#!/usr/bin/perl

=encoding utf8

=head1 NAME

goodpasswd - generate good access codes


=head1 SYNOPSIS

B<goodpasswd>


=head1 DESCRIPTION

B<goodpasswd> generates access codes that:

=over 3

=item Z<>* are hard to guess

=item Z<>* will be displayed unambigously in any (normal) font

=item Z<>* will survive being passed through a bad fax machine

=item Z<>* will survive being passed through handwriting

=item Z<>* will survive unquoted in most scripts

=item Z<>* has characters from the character classes UPPER lower number and sign

=back

=head2 Characters considered too close

These character couples are too similar either in different fonts or
in a bad copy and are thus forbidden: B8 cC g9 6G kK lI l1 oO O0 pP sS uU vV xX zZ Z2 ,. :; `' S5

These characters cause problems in URLs: @/:

These characters cause problems in shell: ! " # $ & ( ) [ ] { } ? | < > \ * =

These characters cause problems in SQL (wildcard): %

These characters are hard to type: ^ ~ ¨ ¤ § ½ æ ø å Æ Ø Å

=head2 Other restrictions

Never use the same chars twice in a row: e.g. -- is bad.

Do not start with '-' or '+' as that looks like an (long) option

=head1 EXAMPLE

B<goodpasswd> will give output similar to FJiY7j+7DQ-D.

=cut

#
# US-kbd: ~!@#$%^&*()_+ [] {} ;'\ :"| < > ,./ <>? 
# DK-kbd: §!"#¤%&/()=?` å" Å^ æø' ÆØ* < > ,.- ;:_
# Common:  ! # %                      < > ,.

my $pw;
my @chars=split //, 'abdefhijmnqrtyADEFHJLMNQRTY347+-';
do { 
    $pw = "";
    for (1..12) {
	$pw .= $chars[rand $#chars+1]
    }
} while (($pw =~ /^[+-]/ or $pw =~ /(.)\1/) or
	 not($pw =~ /[A-Z]/ and
	     $pw =~ /[a-z]/ and
	     $pw =~ /[0-9]/ and
	     $pw =~ /[^a-zA-Z0-9]/));

print "$pw\n";