This passes use strict and -w.

sub readPackage
{
    my $filename = shift;

    open FILE, $filename || die "cannot open $filename\n";
    local $/ = undef;
    my $file = <FILE>;
    close(FILE);
    
    $file ||= '';

    while(
        $file =~
        m/^[^\n]*(FUNCTION|PROCEDURE)(.*?)IS/msgi
    )
    {
        my ($type, $declaration) = ($1, $2);
        print "$1 $2;\n\n";
    }
}
[download]

The other option is

    my @procedures = 
        ( $file =~
        m/^[^\n]*(FUNCTION|PROCEDURE)(.*?)IS/msgi )
[download]

This will create an array of procedure type, declaration pairs, as a flat list. So grap them 2 at a time, and process.

The key to the regex is that using /s allows '.' to match across newlines, and /m allows the '^' to match the beginning of the line. I use the character class [^\n] to act like the '.' but without /s's special abilities.

This was based on reading the regex sections in the cookbook. This version does not handle comments. The one the I use at work is actually for 'use lib' lines and does handle comments.