Echo every other line

yitzle · 11-07-2007, 11:24 PM

I've got some text that I'm operating on through a pipe, eg
cat - | grep "someString" | wc -l
On thing I'd like to do is weed out every other line.
Is there some way to do this?
I was thinking of somehow using sed and changing the separator to "", the telling it to run s,([^\n]\n)[^\n]\n,\1,

Would that work? How do I change the separator?

Disillusionist · 11-08-2007, 01:47 AM

Not sure why you would want to just show every other line, however have you thought of looping:

Code:

cat - |grep "something"|while read l_line
do
   # Perform a test of some sort before echoing!
   echo $l_line
done

Suspecting homework therefore this is not a complete solution.

If this is not homework please explain why you want to show every other line (what this is for in a real world example)

Tinkster · 11-08-2007, 02:37 AM

Quote:

Originally Posted by yitzle

I've got some text that I'm operating on through a pipe, eg
cat - | grep "someString" | wc -l
On thing I'd like to do is weed out every other line.
Is there some way to do this?
I was thinking of somehow using sed and changing the separator to "", the telling it to run s,([^\n]\n)[^\n]\n,\1,

Would that work? How do I change the separator?

Way too complicated...

man sed
/step

Cheers,
Tink

Disillusionist · 11-08-2007, 02:51 AM

Cheers Tink

I didn't know sed could do that so easily!

You learn something everyday.

yitzle · 11-08-2007, 12:52 PM

Quote:

Originally Posted by Disillusionist

Not sure why you would want to just show every other line, however have you thought of looping:

Code:

cat - |grep "something"|while read l_line
do
   # Perform a test of some sort before echoing!
   echo $l_line
done

Suspecting homework therefore this is not a complete solution.

If this is not homework please explain why you want to show every other line (what this is for in a real world example)

Its not homework. I got output from a turn based game/program that shows the result of me vs opponent. Every other line is me, round next.
I want to count the occurances of a specific result of the me lines.
To flesh out your example, you could use a binary toggle to print every other time the loop runs.
Though if there was a course that taught this sort of stuff, I'd love to take it...

Quote:

Originally Posted by Tinkster

Way too complicated...

man sed
/step

Cheers,
Tink

Neat! Thanks!
I didn't even know you could search man (or less, I guess) like that!

One of these days, I'll have to learn sed/awk in detail.

Tinkster · 11-08-2007, 01:00 PM

Quote:

Originally Posted by yitzle

Neat! Thanks!
I didn't even know you could search man (or less, I guess) like that!

One of these days, I'll have to learn sed/awk in detail.

You're most welcome, and "Yes, you ought to" <BEG>

power at your finger tips ;}

Cheers,
Tink

yitzle · 11-08-2007, 01:04 PM

Quote:

Originally Posted by Tinkster

You're most welcome, and "Yes, you ought to" <BEG>

power at your finger tips ;}

Cheers,
Tink

Once on this topic, what about awk?
Does awk replace sed? Or do they have rather seperate uses?
Should I bother learn both? Which is better to learn?

dgar · 11-08-2007, 01:11 PM

I was going to resort to an iterator in awk, and then testing even/odd with modulus 2.

Those ancients of UNIX sure thought ahead.

Tinkster · 11-08-2007, 01:33 PM

Quote:

Originally Posted by yitzle

Once on this topic, what about awk?
Does awk replace sed? Or do they have rather seperate uses?
Should I bother learn both? Which is better to learn?

Always depends on the complexity of the task at hand.

Think of awk as a missing link between sed and perl :}

As dgar said ... you could achieve the same result with
awk doing something like (mind you, I don't use an iterator).

awk '{if( NR%2 ){print}}'

The big difference is that awk has a concept of fields
which wasn't required in this instance. Plus it's quite
easy to use multi-line records in awk which can be very
tedious to deal with using sed.

For simpler tasks use the simpler tool (oversimplification).
It has a (marginally) smaller file-size and memory footprint,
so will commonly be somewhat quicker ...

So I'd suggest you learn them both, really.

Cheers,
Tink

yitzle · 11-08-2007, 02:15 PM

Thanks for the post! Most informative.

PTrenholme · 11-08-2007, 03:06 PM

Speakin of awk, your problem is easily solved using it:

Code:

$ for ((i=1;i<101;++i)); do echo $i >>test.txt;done
$ cat test.txt | head
1
2
3
4
5
6
7
8
9
10
$ gawk 'NR%2==0 {print}' test.txt | head
2
4
6
8
10
12
14
16
18
20

Tinkster · 11-08-2007, 03:24 PM

Quote:

Originally Posted by PTrenholme

Speakin of awk, your problem is easily solved using it:

Code:

$ for ((i=1;i<101;++i)); do echo $i >>test.txt;done
$ gawk 'NR%2==0 {print}' test.txt | head

Heh ... slightly more concise than my version above.
But we don't really need the "==0" or the print, either.

Code:

awk 'NR%2 ' file

will do ... it's just not quite as readable if you don't know awk
very well :}

Cheers,
Tink

syg00 · 11-08-2007, 04:16 PM

Quote:

Originally Posted by yitzle

I didn't even know you could search man (or less, I guess) like that!

While you're in the mood, add an "-i" subcommand first. I have it as default for things like sed and less.

Tinkster · 11-08-2007, 04:20 PM

Quote:

Originally Posted by syg00

While you're in the mood, add an "-i" subcommand first. I have it as default for things like sed and less.

/me coughs ...

But that switch has a completely different meaning in those
two tools?

less -> case insensitive search
sed -> in-place editing of the file

Cheers,
Tink

syg00 · 11-08-2007, 04:25 PM

Sorry Tink; I haven't got to having my first coffee of the morning ...

Let's try that again - "grep and less".

wanders off to kick the ball for the mutt ... leave the computer alone ...