AWK - square brackets problem

Author

square brackets problem

Dave.Kritzberg@gmail.com

2006-05-16, 3:56 am

Hello regular expression experts,

I am trying to clean up a data file, using a combination of sed and
awk in a bash shell script. I'm a beginner at both, and I'm stuck
on a particular expression.

In a few records in the data, a field has the literal value [NULL] --
which I would like to replace with 0 (zero). However, [ and ]
comprise a regular expression, and \[ \] are not accepted as an escape
in either awk or sed, apparently, based on warnings I'm receiving at
the command line. I have tried it this way in sed:

s/[NULL]/0/g

.... and this way in awk:

BEGIN {FS=OFS="\t"}
{
{ gsub("[NULL]","0",$22)
print }
}
END {}

or:

BEGIN {FS=OFS="\t"}
{
{ gsub("\[NULL\]","0",$22)
print }
}
END {}

None of these work. Can anyone offer me a way to fix this?

Thanks in advance,
Dave

--
Dave Kritzberg
University of Colorado Economics
dijon.colorado.edu

Kenny McCormack

2006-05-16, 3:56 am

In article <1147752158.723592.243110@j33g2000cwa.googlegroups.com>,
<Dave.Kritzberg@gmail.com> wrote:
>Hello regular expression experts,
>
>I am trying to clean up a data file, using a combination of sed and
>awk in a bash shell script. I'm a beginner at both, and I'm stuck
>on a particular expression.

Lose the "combination of sed and awk". You never need sed, once you've
taken the step forward to awk.

>In a few records in the data, a field has the literal value [NULL] --
>which I would like to replace with 0 (zero). However, [ and ]
>comprise a regular expression, and \[ \] are not accepted as an escape
>in either awk or sed, apparently, based on warnings I'm receiving at
>the command line.
>
>BEGIN {FS=OFS="\t"}
>{

Lose this.

> { gsub("\[NULL\]","0",$22)
{ gsub(/\[NULL\]/,0,$22)
> print }
>}

Lose this.

>END {}

Lose this.

>None of these work. Can anyone offer me a way to fix this?

Basically, you should be using /'s to enclose REs, not "s.
(I will leave it others to dot the Is and cross the Ts as to why this is so)

Alternatively, you could try [[] and []].

Robert Katz

2006-05-16, 3:57 am

Dave.Kritzberg@gmail.com wrote:
> Hello regular expression experts,
>
> I am trying to clean up a data file, using a combination of sed and
> awk in a bash shell script. I'm a beginner at both, and I'm stuck
> on a particular expression.
>
> In a few records in the data, a field has the literal value [NULL] --
> which I would like to replace with 0 (zero). However, [ and ]
> comprise a regular expression, and \[ \] are not accepted as an escape
> in either awk or sed, apparently, based on warnings I'm receiving at
> the command line. I have tried it this way in sed:
>
> s/[NULL]/0/g

sed 's/\[NULL]/0/'

>
> ... and this way in awk:
>
> BEGIN {FS=OFS="\t"}
> {
> { gsub("[NULL]","0",$22)
> print }
> }
> END {}
>
> or:
>
> BEGIN {FS=OFS="\t"}
> {
> { gsub("\[NULL\]","0",$22)
> print }
> }
> END {}
>
> None of these work. Can anyone offer me a way to fix this?
>

awk 'sub(/\[NULL]/, "0")1'

--
Regards,

---Robert

Sponsored Links