You are making something easy into something hard. First, as you read the input file, output changes to a temporary file. Then save a copy of the original file (rename it to something else) and replace it with the new temporary file. This little "dance" with renaming things is done so that at any time if the program should crash (or maybe system reboots!), it is always possible to get the original data back and re-run the program. As a general rule, do not write modified contents directly back into the original file because there will be a window when no data at all exists! If the data set is something important, that might be a very, very bad thing!

There is no need for a substitute regex here. Nor is there a need for a "/g" global match. Just add a # at the front if the line contains at least one boundary separated token of "test".

Update:I modified the code with another thought...when writing filters like this, it is usually a good idea to have the property that running the filter more than once is "harmless". In this case, I prevent multiple # from showing up at the front of the line by just a little addition to the "if" clause.

#!/usr/bin/perl -w
use strict;

my $file_name = "test.txt";
open(my $in,  '<', $file_name) || die "cannot open $file_name for read
+ $!";
open(my $out, '>', "$file_name.tmp") || die "cannot open $file_name.tm
+p for write $!";

while( <$in>) 
{
   print $out "#" if(/\btest\b/i && !/^#/); #only one # at front of li
+ne
   print $out $_;
}
close ($in);
close ($out);

# save copy of original file
rename ($file_name,"$file_name.bak")  || die "problem with rename $!";
# replace original with the modified version
rename ("$file_name.tmp", $file_name) || die "problem with rename $!";

= file test.txt before running program
testing is requrired 
see/for/test 
test/data/istesting 

= file test.txt after running program
testing is requrired 
#see/for/test 
#test/data/istesting
[download]