Can't remove empty lines with sed regex

Question

I have a file like this:

2733617     3.00    3   3

2733617 E1b1    8.00    8   16
2733617 E1b1b1b 2.00    2   4

2733617 I1  294.00  296 590
2733617 I2  1.00    1   2

2733617 I2a1    2.00    2   4

sed '/^$/d' does not work for me. Outfile looks just like infile. It should remove empty lines.

check that there is no whitespace (including carriage returns) on those "empty" lines: `od -c filename` — glenn jackman, May 25 '12 at 14:17
Thank you for a command good to know. But there is only "\n" in empty lines. — AWE, May 25 '12 at 14:38
`sed -n l file` will show what `sed` is seeing once it has removed the newlines. — potong, May 25 '12 at 15:48
How strange. Your command should work. Could you post the output of `od -x`? — Jeremy Stein, Jun 04 '12 at 17:16

score 4 · Answer 1 · answered Jun 24 '13 at 22:41

4

to delete blank lines:

sed '/^[[:space:]]*$/d'

answered Jun 24 '13 at 22:41

glenn jackman

238,783
38
220
352

Glenn, excuse me. I'm total noob with `sed`. Could you explain me what does the `d` modifier do? I've googled it with no result :) – Rango Nov 21 '15 at 18:53
I wonder why if the very last line is empty, it does not get removed. – Shayan Feb 09 '23 at 12:22
I'm editing in place btw. – Shayan Feb 09 '23 at 12:57

score 3 · Answer 2 · edited Jun 24 '13 at 20:58

3

Unfortunately the manual says that using ranges like [!-~] is not safe. However, just printing the lines that containing printable characters using :print: worked for me in the end:

sed -n '/[[:print:]]/p'

edited Jun 24 '13 at 20:58

answered Nov 16 '12 at 11:22

coltox

360
3
8

This is such an old question I have no idea where this mysterious file is today. – AWE Nov 16 '12 at 11:43

score 2 · Answer 3 · answered Jun 04 '12 at 17:15

2

Since there appear to be unknown non-printable characters in the "blank" lines, you could re-write your sed command to only display lines with printable characters in them:

sed -n '/[!-~]/p'

answered Jun 04 '12 at 17:15

Jeremy Stein

19,171
16
68
83

score 0 · Answer 4 · answered May 25 '12 at 14:01

0

You can also try:

 sed -n '/^./p'

which prints only lines which have at least one char at the beginning of the line. (BTW sed '/^$/d' works for me)

answered May 25 '12 at 14:01

Grzegorz Grzybek

6,152
3
29
42

1. You can also use `grep -v '^$'` 2. How do you use `sed` command? With UNIX pipes? (`cat file | sed ...`)? Maybe you're piping standard error output instead of standard output? – Grzegorz Grzybek May 25 '12 at 17:21
Grep gave me the same result. I "sed 'blahblah' file > newfile". – AWE May 25 '12 at 22:43

Can't remove empty lines with sed regex

4 Answers4

Linked