#============================================================
# MyCharConvEUCJP
#============================================================
package MyCharConvEUCJP;
use MyCharConv;
@ISA = qw(MyCharConv);

use strict;
use Encode;

use Jcode;
#use Text::Kakasi;

#===============================================
# ¥¹¥¯¥ê¥×¥ÈÂç°èÊÑ¿ô
#===============================================
my $ScriptCodeCharCode = Jcode::getcode("·î²Ð¿åÌÚ¶âÅÚÆü¡¡");

my $UseKakasiModule = 0;
my $KakasiPath      = "/usr/local/bin/kakasi";
my $KakasiCharCode  = "euc";
if($^O =~ /^MSWin/) {
	$KakasiPath = "c:\\kakasi\\bin\\kakasi.exe";
	$KakasiCharCode  = "sjis";
}

my %ConvTbl = (
	'a'   => '¤¢',
	'i'   => '¤¤',
	'u'   => '¤¦',
	'e'   => '¤¨',
	'o'   => '¤ª',
	'ka'  => '¤«',
	'ki'  => '¤­',
	'ku'  => '¤¯',
	'ke'  => '¤±',
	'ko'  => '¤³',
	'sa'  => '¤µ',
	'shi' => '¤·',
	'si'  => '¤·',
	'su'  => '¤¹',
	'se'  => '¤»',
	'so'  => '¤½',
	'ta'  => '¤¿',
	'chi' => '¤Á',
	'ti'  => '¤Á',
	'tu'  => '¤Ä',
	'te'  => '¤Æ',
	'to'  => '¤È',
	'na'  => '¤Ê',
	'ni'  => '¤Ë',
	'nu'  => '¤Ì',
	'ne'  => '¤Í',
	'no'  => '¤Î',
	'ha'  => '¤Ï',
	'hi'  => '¤Ò',
	'hu'  => '¤Õ',
	'he'  => '¤Ø',
	'ho'  => '¤Û',
	'ma'  => '¤Þ',
	'mi'  => '¤ß',
	'mu'  => '¤à',
	'me'  => '¤á',
	'mo'  => '¤â',
	'ya'  => '¤ä',
	'yi'  => '¤¤',
	'yu'  => '¤æ',
	'ye'  => '¤¨',
	'yo'  => '¤è',
	'ra'  => '¤é',
	'ri'  => '¤ê',
	'ru'  => '¤ë',
	're'  => '¤ì',
	'ro'  => '¤í',
	'wa'  => '¤ï',
	'wi'  => '¤¦¤£',
	'wu'  => '¤¦',
	'we'  => '¤¦¤§',
	'wo'  => '¤ò',
	'n'   => '¤ó',
	'm'   => '¤ó',

	'da'  => '¤À',
	'di'  => '¤Â',
	'du'  => '¤Å',
	'de'  => '¤Ç',
	'do'  => '¤É',

	'ba'  => '¤Ð',
	'bi'  => '¤Ó',
	'bu'  => '¤Ö',
	'be'  => '¤Ù',
	'bo'  => '¤Ü',

	'ca'  => '¤­¤ã',
	'ci'  => '¤Á',
	'cu'  => '¤­¤å',
	'ce'  => '¤»',
	'co'  => '¤³',

	'fa'  => '¤Õ¤¡',
	'fi'  => '¤Õ¤£',
	'fu'  => '¤Õ',
	'fe'  => '¤Õ¤§',
	'fo'  => '¤Õ¤©',

	'ga'  => '¤¬',
	'gi'  => '¤®',
	'gu'  => '¤°',
	'ge'  => '¤²',
	'go'  => '¤´',

	'ja'  => '¤¸¤ã',
	'ji'  => '¤¸',
	'ju'  => '¤¸¤å',
	'je'  => '¤¸¤§',
	'jo'  => '¤¸¤ç',

	'la'  => '¤é',
	'li'  => '¤ê',
	'lu'  => '¤ë',
	'le'  => '¤ì',
	'lo'  => '¤í',

	'pa'  => '¤Ñ',
	'pi'  => '¤Ó',
	'pu'  => '¤Ö',
	'pe'  => '¤Ú',
	'po'  => '¤Ý',

	'qa'  => '¤¯¤¡',
	'qi'  => '¤¯¤£',
	'qu'  => '¤¯',
	'qe'  => '¤¯¤§',
	'qo'  => '¤¯¤©',

	'va'  => '¥ô¤¡',
	'vi'  => '¥ô¤£',
	'vu'  => '¥ô',
	've'  => '¥ô¤§',
	'vo'  => '¥ô¤©',

	'xa'  => '¤·¤ã',
	'xi'  => '¤·¤£',
	'xu'  => '¤¹¤¥',
	'xe'  => '¤·¤§',
	'xo'  => '¤¹¤©',

