#!/usr/bin/perl
use v5.14;
use strict;

use utf8;
binmode STDOUT, ':utf8';

sub wghtrandsel { # selects in a weighted randomized manner
    my $multices = ref $_[-1] eq 'HASH' ? pop : {};  
    my @simple = split "", shift//"";

    while ( my ($num, $mult) = each $multices ) {
        my @values = ref $mult eq 'ARRAY' ? @$mult : $mult;
        push @simple, ($_) x $num for @values;
    }

    return sub {
        my $c = $simple[ int rand @simple ];
        return (ref $c ? $c->() : $c) // "";
    }
}

sub chained { 
    my @args = @_;
    return sub { join "", map { ref $_ ? $_->() : $_ } @args };
}

my $weak_vowel = wghtrandsel("aaaeeeiiooouu");
my $strong_vowel = wghtrandsel("áóéàòèâôêãõẽíúîû");
                                 # PM bug? ^^^^^^^ => e + ~ 
my $weak_diphthong = wghtrandsel({
    3 => chained(
        wghtrandsel("ae"),
        wghtrandsel({ 5 => "i", 1 => "u" })
    ),
    2 => chained( "o", wghtrandsel({ 5 => "u", 1 => "i" }) ),
    1 => [ "uy", "iw" ],
});
my $strong_diphthong = wghtrandsel({
    3 => chained(
        wghtrandsel("áé"),
        wghtrandsel({ 5 => "i", 2 => "u" })
    ),
    1 => chained( "ó", wghtrandsel({ 5 => "u", 3 => "i" }) ),
    1 => [ "úy", "úa", "íw", "ía" ],
});
my $nonplosive = wghtrandsel("cfjlmnrsvxz", { 1 => ['hn','hr'] });
my $plosive = wghtrandsel("bdgkpqt");
my $initial_consonant = wghtrandsel({
    5 => $nonplosive,
    2 => wghtrandsel({ 1 => [qw[ch fh jh sh vh zh h]] }),
    3 => chained(
        $plosive,
        wghtrandsel({ 6 => undef, 1 => ["w", "y", "l", "r"] }),
    ),
    1 => ["y","w"]
});

my $stressed_syllable = chained(
    $initial_consonant,
    wghtrandsel({ 2 => $strong_vowel, 1 => $strong_diphthong }),
    wghtrandsel({ 3 => $nonplosive, 5 => undef }),
);

my $stressed_initial_syllable = wghtrandsel({
    3 => $stressed_syllable,
    1 => chained(
        wghtrandsel({ 3 => $strong_vowel, 2 => $strong_diphthong }),
        wghtrandsel({ 4 => undef, 1 => $nonplosive })
    ),
}); 

my $unstressed_syllable = chained(
    $initial_consonant,
    wghtrandsel({
        3 => chained(
            $weak_vowel,
            wghtrandsel({ 8 => undef, 1 => $nonplosive })
        ),
        2 => $weak_diphthong,
    }),
);      

my $word = wghtrandsel({
    1 => [ $stressed_initial_syllable, $unstressed_syllable ],
    4 => chained(
        wghtrandsel({
            5 => $stressed_initial_syllable,
            2 => chained($unstressed_syllable, $stressed_syllable),
        }),
        $unstressed_syllable
    ),
});

say $word->() for 1 .. 1000;

1;