Select every nth line

[rkdash]

Hi,

I have a large ascii (3-column ) file ( > 2 Gb). So I want to resample it to select, for example, every 5th line. How to do this ?

Thanks

 

[futurelet]

[tt]
awk '!(NR%5)' file
[/tt]

 

[olded]

I know this is an awk forum, but I'll bet the sed version is faster on a 2-gig file:

sed -n 'n;n;n;n;p' file

 

[rkdash]

Thanks futurelet and olded.

I tried with:

sed -n '1~5p' infile > outfile

and it worked.

 

[futurelet]

> I know this is an awk forum, but I'll bet the sed version
> is faster on a 2-gig file:

Why?

 

[olded]

>Why?

Because sed has a smaller foot print than awk. Of course, that's just a guess on my part; you'd have to test it.

If this problem was a one time only deal, I wouldn't worry about it.

 

[futurelet]

I'm not sure what you mean by footprint. On my system, sed.exe is 1,060,864 bytes; one version of mawk is 67,072 bytes (almost a 16:1 ratio).

Awk spends some time parsing every line into fields, but that time is probably negligible compare to the time needed to read and write the line.

 

[olded]

Hmmm... That's interesting.

On my solaris 7 system, here are the sizes:

76408 Sep 1 1998 /bin/awk
104656 Feb 15 2002 /bin/nawk
24796 Feb 28 2002 /bin/sed

As I said, it would be an interesting technical exercise to see which is faster. I wish I had the time ....

How about that? Two guys are arguing about who's "tool" is smaller. -) -)

Ed

 

[futurelet]

Thanks for the info. I was wondering whether sed was really so complex that it had to be that huge.

The version I use under DOS is
GNU sed version 4.0.7 (DJGPP port 2003-09-09 (r1))

It must be grossly bloated due to linked-in libraries.