File Coverage

blib/lib/Lingua/LO/NLP/Romanize/IPA.pm

Criterion	Covered	Total	%
statement	41	41	100.0
branch	4	4	100.0
condition			n/a
subroutine	14	14	100.0
pod	2	2	100.0
total	61	61	100.0

package Lingua::LO::NLP::Romanize::IPA;
use strict;
use warnings;
use 5.012000;
use utf8;
use feature qw/ unicode_strings say /;
use charnames qw/ :full lao /;
use version 0.77; our $VERSION = version->declare('v1.0.1');
use Carp;
use Lingua::LO::NLP::Analyze;
use parent 'Lingua::LO::NLP::Romanize::PCGN';

=encoding utf8

=head1 NAME

Lingua::LO::NLP::Romanize::IPA - Convert Lao syllables to the International Phonetic Alphabet

=head1 FUNCTION

This class is not supposed to be used directly. Rather use
L as a factory:

    my $o = Lingua::LO::NLP::Romanize->new(variant => 'IPA');

=cut

my %CONSONANTS = (
   'ກ'  => 'k', # TODO: split /g/+/k/ here?
   'ຂ'  => [qw/ kʰ k /],
   'ຄ'  => [qw/ kʰ k /],
   'ງ'  => 'ŋ',
   'ຈ'  => [qw/ tɕ t /],
   'ສ'  => [qw/ s t /],
   'ຊ'  => [qw/ s t /],
   'ຍ'  => 'ɲ',
   'ດ'  => [qw/ d t /],
   'ຕ'  => 't',
   'ຖ'  => [qw/ tʰ t /],
   'ທ'  => [qw/ tʰ t /],
   'ນ'  => 'n',
   'ບ'  => [qw/ b p /],
   'ປ'  => 'p',
   'ຜ'  => 'pʰ',
   'ຝ'  => [qw/ f p /],
   'ພ'  => [qw/ pʰ p /],
   'ຟ'  => [qw/ f p /],
   'ມ'  => 'm',
   'ຢ'  => 'j',
   'ລ'  => 'l',
   "\N{LAO SEMIVOWEL SIGN LO}"  => 'l',
   'ວ'  => [qw/ ʋ w /],
   'ຫ'  => 'h',
   'ອ'  => 'ʔ',
   'ຮ'  => 'h',
   'ຣ'  => 'r',   # TODO l?
   'ໜ'  => 'n',
   'ໝ'  => 'm',
   'ຫຼ'  => 'l',
   'ຫຍ' => 'ɲ',
   'ຫນ' => 'n',
   'ຫມ' => 'm',
   'ຫຣ' => 'r',   # TODO l?
   'ຫລ' => 'l',
   'ຫວ' => 'w', # TODO ʋ?
);

my %VOWELS = (
    ### Monophthongs
    'Xະ'   => 'aʔ',
    'Xັ'    => 'a',
    'Xາ'   => 'aː',
    'Xາວ'  => 'aːo',

    'Xິ'    => 'i',
    'Xີ'    => 'iː',
    'Xິວ'   => 'iu', # TODO correct?
    'Xີວ'   => 'iːu', # TODO correct?

    'Xຶ'    => 'ɯ',
    'Xື'    => 'ɯː',

    'Xຸ'    => 'u',
    'Xູ'    => 'uː',

    'ເXະ'  => 'eʔ',
    'ເXັ'   => 'e',
    'ເX'   => 'eː',

    'ແXະ'  => 'ɛʔ',
    'ແXັ'   => 'ɛ',
    'ແX'   => 'ɛː',
    'ແXວ'  => 'ɛːo',

    'ໂXະ'  => 'oʔ',
    'Xົ'    => 'o',
    'ໂX'   => 'oː',
    'ໂXຍ'  => 'oːi', # TODO correct?
    'Xອຍ'  => 'oːi',

    'ເXາະ' => 'ɔʔ',
    'Xັອ'   => 'ɔ',
    'Xໍ'    => 'ɔː',
    'Xອ'   => 'ɔː',

    'ເXິ'   => 'ɤ',
    'ເXີ'   => 'ɤː',
    'ເXື'   => 'ɯːə', # TODO correct?

    'ເXັຍ'  => 'iə',
    'Xັຽ'   => 'iə',
    'ເXຍ'  => 'iːə',
    'Xຽ'   => 'iːə',
    'Xຽວ'  => 'iːəo', # TODO correct?

    'ເXີຍ'  => 'ɤːi',
    'ເXິຍ'  => 'ɤi',

    'ເXຶອ'  => 'ɯə',
    'ເXືອ'  => 'ɯːə',
    'ເXືອຍ' => 'ɯːəi',

    'Xົວະ'  => 'uəʔ',
    'Xົວ'   => 'uːə',
    'Xວ'   => 'uːə',
    'Xວຍ'  => 'uːəi',

    'ໄX'   => 'aj',
    'ໃX'   => 'aj',
    'Xາຍ'  => 'aːj',
    'Xັຍ'   => 'aj',

    'ເXົາ'  => 'aw',
    'Xຳ'   => 'am', # composed U+0EB3
    'Xໍາ'   => 'am',
);
{
    # Replace "X" in %VOWELS keys with DOTTED CIRCLE. Makes code easier to edit.
    my %v;
    foreach my $v (keys %VOWELS) {
        (my $w = $v) =~ s/X/\N{DOTTED CIRCLE}/;
        $v{$w} = $VOWELS{$v};
    }
    %VOWELS = %v;
}

my %TONE_DIACRITICS = (
    LOW => "\N{COMBINING GRAVE ACCENT}",
    MID => "\N{COMBINING MACRON}",                          MID_STOP => "\N{COMBINING MACRON}",
    HIGH => "\N{COMBINING ACUTE ACCENT}",                   HIGH_STOP => "\N{COMBINING ACUTE ACCENT}",
    RISING => "\N{COMBINING CARON}",
    HIGH_FALLING => "\N{COMBINING CIRCUMFLEX ACCENT}",
    MID_FALLING => "\N{COMBINING CIRCUMFLEX ACCENT BELOW}",
);

=head2 new

You don't call this constructor directly but via L.
It adds the following attribute:

=over 4

=item C: boolean indicating whether to add dicacritics for tone

=back

=cut

sub new {
    my ($class, %args) = @_;
    my $self = bless {}, $class;
    $self->{romanize_vowel} = $args{tone} ? \&_vowel_with_tone : \&_vowel_without_tone;
    return $self;
}

sub _vowel_with_tone {
    my ($lao_vowel, $tone) = @_;
    my $vowel = $VOWELS{ $lao_vowel };
    # Insert tone diacritic after first character
    substr($vowel, 1, 0) = $TONE_DIACRITICS{ $tone };
    return $vowel;
}

sub _vowel_without_tone { return $VOWELS{ $_[0] } }

=head2 romanize_consonant

Overrides L to access
module-local data.

=cut

sub romanize_consonant {
    my (undef, $cons, $position) = @_;
    my $consdata = $CONSONANTS{ $cons };
    return ref $consdata ? $consdata->[$position] : $consdata;
}

1;