183 lines
6.2 KiB
Plaintext
183 lines
6.2 KiB
Plaintext
# © 2016 and later: Unicode, Inc. and others.
|
||
# License & terms of use: http://www.unicode.org/copyright.html#License
|
||
#
|
||
# File: mk_mk_Latn_BGN.txt
|
||
# Generated from CLDR
|
||
#
|
||
|
||
#
|
||
########################################################################
|
||
# BGN/PCGN 1981 System
|
||
#
|
||
# Macedonian was official established as a literary language in
|
||
# Yugoslavia during World War II and is now the official language
|
||
# of Macedonia. Its alphabet is identical to Serbian, except
|
||
# that the letters Ђђ and Ћћ are replaced by Ѓѓ and Ќќ, and
|
||
# the letter Ѕѕ and the apostrophe are added.
|
||
#
|
||
# The Macedonian Alphabet as defined by the BGN (Page 69):
|
||
#
|
||
# АБВГДЃЕЖЗЅИЈКЛЉМНЊОПРСТЌУФХЦЧЏШ
|
||
# абвгдѓежзѕијклљмнњопрстќуфхцчџш’
|
||
#
|
||
# Originally prepared by Michael Everson <everson@evertype.com>
|
||
########################################################################
|
||
#
|
||
# MINIMAL FILTER: Macedonian-Latin
|
||
#
|
||
:: [АБВГДЃЕЖЗЅИЈКЛЉМНЊОПРСТЌУФХЦЧЏШабвгдѓежзѕијклљмнњопрстќуфхцчџш’] ;
|
||
:: NFD (NFC) ;
|
||
#
|
||
#
|
||
########################################################################
|
||
#
|
||
########################################################################
|
||
#
|
||
# Define All Transformation Variables
|
||
#
|
||
########################################################################
|
||
#
|
||
$prime = ’ ;
|
||
$upperConsonants = [БВГДЃЖЗЅЈКЛЉМНЊПРСТЌФХЦЧЏШ] ;
|
||
$lowerConsonants = [бвгдѓжзѕјклљмнњпрстќфхцчџш’] ;
|
||
$consonants = [$upperConsonants $lowerConsonants] ;
|
||
$upperVowels = [АЕИОУ] ;
|
||
$lowerVowels = [аеиоу] ;
|
||
$vowels = [$upperVowels $lowerVowels] ;
|
||
$lower = [$lowerConsonants $lowerVowels] ;
|
||
#
|
||
#
|
||
# Use this $wordBoundary until bug 2034 is fixed in ICU:
|
||
# http://bugs.icu-project.org/cgi-bin/icu-bugs/transliterate?id=2034;expression=boundary;user=guest
|
||
#
|
||
$wordBoundary = [^[:L:][:M:][:N:]] ;