	'za'  => '¤¶',
	'zi'  => '¤¸',
	'zu'  => '¤º',
	'ze'  => '¤¼',
	'zo'  => '¤¾',

	'cha' => '¤Á¤ã',
	'chi' => '¤Á',
	'chu' => '¤Á¤å',
	'che' => '¤Á¤§',
	'cho' => '¤Á¤ç',

	'tsa' => '¤Ä¤¡',
	'tsi' => '¤Á',
	'tsu' => '¤Ä',
	'tse' => '¤Ä¤§',
	'tso' => '¤Ä¤©',

	'tha' => '¤·¤ã',
	'thi' => '¤·',
	'thu' => '¤Á¤å',
	'the' => '¤»',
	'tho' => '¤½',
	
	'kya' => '¤­¤ã',
	'kyi' => '¤­¤£',
	'kyu' => '¤­¤ã',
	'kye' => '¤­¤§',
	'kyo' => '¤­¤ç',
	'sya' => '¤·¤ã',
	'syi' => '¤·¤£',
	'syu' => '¤·¤å',
	'sye' => '¤·¤§',
	'syo' => '¤·¤ç',
	'sha' => '¤·¤ã',
	'shi' => '¤·',
	'shu' => '¤·¤å',
	'she' => '¤·¤§',
	'sho' => '¤·¤ç',
	'tya' => '¤Á¤ã',
	'tyi' => '¤Á¤£',
	'tyu' => '¤Á¤å',
	'tye' => '¤Á¤§',
	'tyo' => '¤Á¤©',
	'nya' => '¤Ë¤ã',
	'nyi' => '¤Ë¤£',
	'nyu' => '¤Ë¤å',
	'nye' => '¤Ë¤§',
	'nyo' => '¤Ë¤ç',
	'hya' => '¤Ò¤ã',
	'hyi' => '¤Ò¤£',
	'hyu' => '¤Ò¤å',
	'hye' => '¤Ò¤§',
	'hyo' => '¤Ò¤ç',
	'mya' => '¤ß¤ã',
	'myi' => '¤ß¤£',
	'myu' => '¤ß¤£',
	'mye' => '¤ß¤§',
	'myo' => '¤ß¤ç',
	'rya' => '¤ê¤ã',
	'ryi' => '¤ê¤£',
	'ryu' => '¤ê¤å',
	'rye' => '¤ê¤§',
	'ryo' => '¤ê¤ç',
	'wya' => '¤ï¤¡',
	'wyi' => '¤ï¤£',
	'wyu' => '¤ï¤¥',
	'wye' => '¤ï¤§',
	'wyo' => '¤ï¤©',

	'nji' => '¤ó¤¸',
	'nta' => '¤ó¤¿',
	);
foreach my $key (keys %ConvTbl) {
	Jcode::convert(\$ConvTbl{$key}, "utf8", $ScriptCodeCharCode);
}
my $ltu = "¤Ã";
Jcode::convert(\$ltu, "utf8", $ScriptCodeCharCode);
my $oh  = "¤ª¤ª";
Jcode::convert(\$oh,  "utf8", $ScriptCodeCharCode);
my $nn  = "¤ó";
Jcode::convert(\$nn,  "utf8", $ScriptCodeCharCode);

#===============================================
# ¥³¥ó¥¹¥È¥é¥¯¥¿¡¦¥Ç¥¹¥È¥é¥¯¥¿
#===============================================
sub new
{
	my ($module, $buff) = @_;
	my $this = {};
	bless $this;

	$this->{pJcode} = new Jcode;
	$this->{pUTF8}  = find_encoding('utf8');
	$this->{pEUCjp} = find_encoding('eucjp');
	$this->{pSJIS}  = find_encoding('sjis');

	return $this;
}

sub DESTROY
{
	my $this = shift;
	$this->Close();
}

sub Initialize
{
	my ($this) = @_;
}

#===============================================
# ÊÑ¿ô¼èÆÀ´Ø¿ô
#===============================================

#===============================================
# °ìÈÌ¥á¥ó¥Ð´Ø¿ô
#===============================================
sub Hankaku2Zenkaku
{
	my ($this, $s, $sourcecharcode, $ConvertKana) = @_;
	$sourcecharcode = $this->getcode($s) if(!defined $sourcecharcode);
	$ConvertKana = 1 if(!defined $ConvertKana);

