Avoid carriage return until ^M is found (CentOS 6, bash 4.1)

gacanepa · 10-25-2013, 08:10 AM

Hi everyone,
I have the following contents in a text file (as seen when viewed using vim):

Code:

one two three ^M
four five six ^M
seven
eight
nine ^M
ten eleven twelve ^M

(That is just a small portion of the file)
How can I obtain the following result?

Code:

one two three ^M
four five six ^M
seven eight nine ^M
ten eleven twelve ^M

If there are words in a certain line that does not end with ^M, they should be all in one line until a ^M appears.
Thanks in advance.

druuna · 10-25-2013, 11:05 AM

Give this a try:

Code:

sed -i ':loop;/^M/!{N;bloop};s/\n/ /g' infile

The ^M is special and can be created like this: type ctrl-v then ctrl-m

As a side effect the ^M's are not longer visible when opened in vi. This because the file is now a dos file and not a mixed dos/linux file.

Below a dump of the input file.

Before:

Code:

$ od -c infile
0000000   o   n   e       t   w   o       t   h   r   e   e      \r  \n
0000020   f   o   u   r       f   i   v   e       s   i   x      \r  \n
0000040   s   e   v   e   n  \n   e   i   g   h   t  \n   n   i   n   e
0000060      \r  \n   t   e   n       e   l   e   v   e   n       t   w
0000100   e   l   v   e      \r  \n
0000107

After:

Code:

0000000   o   n   e       t   w   o       t   h   r   e   e      \r  \n
0000020   f   o   u   r       f   i   v   e       s   i   x      \r  \n
0000040   s   e   v   e   n       e   i   g   h   t       n   i   n   e
0000060      \r  \n   t   e   n       e   l   e   v   e   n       t   w
0000100   e   l   v   e      \r  \n
0000107

The \r \n sequence represents the dos newline (the ^M). A single \n is a linux newline.

grail · 10-25-2013, 12:34 PM

You could also try awk:

Code:

awk -F"[ \n]" 'BEGIN{ORS=RS="\r\n"}$1=$1' file