|
||
#
|
||
#
|
||
########################################################################
|
||
#
|
||
########################################################################
|
||
#
|
||
# Start of Alphabetic Transformations
|
||
#
|
||
########################################################################
|
||
#
|
||
А → A ; # CYRILLIC CAPITAL LETTER A
|
||
а → a ; # CYRILLIC SMALL LETTER A
|
||
Б → B ; # CYRILLIC CAPITAL LETTER BE
|
||
б → b ; # CYRILLIC SMALL LETTER BE
|
||
В → V ; # CYRILLIC CAPITAL LETTER VE
|
||
в → v ; # CYRILLIC SMALL LETTER VE
|
||
Г → G ; # CYRILLIC CAPITAL LETTER GHE
|
||
г → g ; # CYRILLIC SMALL LETTER GHE
|
||
Д → D ; # CYRILLIC CAPITAL LETTER DE
|
||
д → d ; # CYRILLIC SMALL LETTER DE
|
||
#
|
||
#
|
||
########################################################################
|
||
#
|
||
# BGN Page 70 Rule 1:
|
||
#
|
||
# The character ѓ should be romanized g when it occurs before е
|
||
# and и. In all other instances, it should be romanized đ (Đ).
|
||
#
|
||
########################################################################
|
||
#
|
||
Ѓ}[ЕеИи] → G ; # CYRILLIC CAPITAL LETTER GJE
|
||
ѓ}[ЕеИи] → g ; # CYRILLIC SMALL LETTER GJE
|
||
Ѓ → Đ ; # CYRILLIC CAPITAL LETTER GJE
|
||
ѓ → đ ; # CYRILLIC SMALL LETTER GJE
|
||
#
|
||
#
|
||
########################################################################
|
||
#
|
||
# End Rule 1
|
||
#
|
||
########################################################################
|
||
#
|
||
Е → E ; # CYRILLIC CAPITAL LETTER DE
|
||
е → e ; # CYRILLIC SMALL LETTER DE
|
||
Ж → Ž ; # CYRILLIC CAPITAL LETTER ZHE
|
||
ж → ž ; # CYRILLIC SMALL LETTER ZHE
|
||
З → Z ; # CYRILLIC CAPITAL LETTER ZE
|
||
з → z ; # CYRILLIC SMALL LETTER ZE
|
||
Ѕ} $lower → Dz ; # CYRILLIC CAPITAL LETTER DZE
|
||
Ѕ → DZ ; # CYRILLIC CAPITAL LETTER DZE
|
||
ѕ → dz ; # CYRILLIC SMALL LETTER DZE
|
||
И → I ; # CYRILLIC CAPITAL LETTER I
|
||
и → i ; # CYRILLIC SMALL LETTER I
|
||
Ј → J ; # CYRILLIC CAPITAL LETTER JE
|
||
ј → j ; # CYRILLIC SMALL LETTER JE
|
||
К → K ; # CYRILLIC CAPITAL LETTER KA
|
||
к → k ; # CYRILLIC SMALL LETTER KA
|
||
Л → L ; # CYRILLIC CAPITAL LETTER EL
|
||
л → l ; # CYRILLIC SMALL LETTER EL
|
||
Љ} $lower → Lj ; # CYRILLIC CAPITAL LETTER LJE
|
||
Љ → LJ ; # CYRILLIC CAPITAL LETTER LJE
|
||
љ → lj ; # CYRILLIC SMALL LETTER LJE
|
||
М → M ; # CYRILLIC CAPITAL LETTER EM
|
||
м → m ; # CYRILLIC SMALL LETTER EM
|
||
Н → N ; # CYRILLIC CAPITAL LETTER EN
|
||
н → n ; # CYRILLIC SMALL LETTER EN
|
||
Њ} $lower → Nj ; # CYRILLIC CAPITAL LETTER NJE
|
||
Њ → NJ ; # CYRILLIC CAPITAL LETTER NJE
|
||
њ → nj ; # CYRILLIC SMALL LETTER NJE
|
||
О → O ; # CYRILLIC CAPITAL LETTER O
|
||
о → o ; # CYRILLIC SMALL LETTER O
|
||
П → P ; # CYRILLIC CAPITAL LETTER PE
|
||
п → p ; # CYRILLIC SMALL LETTER PE
|
||
Р → R ; # CYRILLIC CAPITAL LETTER ER
|
||
р → r ; # CYRILLIC SMALL LETTER ER
|
||
С → S ; # CYRILLIC CAPITAL LETTER ES
|
||
с → s ; # CYRILLIC SMALL LETTER ES
|
||
Т → T ; # CYRILLIC CAPITAL LETTER TE
|
||
т → t ; # CYRILLIC SMALL LETTER TE
|
||
#
|
||
#
|
||
########################################################################
|
||
#
|
||
# BGN Page 70 Rule 2:
|
||
#
|
||
# The character ќ should be romanized k when it occurs before е
|
||
# and и. In all other instances, it should be romanized c\u0301.
|
||
#
|
||
########################################################################
|
||
#
|
||
Ќ}[ЕеИи] → K ; # CYRILLIC CAPITAL LETTER KJE
|
||
ќ}[ЕеИи] → k ; # CYRILLIC SMALL LETTER KJE
|
||
Ќ → C\u0301 ; # CYRILLIC CAPITAL LETTER KJE
|
||
ќ → c\u0301 ; # CYRILLIC SMALL LETTER KJE
|
||
#
|
||
#
|
||
########################################################################
|
||
#
|
||
# End Rule 2
|
||
#
|
||
########################################################################
|
||
#
|
||
У → U ; # CYRILLIC CAPITAL LETTER U
|
||
у → u ; # CYRILLIC SMALL LETTER U
|
||
Ф → F ; # CYRILLIC CAPITAL LETTER EF
|
||
ф → f ; # CYRILLIC SMALL LETTER EF
|
||
Х → H ; # CYRILLIC CAPITAL LETTER HA
|
||
х → h ; # CYRILLIC SMALL LETTER HA
|
||
Ц → C ; # CYRILLIC CAPITAL LETTER TSE
|
||
ц → c ; # CYRILLIC SMALL LETTER TSE
|
||
Ч → Č ; # CYRILLIC CAPITAL LETTER CHE
|
||
ч → č ; # CYRILLIC SMALL LETTER CHE
|
||
Џ} $lower → Dž ; # CYRILLIC CAPITAL LETTER SHA
|
||
Џ → DŽ ; # CYRILLIC CAPITAL LETTER SHA
|
||
џ → dž ; # CYRILLIC SMALL LETTER SHA
|
||
Ш → Š ; # CYRILLIC CAPITAL LETTER SHA
|
||
ш → š ; # CYRILLIC SMALL LETTER SHA
|
||
#
|
||
#
|
||
########################################################################
|
||
#
|
||
# BGN Page 69 Rule 32, maps the symbol onto itself and
|
||
# is ignored here for computational efficiency.
|
||
#
|
||
# $prime → $prime ; # RIGHT SINGLE QUOTATION MARK
|
||
#
|
||
########################################################################
|
||
|