awk '$2 == "NA" ... error

UnixKiwi · 02-03-2008, 06:52 PM

Hi I am trying to extract a subset of a dataset where the selection criteria is that the second column equals NA (all other options are numerical, NA is the only option containg letters).

The following script has worked for me previously on a much older unix machine but refuses to work for me now

awk '$2=="NA" {print $0}' newdat > temp

It seems that it won't recognise NA as it will work when I substitute one of the numbers in the place of NA (but of course it extracts the wrong data). I also used to be able to achieve similiar the same thing by using

awk '$2 > 99999999 {print $0}' newdat > temp

where 99999999 exceeded the highest numerical value in column 2. Unfortunately this does not work anymore either.

Can anyone see what I am doing wrong?

ghostdog74 · 02-03-2008, 07:43 PM

you might want to try $2 ~ /NA/, if not show your sample data file

UnixKiwi · 02-03-2008, 09:05 PM

Nup, still doesn't work

Sample Data

30448 992 30432 2 275 275 275 21 275 1994 1989 1993
10276 9929 9347 2 275 89 275 275 2004 2002 2000
10275 NA 9610 1 275 275 89 275 275 2004 2002 2001
37563 99999 36692 2 152 275 152 152 275 2002 2000 2000
37732 NA 36692 2 152 275 152 152 275 2002 2000 2000

gilead · 02-03-2008, 09:09 PM

That's odd, the following works for me:

Code:

$ cat data
30448 992 30432 2 275 275 275 21 275 1994 1989 1993
10276 9929 9347 2 275 89 275 275 2004 2002 2000
10275 NA 9610 1 275 275 89 275 275 2004 2002 2001
37563 99999 36692 2 152 275 152 152 275 2002 2000 2000
37732 NA 36692 2 152 275 152 152 275 2002 2000 2000
$ awk '$2 ~ /NA/ {print $0}' data
10275 NA 9610 1 275 275 89 275 275 2004 2002 2001
37732 NA 36692 2 152 275 152 152 275 2002 2000 2000

What version of awk are you running?

Code:

$ awk --version
GNU Awk 3.1.5
Copyright (C) 1989, 1991-2005 Free Software Foundation.

Tinkster · 02-04-2008, 05:23 PM

Quote:

Originally Posted by UnixKiwi

Nup, still doesn't work

Sample Data

30448 992 30432 2 275 275 275 21 275 1994 1989 1993
10276 9929 9347 2 275 89 275 275 2004 2002 2000
10275 NA 9610 1 275 275 89 275 275 2004 2002 2001
37563 99999 36692 2 152 275 152 152 275 2002 2000 2000
37732 NA 36692 2 152 275 152 152 275 2002 2000 2000

This feature of awk is so basic and next to trivial that
I strongly doubt any implementation would have broken it.

I'd be looking at my input data more closely, and check
whether something sill is going on there, and awk for example
sees everything as ONE long line, or can't see the the fields
because a unicode character for space snuck in that's not
part of your locale ....

Cheers,
Tink