• Home
  • Line#
  • Scopes#
  • Navigate#
  • Raw
  • Download
1# © 2016 and later: Unicode, Inc. and others.
2# License & terms of use: http://www.unicode.org/copyright.html
3# Generated using tools/cldr/cldr-to-icu/build-icu-data.xml
4#
5# File: ru_ru_Latn_BGN.txt
6# Generated from CLDR
7#
8
9# BGN/PCGN 1947 System
10#
11# The BGN/PCGN system for Russian was adopted by the BGN in 1944 and
12# by the PCGN in 1947 for use in romanizing names written in the
13# Russian Cyrillic alphabet.
14#
15# The Russian Alphabet as defined by the BGN (Page 93):
16#     АБВГДЕЁЖЗИЙКЛМНОПРСТУФХЦЧШЩЪЫЬЭЮЯ
17#     абвгдеёжзийклмнопрстуфхцчшщъыьэюя
18#
19# Originally prepared by Michael Everson everson@evertype.com
20# Fixed by Frank Yung-Fong Tang ftang@google.com
21#
22# Test Data from http://en.wikipedia.org/wiki/BGN/PCGN_romanization_of_Russian
23########################################################################
24# MINIMAL FILTER: Russian-Latin
25::[АБВГДЕЁЖЗИЙКЛМНОПРСТУФХЦЧШЩЪЫЬЭЮЯабвгдеёжзийклмнопрстуфхцчшщъыьэюя];
26::NFC;
27########################################################################
28# Define All Transformation Variables
29########################################################################
30$prime = ʹ ;
31$doublePrime = ʺ ;
32$wordBoundary = [^[:L:][:M:][:N:]] ;
33$upperVowels = [АЕЁЭИОУЫЮЯ] ;
34$lowerVowels = [аеёэиоуыюя] ;
35$vowels = [$upperVowels $lowerVowels] ;
36$upperConsonants = [[:Uppercase:]-$vowels] ;
37$lowerConsonants = [[:Lowercase:]-$vowels] ;
38$consonants = [$upperConsonants $lowerConsonants] ;
39$upper = [:Uppercase:];
40$lower = [:Lowercase:];
41########################################################################
42# Rules moved to front to avoid masking
43########################################################################
44$lowerVowels { ы → ·y ;
45$upperVowels { [Ыы] } $lower → ·y ;
46$upperVowels { [Ыы] } → ·Y ;
47[$consonants - [Йй]]{Э → ·E ;
48[$consonants - [Йй]]{э → ·e ;
49[$upperVowels [ЙЪЬ]] { Е } $upper → YE ; # CYRILLIC CAPITAL LETTER IE
50[$upperVowels [ЙЪЬ]] { Е → Ye ; # CYRILLIC CAPITAL LETTER IE
51[$upperVowels $lowerVowels [ЙйЪъЬь]] { е → ye ; # CYRILLIC SMALL LETTER IE
52[$upperVowels [ЙЪЬ]] { Ё } $upper → YË ; # CYRILLIC CAPITAL LETTER IO
53[$upperVowels [ЙЪЬ]] { Ё → Yë ; # CYRILLIC CAPITAL LETTER IO
54[$upperVowels $lowerVowels [ЙйЪъЬь]] { ё → yë ; # CYRILLIC SMALL LETTER IO
55# Since in the above rule we look at the Cyrillic context before the E/Ё/ё,
56# we have to transform these in a separate pass before we change the vowels.
57# The ::Null forces a separate pass.
58::Null;
59########################################################################
60# Start of Alphabetic Transformations
61########################################################################
62А → A ; # CYRILLIC CAPITAL LETTER A
63а → a ; # CYRILLIC SMALL LETTER A
64Б → B ; # CYRILLIC CAPITAL LETTER BE
65б → b ; # CYRILLIC SMALL LETTER BE
66В → V ; # CYRILLIC CAPITAL LETTER VE
67в → v ; # CYRILLIC SMALL LETTER VE
68Г → G ; # CYRILLIC CAPITAL LETTER GHE
69г → g ; # CYRILLIC SMALL LETTER GHE
70Д → D ; # CYRILLIC CAPITAL LETTER DE
71д → d ; # CYRILLIC SMALL LETTER DE
72########################################################################
73# BGN Page 94 Rule 1:
74# # The character e should be romanized ye
75# initially, after the vowel # characters a, e, ё, и, о, у, ы, э, ю,
76# and я, and after й, ъ, and ь.
77# In all other instances, it should
78# be romanized e.
79########################################################################
80# BUG(ftang)- the following two lines said BEFORE the vowels, instead of AFTER
81# Е}[$upperVowels [ЙЪЬ]] → YE ; # CYRILLIC CAPITAL LETTER IE
82# Е}[$lowerVowels [йъь]] → Ye ; # CYRILLIC CAPITAL LETTER IE
83$wordBoundary{Е} $upper → YE ; # CYRILLIC CAPITAL LETTER IE
84$wordBoundary{Е → Ye ; # CYRILLIC CAPITAL LETTER IE
85Е → E ; # CYRILLIC CAPITAL LETTER IE
86#
87# BUG(ftang)- the following line said BEFORE the vowels, instead of AFTER
88# е}[$upperVowels $lowerVowels [ЙйЪъЬь]] → ye ; # CYRILLIC SMALL LETTER IE
89$wordBoundary{е → ye ; # CYRILLIC SMALL LETTER IE
90е → e ; # CYRILLIC SMALL LETTER IE
91########################################################################
92# End of Rule 1
93########################################################################
94########################################################################
95# BGN Page 94 Rule 2:
96#
97# The character ё is not considered a separate character of the
98# Russian alphabet and the dieresis is generally not shown. When the
99# dieresis is shown, the character should be romanized yë initially,
100# after the vowel characters a, e, ё, и, о, у, ы, э, ю, and я, and
101# after й, ъ, and ь, In all other instances, it should be romanized
102# ё. When the dieresis is not shown, the character may still be
103# romanized in the preceding manner or, alternatively, in accordance
104# with note 1.
