Question

What I love about awk is you can fetch all the lines from a file that satisfies the condition on some aritrary field you specify.
For example,

awk '$3~/hi/' < test.txt # print all lines where the third field matches the pattern "hi"

or

awk '$2>=2' < test.txt  # print all lines where the second field is greater or equal to 2

As a beginner who’s learning about the power of unix, I am absolutely fascinated about this.
Now I am wondering if there is an easy way to perform regex substitutions only on some arbitrary fields you specify? For example, I want to do regex substitution on the third field only.
my current method is to “cut” the field I want and perform substitution on that using perl or sed, which then I “paste” to the original file. But I am wondering if there is more efficient way to achieve this.

Thanks

Answer 1

Since you tagged this question with ‘perl’ (in addition to ‘sed’, ‘awk’, ‘unix’, and ‘command-line’), I’ll assume you’re interested in answers that incorporate any of the above tools.

Perl has an auto-split command-line switch (-a):

perl -lane 'print if $F[2] =~ /some pattern/' filename

…or…

perl -lane 'print if $F[1] >= 42' filename

-a causes an auto-split into the @F array. -n causes Perl to iterate over the lines of the file you feed it. The rest is programming. 😉

Now for substitution:

perl -i.bak -lane '$F[2] =~ s/match/subst/; print join q/ /, @F' filename

Or, a little shorter using the -p switch, which tells Perl to print each line as it appears in $_. That means if you alter @F, you’ll have to copy it back into $_:

perl -i.bak -pale '$F[2] =~ s/match/subst/ && $_="@F"' filename