#print("c[$ScriptCodeCharCode][$sourcecharcode]<br>\n");
$this->convert(\$s, $ScriptCodeCharCode, $sourcecharcode);
#	$s = Jcode->new($s)->tr('a-zA-Z0-9 @!\"#$%&\'()=-^\\|{}:;+*?_/<>', 
#				'£á-£ú£Á-£Ú£°-£¹¡¡¡÷!¡É#$¡ó¡õ¡Ç()=¡Ý¡°¡ï¡Ã¡Ð¡Ñ¡§¡¨+¡ö?¡²/¡ã¡ä')->h2z() . '';
	$s = Jcode->new($s)->tr('a-zA-Z0-9 ', '£á-£ú£Á-£Ú£°-£¹¡¡');
	$s = $s->tr('@!\"#$%&\'',             '¡÷¡ª¡É¡ô¡ð¡ó¡õ¡Ç');
	$s = $s->tr('()=-^\\|{}:;+*?_/<>',    '()=¡Ý¡°¡ï¡Ã¡Ð¡Ñ¡§¡¨+¡ö¡©¡²¡¿¡ã¡ä');
	if($ConvertKana) {
		$s = $s->h2z() . '';
	}
	else {
		$s = $s . '';
	}
$this->convert(\$s, $sourcecharcode, $ScriptCodeCharCode);
	return $s;
#	return Jcode->new($s)->h2z();
}

sub Zenkaku2Hankaku
{
	my ($this, $s, $sourcecharcode, $ConvertKana) = @_;
	$sourcecharcode = $this->getcode($s) if(!defined $sourcecharcode);
	$ConvertKana = 1 if(!defined $ConvertKana);

$this->convert(\$s, $ScriptCodeCharCode, $sourcecharcode);
#	$s = Jcode->new($s)->tr('£á-£ú£Á-£Ú£°-£¹¡¡¡÷¡ª¡É¡ô¡ð¡ó¡õ¡Ç¡Ê¡Ë¡á¡Ý¡°¡ï¡Ã¡Ð¡Ñ¡§¡¨¡Ü¡ö¡©¡²¡¿¡ã¡ä', 
#				'a-zA-Z0-9 @!"#$%&\'()=-^\\|{}:;+*?_/<>')->z2h() . '';
	$s = Jcode->new($s)->tr('£á-£ú£Á-£Ú£°-£¹¡¡',    'a-zA-Z0-9 ');
	$s = $s->tr('¡÷¡ª¡É¡ô¡ð¡ó¡õ¡Ç',                 '@!\"#$%&\'');
	$s = $s->tr('()=¡Ý¡°¡ï¡Ã¡Ð¡Ñ¡§¡¨+¡ö¡©¡²¡¿¡ã¡ä', '()=-^\\|{}:;+*?_/<>');
	if($ConvertKana) {
		$s = $s->z2h() . '';
	}
	else {
		$s = $s . '';
	}
	$this->convert(\$s, $sourcecharcode, $ScriptCodeCharCode);
	return $s;
#	return Jcode->new($s)->z2h();
}

sub Zenkaku2HankakuKana
{
	my ($this, $s, $sourcecharcode) = @_;
	$sourcecharcode = $this->getcode($s) if(!defined $sourcecharcode);

	$s = Jcode->new($s)->z2h() . '';
	$this->convert(\$s, $sourcecharcode);
	return $s;
}

sub HankakuKana2Zenkaku
{
	my ($this, $s, $sourcecharcode) = @_;
	$sourcecharcode = $this->getcode($s) if(!defined $sourcecharcode);

$this->convert(\$s, $ScriptCodeCharCode, $sourcecharcode);
	$s = Jcode->new($s)->h2z() . '';
	$this->convert(\$s, $sourcecharcode, $ScriptCodeCharCode);
	return $s;
}

sub Hiragana2Katakana
{
	my ($this, $s, $sourcecharcode) = @_;
	$sourcecharcode = $this->getcode($s) if(!defined $sourcecharcode);

	$s = Jcode->new($s)->tr('¤¡-¤ó', '¥¡-¥ó') . '';
	$this->convert(\$s, $sourcecharcode, $ScriptCodeCharCode);
	return $s;
}

sub Katakana2Hiragana
{
	my ($this, $s, $sourcecharcode) = @_;
	$sourcecharcode = $this->getcode($s) if(!defined $sourcecharcode);

$this->convert(\$s, $ScriptCodeCharCode, $sourcecharcode);
	$s = $this->HankakuKana2Zenkaku($s, $sourcecharcode);
	$s = Jcode->new($s)->tr('¥¢-¥ó', '¤¢-¤ó') . '';
	$this->convert(\$s, $sourcecharcode);
	return $s;
}

#my $ZenkakuKigou    = "[¡ª-¡Á¡¡]";
#my $ZenkakuSpace    = "([!-~])¡¡([!-~])";
#my $ZenkakuAlphabet = "[£Á-£Ú£á-£ú£°-£¹¡¡¡Ý¡¾¡¤¡¢¡£¡¥]";
#Jcode::convert(\$ZenkakuKigou,    'utf8', $ScriptCodeCharCode);
#Jcode::convert(\$ZenkakuSpace,    'utf8', $ScriptCodeCharCode);
#Jcode::convert(\$ZenkakuAlphabet, 'utf8', $ScriptCodeCharCode);
sub Zen2Han
{
	my ($this, $s, $sourcecharcode) = @_;
	$sourcecharcode = $this->getcode($s) if(!defined $sourcecharcode);

