking) LF character, and it gets recognized correctly when
loading it again.

-- 
David Kastrup, Kriemhildstr. 15, 44793 Bochum

Re: mac/dos/unix newline conversion without specify from

[Xah Lee]

David Kastrup wrote:
<<That's because html-mode sets require-final-newline to t. >>

Thanks for finding the cause.

Anyhow, for what's worth, i think this is still something emacs (in
particular Mac versions) should fixup. Basically, the scenario is that
a user opens mac classic html files, work on it, save it, then next
time he opens the files shows ^M. (this actually happens to me a lot
since i work on a server with non-professional coders and they use
BBEdit that is still set to CR as newline) In contrast, i open the
same file (with CR as eol but a LF at the end) in Xcode, TextWrangler,
TextEdit, all opens correctly and indicate as Mac files.

* perhaps require-final-newline should insert the right EOL char based
on the encoding.

* perhaps emacs file opening routine should be more smart. (i.e. not
using the last EOL, or check more lines, to determine the EOL char)

(note: this post contains the french double quote char "<<>>" (unicode
00AB 00BB). Google groups may have botched it to <<>> or omitted it)

  Xah
  xah@xahlee.org
\xAD\xF4 http://xahlee.org/



On Dec 5, 9:05 am, David Kastrup <d...@gnu.org> wrote:
> Xah Lee <x...@xahlee.org> writes:
> > Xah Lee wrote:
> > <<Open a unix file, do replace-string from LF to CR. Save the file.
> > Then start emacs with -q, open the file. It doesn't interprete the
> > file as a mac os classic file but instead display newline char as ^M.>>
>
> > David Kastrup wrote:
> > <<
> > It does switch to Mac endings here in this case.  Just tried it.
>
> > mm... interesting. Here's a sample file for what's worth:
> >http://xahlee.org/emacs/x-unixmacdos-eol/x1mac
>
> That's because html-mode sets require-final-newline to t.  If you reset
> it to nil and then save, you get the file saved without a final
> (non-Macish-looking) LF character, and it gets recognized correctly when
> loading it again.
>
> --
> David Kastrup, Kriemhildstr. 15, 44793 Bochum

Re: mac/dos/unix newline conversion without specify from

[Stefan Monnier]

> <<That's because html-mode sets require-final-newline to t. >>
> Thanks for finding the cause.

That's probably not the cause of your original problem.  It's the cause
of the problem you saw when you tried the replace-string thingy to
replace "all" CRs into LFs.  If you used the C-x RET f mac RET method
I suggested earlier to convert unix-style files to mac-style files, this
problem shouldn't appear.

> Anyhow, for what's worth, i think this is still something emacs (in
> particular Mac versions) should fixup. Basically, the scenario is that
> a user opens mac classic html files, work on it, save it, then next
> time he opens the files shows ^M. (this actually happens to me a lot

That's a clear bug, indeed.  But I can guarantee that it normally works,
so please give us more details so we can track down the problem.
The only know case where a Mac file is misidentified is if it contains
LF chars, which shouldn't happen in normal circumstances.

> since i work on a server with non-professional coders and they use
> BBEdit that is still set to CR as newline) In contrast, i open the
> same file (with CR as eol but a LF at the end) in Xcode, TextWrangler,
> TextEdit, all opens correctly and indicate as Mac files.

The final LF is clearly an error in the file.  You need to track down
its origin.

> * perhaps require-final-newline should insert the right EOL char based
> on the encoding.

It does.


        Stefan

Re: mac/dos/unix newline conversion without specify from

[Stefan Monnier]

> Open a unix file, do replace-string from LF to CR. Save the file. Then
> start emacs with -q, open the file. It doesn't interprete the file as
> a mac os classic file but instead display newline char as ^M.

That's a good method if you like to get into trouble.

For those who prefer to do things right, then you can C-x RET f mac RET
and then C-x C-s, or alternatively just click twice on the ":" in the
left part of the mode line (the first time makes it change from unix to
dos, the second from dos to mac).


        Stefan

Re: mac/dos/unix newline conversion without specify from

[Xah Lee]

what you described wouldn't work if the file is not opened correctly.

in particular, open a file with CR as line return but containing a LF
as the last char. Then, open it in emacs and emacs will show ^M. At
this point, changing the encoding as you described only changes the
last char.

