I think this regular expression pattern is probably more suitable.

    (\$\d{1,3}(?:,\d\d\d)*\.\d\d\b)

use strict;
use warnings;

use English qw( -no_match_vars );

...

while (my $record = <$input_fh>) {
    my @prices = $record =~ m/(\$\d{1,3}(?:,\d\d\d)*\.\d\d\b)/g;

    $number_of_prices = scalar @prices;

    $number_of_prices == 3
        or warn "Incorrect number of prices ($number_of_prices) at"
              . " line $INPUT_LINE_NUMBER of file $input_file\n";

    ...
}

exit 0;
[download]

But you should probably use Text::CSV_XS instead of a regular expression pattern to parse the semicolon-separated-value records properly and to extract the prices from them that way.

Jim