src/com/ibm/text/resources/Transliterator_Latin_Cyrillic.utf8.txt

#--------------------------------------------------------------------
# Copyright (c) 1999-2001, International Business Machines
# Corporation and others.  All Rights Reserved.
#--------------------------------------------------------------------
# Date: Tue Jan 23 12:42:05 2001
#--------------------------------------------------------------------

# Latin-Cyrillic


# These rules provide general Latin-Cyrillic
# transliteration. The standard Russian transliterations
# are generally used for the letters from Russian,
# with additional Cyrillic characters given consistent
# mappings.

### $S_hacek=Š;
### $s_hacek=š;

### $YO=Ё;
### $J=Ј;
### $A=А;
### $B=Б;
### $V=В;
### $G=Г;
### $D=Д;
### $YE=Е;
### $ZH=Ж;
### $Z=З;
### $YI=И;
### $Y=Й;
### $K=К;
### $L=Л;
### $M=М;
### $N=Н;
### $O=О;
### $P=П;
### $R=Р;
### $S=С;
### $T=Т;
### $U=У;
### $F=Ф;
### $KH=Х;
### $TS=Ц;
### $CH=Ч;
### $SH=Ш;
### $SHCH=Щ;
### $HARD=Ъ;
### $I=Ы;
### $SOFT=Ь;
### $E=Э;
### $YU=Ю;
### $YA=Я;

# Lowercase

### $a=а;
### $b=б;
### $v=в;
### $g=г;
### $d=д;
### $ye=е;
### $zh=ж;
### $z=з;
### $yi=и;
### $y=й;
### $k=к;
### $l=л;
### $m=м;
### $n=н;
### $o=о;
### $p=п;
### $r=р;
### $s=с;
### $t=т;
### $u=у;
### $f=ф;
### $kh=х;
### $ts=ц;
### $ch=ч;
### $sh=ш;
### $shch=щ;
### $hard=ъ;
### $i=ы;
### $soft=ь;
### $e=э;
### $yu=ю;
### $ya=я;

### $yo=ё;
### $j=ј;

# variables
# some are duplicated so lowercasing works

$csoft=[eiyEIY];
$CSOFT=[eiyEIY];

$BECOMES_H=[Ъъ];
$becomes_h=[Ъъ];

$BECOMES_S=[Сс];
$becomes_s=[Сс];

$BECOMES_C=[Чч];
$becomes_c=[Чч];

$BECOMES_VOWEL=[АЭЫОУаэыоу];
$becomes_vowel=[АЭЫОУаэыоу];

$letter=[[:Lu:][:Ll:]];
$lower=[[:Ll:]];

# Modified to combine display transliterator and typing transliterator.
# The display mapping uses accents for the "soft" vowels.
# It does not, although it could, use characters like š instead of digraphs
# like sh.

# #############################################
# Special titlecase forms, not duplicated
# #############################################

Sh''ch<>Шч; # LIU Distinguish Шч from Щ

Ch   <> {Ч}  $lower;
Kh   <> {Х}  $lower;
Shch <> {Щ}$lower;
Sh   <> {Ш}  $lower;
Ts   <> {Ц}  $lower;
Zh   <> {Ж}  $lower;
Yi>И;
Ye>Е;
Yo>Ё;
Yu>Ю;
Ya>Я;

# #############################################
# Rules to Duplicate
# To get the lowercase versions, copy these and lowercase
# #############################################

# variant spellings in English

SHTCH>Щ;
TCH>Ч;
TH>З;
Q>К;
WH>В;
W>В;
X>КС; #+ "X<КС;"

# Separate letters that would otherwise join

SH''<Ш}$BECOMES_C;
T''<Т}$BECOMES_S;
T''<Т}[ЧЩщ]; # LIU add special cases

K''<К}$BECOMES_H;
S''<С}$BECOMES_H;
T''<Т}$BECOMES_H;
Z''<З}$BECOMES_H;

Y''<Й}$BECOMES_VOWEL;

# Main letters

A<>А;
B<>Б;
CH<>Ч;
D<>Д;
E<>Э;
F<>Ф;
G<>Г;
Ì<>И;
I<>Ы;
KH<>Х;
K<>К;
L<>Л;
M<>М;
N<>Н;
O<>О;
P<>П;
R<>Р;
SHCH<>Щ;
SH>Ш; #+ "SH<Ш;"
Š<>Ш;
S<>С;
TS<>Ц;
T<>Т;
U<>У;
V<>В;
#ÌÀÈÒÙ
YE>Е; #+ "YE<Е;"
È<>Е;
YO>Ё; #+ "YO<Ё;"
Ò<>Ё;
YU>Ю; #+ "YU<Ю;"
Ù<>Ю;
YA>Я; #+ "YA<Я;"
À<>Я;
Y<>Й;
ZH<>Ж;
Z<>З;

H<>Ъ;
Ÿ<>Ь;

# Non-russian

J<>Ј;

# variant spellings in English

C}$csoft>С;
C>К;

# #############################################
# Duplicated Rules
# Copy and lowercase the above rules
# #############################################

# variant spellings in english

shtch>щ;
tch>ч;
th>з;
q>к;
wh>в;
w>в;
x>кс; #+ "x<кс;"

# separate letters that would otherwise join

sh''<ш}$becomes_c;
t''<т}$becomes_s;
t''<т}[чщ]; # LIU add special cases

k''<к}$becomes_h;
s''<с}$becomes_h;
t''<т}$becomes_h;
z''<з}$becomes_h;

y''<й}$becomes_vowel;

# main letters

a<>а;
b<>б;
ch<>ч;
d<>д;
e<>э;
f<>ф;
g<>г;
ì<>и;
i<>ы;
kh<>х;
k<>к;
l<>л;
m<>м;
n<>н;
o<>о;
p<>п;
r<>р;
shch<>щ;
sh>ш; #+ "sh<ш;"
š<>ш;
s<>с;
ts<>ц;
t<>т;
u<>у;
v<>в;
#ìàèòù
ye>е; #+ "ye<е;"
è<>е;
yo>ё; #+ "yo<ё;"
ò<>ё;
yu>ю; #+ "yu<ю;"
ù<>ю;
ya>я; #+ "ya<я;"
à<>я;
y<>й;
zh<>ж;
z<>з;

h<>ъ;
ÿ<>ь;

# non-russian

j<>ј;

# variant spellings in english

c}$csoft>с;
c>к;


# #############################################
# End of Duplicated Rules
# #############################################

#generally the last rule
''>;

# eof