 Xah

On Dec 7, 8:27 am, Stefan Monnier <monn...@iro.umontreal.ca> wrote:
> > Open a unix file, do replace-string from LF to CR. Save the file. Then
> > start emacs with -q, open the file. It doesn't interprete the file as
> > a mac os classic file but instead display newline char as ^M.
>
> That's a good method if you like to get into trouble.
>
> For those who prefer to do things right, then you can C-x RET f mac RET
> and then C-x C-s, or alternatively just click twice on the ":" in the
> left part of the mode line (the first time makes it change from unix to
> dos, the second from dos to mac).
>
>         Stefan

Re: mac/dos/unix newline conversion without specify from

[Stefan Monnier]

> what you described wouldn't work if the file is not opened correctly.

It also doesn't work if you want to cook an egg.  But that's not very
relevant, is it?  As you can see in the text I quoted, I was replying to
method that starts with "Open a unix file, ...".


        Stefan


> in particular, open a file with CR as line return but containing a LF
> as the last char. Then, open it in emacs and emacs will show ^M. At
> this point, changing the encoding as you described only changes the
> last char.

>  Xah

> On Dec 7, 8:27 am, Stefan Monnier <monn...@iro.umontreal.ca> wrote:
>> > Open a unix file, do replace-string from LF to CR. Save the file. Then
>> > start emacs with -q, open the file. It doesn't interprete the file as
>> > a mac os classic file but instead display newline char as ^M.
>> 
>> That's a good method if you like to get into trouble.
>> 
>> For those who prefer to do things right, then you can C-x RET f mac RET
>> and then C-x C-s, or alternatively just click twice on the ":" in the
>> left part of the mode line (the first time makes it change from unix to
>> dos, the second from dos to mac).
>> 
>> Stefan

Re: mac/dos/unix newline conversion without specify from

[Xah Lee]

Dear Stefan Monnier,

When is the last time you sucked your mom's tits?

  Xah
  xah@xahlee.org
\xAD\xF4 http://xahlee.org/

On Dec 7, 2:30 pm, Stefan Monnier <monn...@iro.umontreal.ca> wrote:
> > what you described wouldn't work if the file is not opened correctly.
>
> It also doesn't work if you want to cook an egg.  But that's not very
> relevant, is it?  As you can see in the text I quoted, I was replying to
> method that starts with "Open a unix file, ...".
>
>         Stefan
>
> > in particular, open a file with CR as line return but containing a LF
> > as the last char. Then, open it in emacs and emacs will show ^M. At
> > this point, changing the encoding as you described only changes the
> > last char.
> >  Xah
> > On Dec 7, 8:27 am, Stefan Monnier <monn...@iro.umontreal.ca> wrote:
> >> > Open a unix file, do replace-string from LF to CR. Save the file. Then
> >> > start emacs with -q, open the file. It doesn't interprete the file as
> >> > a mac os classic file but instead display newline char as ^M.
>
> >> That's a good method if you like to get into trouble.
>
> >> For those who prefer to do things right, then you can C-x RET f mac RET
> >> and then C-x C-s, or alternatively just click twice on the ":" in the
> >> left part of the mode line (the first time makes it change from unix to
> >> dos, the second from dos to mac).
>
> >> Stefan

Re: mac/dos/unix newline conversion without specify from

[Eli Zaretskii]

> From: Xah Lee <xah@xahlee.org>
> Date: Fri, 7 Dec 2007 16:38:57 -0800 (PST)
> 
> Dear Stefan Monnier,
> 
> When is the last time you sucked your mom's tits?

Please never again use such tone in this forum.  There was nothing in
Stefan's reply to justify such indecency, not to mention OT content.
If you were somehow offended, please express that in civilized form.

Re: mac/dos/unix newline conversion without specify from

[Xah Lee]

Sorry. I deleted that from google groups within 3 minutes of post. But
too late i guess.

...

 Xah

On Dec 8, 3:47 am, Eli Zaretskii <e...@gnu.org> wrote:
> > From:XahLee<x...@xahlee.org>
> > Date: Fri, 7 Dec 2007 16:38:57 -0800 (PST)
>
> > Dear Stefan Monnier,
>
> > When is the last time you sucked your mom's tits?
>
> Please never again use such tone in this forum.  There was nothing in
> Stefan's reply to justify such indecency, not to mention OT content.
> If you were somehow offended, please express that in civilized form.

Re: mac/dos/unix newline conversion without specify from

[David Kastrup]

Xah Lee <xah@xahlee.org> writes:

> Sorry. I deleted that from google groups within 3 minutes of post. But
> too late i guess.

This particular group is gated to a mailing list, so canceling is not
really effective at all for a lot of readers.

-- 
David Kastrup, Kriemhildstr. 15, 44793 Bochum