#!/bin/bash

# from to-dict.sh with small modifications
MK_DATA () {
	echo
	sed '/^ /d;
s/$/\
/g;
s/[^]]*\ \ /%h&\
%d/;
s/_[IVX][IVX]* /\
 &/g;
s/ [1-9]\. /\
  &/g;
s/[1-9][0-9]*>/\
      &/g;
s/[ÁÂ×ÇÄÅÖÚÉËÌÍÎÏÐÒÓÔÕÆÞÃÞÛÈÝßØÜÀÑ]>/(&>/g;
s/>>/)/g;
s/\ \_[AISE][a-z]*:/\
  &/g;
s/>/:/g'
}

# from slovnyktodict.awk with small modifications
TO_SLOV () {
awk ' \
function stardict_nbo(string) {
	s1=and(string,255);
	string=rshift(string,8);
	s2=and(string,255);
	string=rshift(string,8);
	s3=and(string,255);
	string=rshift(string,8);
	s4=and(string,255);
	string=rshift(string,8);
	return sprintf("%c", s4) sprintf("%c", s3) sprintf("%c", s2) sprintf("%c", s1);
}
BEGIN{
	ORS="";
	word="";
	dictname="'$1'";
}
{
	if (substr($0,0,2) == "%h") 	{
		gsub(/ +$/,"",$0);
		word=tolower(substr($0,3));
	} else {
		if (substr($0,0,2) == "%d")
			dict[word]=substr($0,3);
		else
			dict[word]=dict[word] "\n" $0;
	}
}
END{
	cnt=asorti(dict,idx);
	if ( cnt == 0) {
		print "Empty dictionary\n";
		exit 1;
	}
	pos=0;
	for (z=1;z<=cnt;++z)
	{
		i=idx[z];
		v=dict[i];
		len=length(v);
		print i sprintf("%c", 0) stardict_nbo(pos) stardict_nbo(len) >dictname".idx";
		print v >dictname".dict";
		pos=pos+len;
	}
	close(dictname".idx");
	com= "stat --printf=\"%s\" " dictname ".idx";
	com| getline size;
	print "StarDict'\''s dict ifo file\nversion=2.4.2\nwordcount="cnt"\nidxfilesize="size"\nbookname="dictname"\nsametypesequence=m\n">dictname".ifo";
}';
}

# from mueller2stardict with small modifications
FROM_MUELLER () {
perl -p -e '
require Unicode::Map8;
my $koi_map = Unicode::Map8->new("koi8-r")    || die;

my $ipa_map = Unicode::Map8->new();

$ipa_map->default_to16( 0x20 );

$ipa_map->addpair( 0x61,  0x61 );
$ipa_map->addpair( 0x62,  0x62 );
$ipa_map->addpair( 0x63,  0x63 );
$ipa_map->addpair( 0x64,  0x64 );
$ipa_map->addpair( 0x65,  0x65 );
$ipa_map->addpair( 0x66,  0x66 );
$ipa_map->addpair( 0x67,  0x67 );
$ipa_map->addpair( 0x68,  0x68 );
$ipa_map->addpair( 0x69,  0x69 );
$ipa_map->addpair( 0x6a,  0x6a );
$ipa_map->addpair( 0x6b,  0x6b );
$ipa_map->addpair( 0x6c,  0x6c );
$ipa_map->addpair( 0x6d,  0x6d );
$ipa_map->addpair( 0x6e,  0x6e );
$ipa_map->addpair( 0x6f,  0x6f );
$ipa_map->addpair( 0x70,  0x70 );
$ipa_map->addpair( 0x71,  0x71 );
$ipa_map->addpair( 0x72,  0x72 );
$ipa_map->addpair( 0x73,  0x73 );
$ipa_map->addpair( 0x74,  0x74 );
$ipa_map->addpair( 0x75,  0x75 );
$ipa_map->addpair( 0x76,  0x76 );
$ipa_map->addpair( 0x77,  0x77 );
$ipa_map->addpair( 0x78,  0x78 );
$ipa_map->addpair( 0x79,  0x79 );
$ipa_map->addpair( 0x7a,  0x7a );

$ipa_map->addpair( 0x2d,  0x2e );
$ipa_map->addpair( 0x2e,  0x2e );


$ipa_map->addpair( 0x54,  0x03b8 );
$ipa_map->addpair( 0x53,  0x0283 );
$ipa_map->addpair( 0x4e,  0x014b );
$ipa_map->addpair( 0xd7,  0x02a7 );
$ipa_map->addpair( 0x44,  0x00F0 );

$ipa_map->addpair( 0x5a,  0x0292 );

$ipa_map->addpair( 0x51,  0x00e6 );
$ipa_map->addpair( 0x49,  0x0131 );
$ipa_map->addpair( 0xc3,  0x028c );
$ipa_map->addpair( 0x55,  0x028a );
$ipa_map->addpair( 0x81,  0x0252 );
$ipa_map->addpair( 0x45,  0x025b );
$ipa_map->addpair( 0xab,  0x0259 );

$ipa_map->addpair( 0x41,  0x0251 );
$ipa_map->addpair( 0xce,  0x025c );
$ipa_map->addpair( 0x8d,  0x0254 );

$ipa_map->addpair( 0xc7,  0x02cc );
$ipa_map->addpair( 0xc8,  0x02c8 );

$ipa_map->addpair( 0xf9,  0x02d0 );
$ipa_map->addpair( 0x3e,  0x02d1 );

sub to_utf {
	return $koi_map->tou($_[0])->utf8;
}

sub ipa_code {
	return $ipa_map->tou($_[0])->utf8;
}
while( <> ) {
	$string = $_;
	if( /\[\S+\]/ ) {
		if( substr( $string, 0, 1 ) ne "[" ) {
			if( $string =~ /([^[]+)/ ){
				print to_utf($1);
			}
		};
		while( $string =~ /\[([^]]+)\]([^[]*)/g ){
			print "[".ipa_code($1)."]".to_utf($2);
		};
	}else{
	    print to_utf($string);
	}
};
'
}

fatal()
{
	printf >&2 '%s\n' "${0##*/}: $*"
	exit 1
}

[ -n "$1" ] ||
	fatal Please specify input file
[ -f "$1" ] ||
	fatal Input file does not exist
[ -n "$2" ] ||
        fatal Please specify output name

if [ -a "$2.idx" -a "$2.ifo" -a "$2.dict" ]; then
	fatal "$2: dictionary already exists"
fi

cat $1 |MK_DATA|FROM_MUELLER|LC_ALL=C LC_COLLATE=ru_RU.UTF8 LC_CTYPE=ru_RU.UTF8 TO_SLOV $2