Find string and replace line in linux

Question

I'd like to do the following, but I can't seem to find an elegant way to do it.

I have a text file that looks like this:

..more values
stuff = 23   
volume = -15
autostart = 12
more values..

Now the "volume" value may change, depending on the circumstance. I need a script that can find that line of text with "volume = xxx" then replace it with the line "volume = 0". What is an elegant solution to this? This line in the text is not guaranteed to be the first line, so I need a way to find it first.

by the time I wrote my answer and tested in a shell, three others were out. I'm getting rusty! — Zlatko, Sep 30 '12 at 17:51

score 13 · Accepted Answer · answered Sep 30 '12 at 17:49

13

sed 's/^volume =.*/volume = 0/g' file.txt

answered Sep 30 '12 at 17:49

Lev Levitsky

63,701
20
147
175

Thanks for the answer, I need to pipe this back into the file correct? – jliu83 Sep 30 '12 at 18:00
1

@jliu83 Don't pipe it to the file directly. You can pipe it to a new file, then you can replace the original file with the new one using `mv`. – Lev Levitsky Sep 30 '12 at 18:03
2

Thanks, and why, might I ask, should I not pipe to the same file? I'm guessing some sort of a race condition concerning reading and writing to the same file? – jliu83 Sep 30 '12 at 18:05
1

@jliu83 Something like that. Also note that `sed` has `-i` option for modifying files "in place", but it actually creates a temporary file, too. For safety, I'd start with the command I show, and if you're happy with the output, you can add the `-i` option. – Lev Levitsky Sep 30 '12 at 18:07
Also, can you explain the `.*` usage? I understand here it means anything that starts with volume, but why not just `*`, why is there a period (ie, why `.*` not `*`)? – jliu83 Sep 30 '12 at 18:21
`*` is a quantifier. It means "match the previous expression 0 or more times". In this case the previous expression will be `.`, which is any character. In `volume =*`, it'd mean "zero or more `=` signs". Feel the difference :) See a short reference on [`sed regex`](http://www.gnu.org/software/sed/manual/html_node/Regular-Expressions.html). – Lev Levitsky Sep 30 '12 at 18:36

score 2 · Answer 2 · answered Sep 30 '12 at 17:49

2

With sed you can say:

sed '/^volume/s/.*/volume = 0/' infile

answered Sep 30 '12 at 17:49

Thor

45,082
11
119
130

can you explain this? usually the "s" option is used before delimiter for substitution. – jliu83 Sep 30 '12 at 18:00
The first `/regex/` decides which line the following commands are applied to, in this case any line starting with `volume`. – Thor Sep 30 '12 at 18:11

score 1 · Answer 3 · answered Sep 30 '12 at 17:50

1

Pipe the contents into this command:

sed -e 's/^volume\s*=\s*-\?[0-9]\+$/volume = 0/'

answered Sep 30 '12 at 17:50

viraptor

33,322
10
107
191

score 1 · Answer 4 · edited Oct 01 '12 at 05:05

1

sed 's/^volume=$.*$$/volume=1/g' inputfile

edited Oct 01 '12 at 05:05

Dan

10,531
2
36
55

answered Oct 01 '12 at 04:59

vara

816
3
12
29

score 0 · Answer 5 · answered Sep 30 '12 at 18:50

0

@jliu83, do some reading on regular expressions. It'll make the pattern matching in the sed commands more understandable.

http://en.wikipedia.org/wiki/Regular_expression

answered Sep 30 '12 at 18:50

itsbruce

4,825
26
35

this is not an answer to the question – Guscie Jan 26 '14 at 17:45

Find string and replace line in linux

5 Answers5

Linked