	$s = $this->Zenkaku2Hankaku($s, $sourcecharcode) . '';
#	$this->convert(\$s, $sourcecharcode);
	return $s;
}

sub Roma2Kana {
	my ($this, $s, $charcode, $sourcecharcode) = @_;
	$charcode = 'sjis'                   if(!defined $charcode);
	$sourcecharcode = $this->getcode($s) if(!defined $sourcecharcode);

	$s = lc $this->Zenkaku2Hankaku($s, $sourcecharcode);

$this->convert(\$s, "utf8", $sourcecharcode);
	$s =~ s/([aiueo])h([bcdfghjklmnpqrstvwxyz])/$1$1$2/g;
	$s =~ s/[nm]([^aiueon])/$nn$1/g;
	$s =~ s/nn/$nn/g;
	$s =~ s/n-([aiueo])/$nn$1/g;

	my ($others, $hit, $rest);
	my $ret = '';
	while(1) {
#print "s=[$s]\n";
		my ($others, $hit, $rest) = ($s =~ /^(.*?)([bcdfghjklmnpqrstvwxyz]*[aiueon])(.*)$/sg);
		if(!defined $hit) {
$this->convert(\$s, $charcode, "utf8");
			$ret .= $s;
			return $ret;
		}

		my $pre = '';
		if($hit =~ /^([bcdfghjklmnpqrstvwxyz])([bcdfghjklmnpqrstvwxyz])(.*)$/) {
			if($1 eq $2) {
#				$pre = $ltu;
#				$hit = $2 . $3;
			}
		}
		if($hit =~ /^(oh)(.*)$/) {
#			$pre .= $oh;
#			$hit = $2;
		}
		$hit = $ConvTbl{$hit} if(defined $ConvTbl{$hit});
$this->convert(\$others, $charcode, "utf8");
$this->convert(\$hit,    $charcode, "utf8");
$this->convert(\$pre,    $charcode, "utf8");
		$ret .= $others . $pre .$hit;
		$s = $rest;
	}
}

sub Conv2Hiragana
{
	my ($this, $s, $sourcecharcode, $targetcharcode) = @_;
$sourcecharcode = Jcode::getcode($s);
$this->convert(\$s, $ScriptCodeCharCode, $sourcecharcode);
	if($this->match($s, "^[¤¡-¤ó¥¡-¥óŽ§-ŽÝ]*\$")) {
$this->convert(\$s, $sourcecharcode, $ScriptCodeCharCode);
		return $this->Katakana2Hiragana($s, $sourcecharcode, $targetcharcode);
	}
$this->convert(\$s, $sourcecharcode, $ScriptCodeCharCode);
	return $this->Kakasi($s, $sourcecharcode, $targetcharcode, qw/-aH -jH -KH -JH -EH -kH/);
#	return $this->Kakasi($s, $sourcecharcode, $targetcharcode, qw/-ieuc -oeuc -aH -jH -KH -JH -EH -kH/);
}

sub Conv2Roma
{
	my ($this, $s, $sourcecharcode, $targetcharcode) = @_;
	return $this->Kakasi($s, $sourcecharcode, $targetcharcode, qw/-Ha -ja -Ka -Ja -Ea -ka/);
#	return $this->Kakasi($s, $sourcecharcode, $targetcharcode, qw/-ieuc -oeuc -Ha -ja -Ka -Ja -Ea -ka/);
}

sub Kakasi {
	my ($this, $s, $sourcecharcode, $targetcharcode, @options) = @_;
	$sourcecharcode = $this->getcode($sourcecharcode) if(!defined $sourcecharcode);
	$targetcharcode = $sourcecharcode if(!defined $targetcharcode);
	return $s if($s eq '');

#print "s1: [$s]<br>\n";
$s =~ s/([^\\])([\(\)\{\}\[\]\>\<\*\?\|\!\$\%\&\~\=\-\'\"\`\/\#])/$1\\$2/g;
$this->convert(\$s, $KakasiCharCode, $sourcecharcode);
#print "s2: [$s]<br>\n";
	if($UseKakasiModule) {
		my $kakasi = Text::Kakasi->new(@options);
		$s  = $kakasi->get($s);
	}
	else {
		my $cmd = "$KakasiPath " . join(' ', @options);
#print "cmd: [$cmd]\n";
#$s =~ s/\(/\\(/g;
#$s =~ s/\)/\\(/g;
#print "[$s]<br>\n";
		$s = `echo $s | $cmd`;
		$s =~ s/[\r\n]+$//s;
	}
$this->convert(\$s, $targetcharcode, $KakasiCharCode);

	return $s;
}

1;