How to strip CR (^M) and leave LF (^J) characters?

Question

I am trying to use Hexl mode to manually remove some special characters from a text file and don't see how to delete anything in Hexl mode.

What I really want is to remove carriage return and keep linefeed characters.
Is Hexl mode the right way to do this?

I'm trying to remove a carriage return and leave a line feed...don't ask - the program that's reading this config file wants it that way. — Brandon Leiran, Jul 23 '09 at 15:32
+1 I had a similar issue. For some odd reason, a bunch of hex characters were prepended to the beginning of some of my text files. I just want to delete those characters, but can't see them in regular buffer mode. Why is the "delete" feature so hard for hexl-mode? It must be there. — User1, Jun 22 '10 at 14:14

CantGetANick · Answer 1 · 2011-11-01T05:13:25.353

99

No need to find replace. Just use.

M-x delete-trailing-whitespace

You can also set the file encoding through

C-x RET f unix

edited Nov 01 '11 at 05:13

answered Dec 10 '10 at 05:28

CantGetANick

1,799
15
25

score 33 · Answer 2 · edited Nov 18 '14 at 03:47

33

Oops. That ^J^M needs to be entered as two literal characters. Use c-q c-j, c-q c-m and for the replacement string, use c-q c-j.

edited Nov 18 '14 at 03:47

yurisich

6,991
7
42
63

answered Jul 23 '09 at 18:29

I can't believe how few votes this has! This would totally save a novice (or forgetful) Emacs user! – Ogre Psalm33 Jan 13 '15 at 14:01
2

Yes, this should be higher. Also, you can just replace ^M and ignore the ^J. – Ashton Honnecke Nov 13 '15 at 17:29

score 10 · Accepted Answer · edited Jul 27 '09 at 19:18

10

No need for hexl-mode for this. Just do a global-search-and-replace of ^J^M with ^J Works for me. :) Then save the file, kill the buffer, and revisit the file so the window shows the new file mode (Unix vs DOS).

edited Jul 27 '09 at 19:18

Sean Bright

118,630
17
138
146

answered Jul 23 '09 at 18:25

2

Why even bother to do this? Do what keysersoze suggests and use dos2unix and/or unix2dos. – Thomas Owens Jul 27 '09 at 19:20
Instead of killing the buffer and revisiting the file, you can click on the EOL type indicator in the mode line to cycle through the different options. This way you won't lose your kill ring. – phils Dec 12 '10 at 00:39
1

would have been helpful to note how to actually do this - `M-%` doesn't work like this for me – baxx Oct 18 '15 at 21:20

score 9 · Answer 4 · answered Jul 27 '09 at 19:14

9

There's also a command-line tool called unix2dos/dos2unix that exists specifically to convert line endings.

answered Jul 27 '09 at 19:14

KeyserSoze

2,501
1
16
17

score 8 · Answer 5 · answered Jul 28 '09 at 07:35

8

Assuming you want a DOS encoded file to be changed into UNIX encoding, use M-x set-buffer-file-coding-system (C-x RET f) to set the coding-system to "unix" and save the file.

answered Jul 28 '09 at 07:35

remvee

829
6
14

score 6 · Answer 6 · answered Jul 23 '09 at 18:07

If you want to remove a carriage return (usually displayed as ^M) and leave the line feed. You can just visit the file w/out any conversion:

M-x find-file-literally /path/to/file

Because a file with carriage returns is generally displayed in DOS mode (hiding the carriage returns). The mode line will likely display (DOS) on the left side.

Once you've done that, the ^M will show up and you can delete them like you would any character.

score 4 · Answer 7 · answered Jul 28 '09 at 00:48

You don't need to use hexl-mode. Instead:

open file in a way that shows you those ^M's. See M-x find-file-literally /path/to/file above. In XEmacs you can also do C-u C-x C-f and select binary encoding.
select the string you want replace and copy it using M-w
do M-% (query replace) and paste what you want to copy using C-y
present Enter when prompted to what replace it with
possible press ! now to replace all occurrences

The point is that even if you don't how to enter what you are trying to replace, you can always select/copy it.

Thanks, didn't know about find-file-literally, actually a better tool for seeing control characters in text files than hexl mode which I'd been using. — Marc Stober, Nov 22 '11 at 15:52

score 3 · Answer 8 · edited Jul 27 '09 at 19:09

3

(in hexl mode) I'm not sure that you can delete characters. I've always converted them to spaces or some other character, switched to the regular text editor, and deleted them there.

edited Jul 27 '09 at 19:09

Cheeso

189,189
101
473
713

answered Jul 23 '09 at 13:34

Chris Arguin

11,850
4
34
50

score 2 · Answer 9 · answered Dec 17 '19 at 08:36

I use this function:

(defun l/cr-sanitise ()
  "Make sure current buffer uses unix-utf8 encoding.
If necessary remove superfluous ^M. Buffer will need to be saved
for changes to be permanent."
  (interactive)
    (set-buffer-file-coding-system 'utf-8-unix)
    (delete-trailing-whitespace)
    (message "Please save buffer to persist encoding changes."))

ayman · Answer 10 · 2009-08-13T23:10:57.087

From http://www.xsteve.at/prg/emacs/xsteve-functions.el:

;02.02.2000
(defun xsteve-remove-control-M ()
  "Remove ^M at end of line in the whole buffer."
  (interactive)
  (save-match-data
    (save-excursion
      (let ((remove-count 0))
        (goto-char (point-min))
        (while (re-search-forward (concat (char-to-string 13) "$") (point-max) t)
          (setq remove-count (+ remove-count 1))
          (replace-match "" nil nil))
        (message (format "%d ^M removed from buffer." remove-count))))))

Add this to your .emacs and run it via M-x xsteve-remove-control-M or bind it to a easier key. It will strip the ^Ms in anymode.

This is a great solution. One command to fix the whole file instantly. — DOKKA, Aug 20 '14 at 17:08

How to strip CR (^M) and leave LF (^J) characters?

10 Answers10

Linked