Gawk output difference

Shell Programming and Scripting
1

Why the below option2 doesn't fetch similar output as option1 ? I am on linux.


$cat test
2013-01-01-00.25.43.643845

Option1:
cat test | gawk -F"-" ' {print $2 " " $3 " " $1}'
01 01 2013

Option2:
cat test | gawk '{FS="-"}  {print $2 " " $3 " " $1}'
2013-01-01-00.25.43.643845
2

In the second command the FS is set after reading the first record.

3

You have to change the field separator before you read the record or you have to force awk to re-evaluate the record to make the change in FS take effect. So, either of the following should work:

gawk 'BEGIN{FS="-"}  {print $2 " " $3 " " $1}' test

or
gawk '{FS="-";$1=$1} {print $2 " " $3 " " $1}' < test
Note that there is no need to use cat and a pipeline instead of the simple command letting awk open the file itself or using redirection in the shell to set the input file before kicking off awk.

4

Should be:

$0=$0
5

Thanks, that helps.

Was just wondering , if this thing is specific to GAWK 'coz nawk works fine on Solaris

$ cat test  | nawk '{FS="-"}  {print $2 " " $3 " " $1}' 
03 01 2013
6

Yes, different awk versions implement field splitting differently:

$ echo a:b | /usr/xpg4/bin/awk '{ FS = ":"; print NF }'
1
$ echo a:b | nawk '{ FS = ":"; print NF }'
2

That's why usually the correct way to assign a non-default field separator is in the BEGIN block or directly on the command line.

$ awk -F- '{ print $2, $3, $1 }' infile
01 01 2013
$ nawk -F- '{ print $2, $3, $1 }' infile
01 01 2013
$ /usr/xpg4/bin/awk -F- '{ print $2, $3, $1 }' infile
01 01 2013