105########################################################################
106# BUG(ftang)- the following two lines said BEFORE the vowels, instead of AFTER
107# Ё}[$upperVowels [ЙЪЬ]] → YË ; # CYRILLIC CAPITAL LETTER IO
108# Ё}[$lowerVowels [йъь]] → Yë ; # CYRILLIC CAPITAL LETTER IO
109$wordBoundary {Ё} [·]? $upper → YË ; # CYRILLIC CAPITAL LETTER IO
110$wordBoundary {Ё} [·]? $lower → Yë ; # CYRILLIC CAPITAL LETTER IO
111Ё → Ë ; # CYRILLIC CAPITAL LETTER IO
112# BUG(ftang)- the following line said BEFORE the vowels, instead of AFTER
113# ё}[$upperVowels $lowerVowels [ЙйЪъЬь]] → yë ; # CYRILLIC SMALL LETTER IO
114$wordBoundary{ё → yë ; # CYRILLIC SMALL LETTER IO
115ё → ë ; # CYRILLIC SMALL LETTER IO
116########################################################################
117# End of Rule 2
118########################################################################
119Ж} $lower → Zh ; # CYRILLIC CAPITAL LETTER ZHE
120Ж → ZH ; # CYRILLIC CAPITAL LETTER ZHE
121ж → zh ; # CYRILLIC SMALL LETTER ZHE
122########################################################################
123# BGN Page 94 Rule 3.4
124# э after any consonant character except
125# й becomes ·е
126########################################################################
127З → Z ; # CYRILLIC CAPITAL LETTER ZE
128з → z ; # CYRILLIC SMALL LETTER ZE
129# BUG(ftang) The following two lines said those consonant becomes ·е
130# [$consonants - [Йй]]}Э → ·Е ;
131# [$consonants - [Йй]]}э → ·е ;
132########################################################################
133# End of Rule 3.4
134########################################################################
135И → I ; # CYRILLIC CAPITAL LETTER I
136и → i ; # CYRILLIC SMALL LETTER I
137########################################################################
138# BGN Page 94 Rule 3:
139#
140# Unusual Russian character sequences occurring primarily in
141# non-Russian-language names may be romanized as shown below in order
142# to provide differentiation from regularly-occurring digraphs and
143# character sequences.
144#
145# BGN Page 94 Rule 3.1
146# й before а, у, ы, or э becomes у·
147########################################################################
148Й}[АаУуЫыЭэ] → Y· ; # CYRILLIC CAPITAL LETTER I
149й}[АаУуЫыЭэ] → y· ; # CYRILLIC SMALL LETTER I
150Й → Y ; # CYRILLIC CAPITAL LETTER I
151й → y ; # CYRILLIC SMALL LETTER I
152########################################################################
153# End Rule 3.1
154########################################################################
155К → K ; # CYRILLIC CAPITAL LETTER KA
156к → k ; # CYRILLIC SMALL LETTER KA
157Л → L ; # CYRILLIC CAPITAL LETTER EL
158л → l ; # CYRILLIC SMALL LETTER EL
159М → M ; # CYRILLIC CAPITAL LETTER EM
160м → m ; # CYRILLIC SMALL LETTER EM
161Н → N ; # CYRILLIC CAPITAL LETTER EN
162н → n ; # CYRILLIC SMALL LETTER EN
163О → O ; # CYRILLIC CAPITAL LETTER O
164о → o ; # CYRILLIC SMALL LETTER O
165П → P ; # CYRILLIC CAPITAL LETTER PE
166п → p ; # CYRILLIC SMALL LETTER PE
167Р → R ; # CYRILLIC CAPITAL LETTER ER
168р → r ; # CYRILLIC SMALL LETTER ER
169С → S ; # CYRILLIC CAPITAL LETTER ES
170с → s ; # CYRILLIC SMALL LETTER ES
171########################################################################
172# BGN Page 94 Rule 3.5
173# тс becomes t·s
174########################################################################
175ТС → T·S ; # CYRILLIC CAPITAL LETTER TE
176Тс → T·s ; # CYRILLIC CAPITAL LETTER TE
177тс → t·s ; # CYRILLIC SMALL LETTER TE
178Т → T ; # CYRILLIC CAPITAL LETTER TE
179т → t ; # CYRILLIC SMALL LETTER TE
180########################################################################
181# End Rule 3.5
182########################################################################
183У → U ; # CYRILLIC CAPITAL LETTER U
184у → u ; # CYRILLIC SMALL LETTER U
185Ф → F ; # CYRILLIC CAPITAL LETTER EF
186ф → f ; # CYRILLIC SMALL LETTER EF
187Х} $lower → