Perl filter with substitution - perl

Question

I am attempting to create a Perl script that filters data presented on STDIN, changing all occurrences of
one string to another and outputting all input lines, changed and unchanged to STDOUT. FROMSTRING and TOSTRING can be PERL-compatible regular expressions. I am unable to get matching output.
Here is an example of what I am trying to achieve.
echo "Today is Saturday" | f.pl 'a' '#'
Output Tod#y is S#turd#y.
echo io | filter.pl '([aeiou])([aeiou])' '$2$1'
Output oi.
#!/usr/bin/perl
use strict;
use warnings;
if (#ARGV != 2){
print STDERR "Usage: ./filter.pl FROMSTRING TOSTRING\n"
}
exit 1;
my $FROM = $ARGV[0];
my $TO = $ARGV[1];
my $inLine = "";
while (<STDIN>){
$inLine = $_;
$inLine =~ s/$FROM/$TO/;
print $inLine
}
exit 0;

Answer 1

First off, the replacement part of a s/.../.../ operation is not a regex; it works like a double-quoted string.
There are a couple of issues with your code.
Your exit 1; statement appears in the middle of the main code, not in the error block. You probably want:
if (#ARGV != 2) {
print STDERR "Usage: ./filter.pl FROMSTRING TOSTRING\n";
exit 1;
}
You're missing a g flag if you want multiple substitutions to happen in the same line:
$inLine =~ s/$FROM/$TO/g;
There's no need to predeclare $inLine; it's only used in one block.
There's also no need to read a line into $_ just to copy it into $inLine.
It's common to use $names_like_this for variables and functions, not $namesLikeThis.
You can use $0 instead of hardcoding the program name in the error message.
exit 0; is redundant at the end.
The following is closer to how I'd write it:
#!/usr/bin/perl
use strict;
use warnings;
if (#ARGV != 2) {
die "Usage: $0 FROMSTRING TOSTRING\n";
}
my ($from, $to) = #ARGV;
while (my $line = readline STDIN) {
$line =~ s/$from/$to/g;
print $line;
}
That said, none of this addresses your second example with '$2$1' as the replacement. The above code won't do what you want because $to is a plain string. Perl won't scan it to look for things like $1 and replace them.
When you write "foo $bar baz" in your code, it means the same thing as 'foo ' . $bar . ' baz', but this only applies to code, i.e. stuff that literally appears in your source code. The contents of $bar aren't re-scanned at runtime to expand e.g. \n or $quux. This also applies to $1 and friends, which are just normal variables.
So how do you get '$2$1' to work?
One way is to mess around with eval, but I don't like it because, well, it's eval: If you're not very careful, it would allow someone to execute arbitrary code by passing the right replacement "string".
Doing it without eval is possible and even easy with e.g. Data::Munge::replace:
#!/usr/bin/perl
use strict;
use warnings;
use Data::Munge qw(replace);
if (#ARGV != 2) {
die "Usage: $0 FROMSTRING TOSTRING\n";
}
my ($from, $to) = #ARGV;
while (my $line = readline STDIN) {
print replace($line, $from, $to, 'g');
}
replace works like JavaScript's String#replace in that it expands special $ sequences.
Doing it by hand is also possible but slightly annoying because you basically have to treat $to as a template and expand all $ sequences by hand (e.g. by using another regex substitution):
# untested
$line =~ s{$from}{
my #start = #-;
my #stop = #+;
(my $r = $to) =~ s{\$([0-9]+|\$)}{
$1 eq '$'
? '$'
: substr($from, $start[$1], $stop[$1] - $start[$1])
}eg;
$r
}eg;
(This does not implement braced groups such as ${1}, ${2}, etc. Those are left as an exercise for the reader.)
This code is sufficiently annoying to write (and look at) that I much prefer using a module like Data::Munge for this sort of thing.

Answer 2

three errors found:
; after error message
exit 1;
$inLine =~ s/$FROM/$TO/g;
like:
#!/usr/bin/perl
use strict;
use warnings;
if (#ARGV != 2){
print STDERR "Usage: ./filter.pl FROMSTRING TOSTRING\n";
exit 1;
}
my $FROM = $ARGV[0];
my $TO = $ARGV[1];
my $inLine = "";
while (<STDIN>){
$inLine = $_;
$inLine =~ s/$FROM/$TO/g;
print $inLine
}
exit 0;