Eliminating CR (new lines) from a file.

KornFire · 01-28-2009, 11:28 PM

Hi all, I made a C++ program in dos (in dev-C++) and uploaded it on Solaris box. On opening that file with 'vim' editor i found that there is some extra new lines after each written code line. I tried to find out is the file is in dos or in unix format, with 'file' command,and i got "<file-name>.h: ASCII C++ program text". After which i used 'dos2unix' command which shows 'converting file to unix format' but new-lines didn't disappear. Any help on this will be most welcome.

Thanks,

chrism01 · 01-29-2009, 12:54 AM

Its possible that if you saved it in vim, vim converted all the lines, so you've got double (Unix) new-lines.
Just upload from DOS and convert immediately before doing anything else.
Better yet, code on solaris in the first place (or any *nix).

colucix · 01-29-2009, 01:31 AM

To see what alien characters are appended to each line, post the result of

Code:

head <file-name>.h | od -c

then we can find a sed regexp to strip them out. As chrism01 already pointed out, download the file again and do the command above before any other attempt.

KornFire · 01-29-2009, 01:32 AM

is there any command (may be 'tr' cmd, i think) to remove repeated (Unix) new line to single one. I know it is possible to remove all new line with:

Code:

 tr -s '\n' < infile > outfile

but it's not exactly i wanted. it removes all blank line from code.

Thanks,

KornFire · 01-29-2009, 01:39 AM

Hi colucix,
The output of 'head' is as below:

Code:

[xyz@blr api]$ head terminal.h.bak | od -c
0000000   /   /   *   *   *   *   *   *   *   *   *   *   *   *   *   *
0000020   *   *   *   *   *   *   *   *   *   *   *   *   *   *   *   *
*
0000120   *   *   *   *   *   *  \n  \n   /   /                   N   a
0000140   m   e   :       T   e   r   m   i   n   a   l   .   h  \n  \n
0000160   /   /  \n  \n   /   /                   D   e   s   c   r   i
0000200   p   t   i   o   n   :       T   h   i   s       f   i   l   e
0000220       p   r   o   v   i   d   e   s       t   h   e       d   e
0000240   c   l   a   r   a   t   i   o   n       o   f       t   h   e
0000260       T   e   r   m   i   n   a   l       o   b   j   e   c   t
0000300   .  \n  \n   /   /  \n  \n
0000307

Please have a look.

Thanks,

colucix · 01-29-2009, 02:22 AM

It confirms you have two consecutive newlines as Chris already guessed. Anyway, I'd do the tr command to strip them out or eventually a sed one-liner like

Code:

sed -i '/^$/d' file.h

but this actually removes all the blank lines. How do you think to remove some blank lines only and not other? Which is the criterium?

KornFire · 01-29-2009, 02:56 AM

If criteria of deletion of new line we are talking about then it should be like deleting only one '\n' among repetition of it. Every Line in unix end with delimiter '\n' and if by any means, i'm getting one more blank line (mean one more '\n' entry) then i think any command that could delete only one '\n' among consecutive '\n' entries will be useful. Though, i'm not sure about it.

colucix · 01-29-2009, 03:48 AM

I've got the point. Anyway, a double \n\n equals to blank lines, so removing blank lines using sed should give you the result you want to achieve. For example try this:

Code:

sed '/^$/d' file.h | head | od -c

and see the difference with the previous output of od -c.

JulianTosh · 01-29-2009, 07:33 AM

tr "\n" ""