How to set bash next bash operation input equal to the output of the prvious operation?

Question

I believe it is really easy, but I didn't find anywhere an answer for me. What I'm trying to do is:

echo file.war | sed s/.war// | rm -rf ???

to pass to the rm -rf the output of the sed command.

Not sure if it is the right way to get this...

Thanks for your help in advance!

It would help if you described exactly what you're trying to do, since it's easy to interpret wrongly from code that doesn't do what you want it to. — Wug, Jun 20 '13 at 14:56

score 3 · Answer 1 · answered Jun 20 '13 at 14:58

3

This is what 'xargs' does.

echo file.war | sed s/\.war$// | xargs rm -rf

And note my changes to your regex. It needs to be anchored and the '.' needs to be escaped.

answered Jun 20 '13 at 14:58

ccarton

3,556
16
17

Yeah, thanks for this. Work as a charm! – mr.nothing Jun 20 '13 at 15:00
2

`\.` is changed by bash to `.`, which sed then treats as a “magic” character. To pass an escaped `.` to sed, you need to escape the backslash: `sed s/\\.war$//`. – Eric Postpischil Jun 20 '13 at 16:02

gniourf_gniourf · Accepted Answer · 2013-06-20T16:02:19.590

Other possibilities, in bash:

Use command substitution:

rm -rf $(echo file.war | sed 's/\.war$//')

Use a variable and shell parameter expansion:
```
file=file.war
rm -rf "${file%.war}"
```
Instead of the echo | sed, use a here-string (and the command substitution):
```
rm -rf $(sed 's/\.war$//' <<< "file.war")
```

This was in fact just to show you the command substitution $(...) thing (avoiding xargs that will fail miserably if you have file names containing funny symbols like spaces). Also to show you that in your case, sed is useless (thanks to the shell parameter expansion) and, if you really need sed, that the echo | sed thing can be avoided in bash.

I don't know what exactly you're trying to achieve. I could imagine:

you're trying to delete all files file such that file.war exist in current directory. In this case, I would do:
```
for file in *.war; do
    rm -rf -- "${file%.war}"
done
```

You have a file called filenames that contains lines like:

file1.war
file2.war
...
filen.war

and you want to delete all files file1, file2, …, filen. Then I would do:

while read -r file; do
    [[ ( $file = *.war ) && ( -f $file ) ]] || continue
    rm -rf -- "${file%.war}"
done < filenames

score 1 · Answer 3 · answered Jun 20 '13 at 15:59

1

An alternative approach is

rm -rf `echo file.war | sed s/\.war$//`

answered Jun 20 '13 at 15:59

Nicola Musatti

17,834
2
46
55

How to set bash next bash operation input equal to the output of the prvious operation?

3 Answers3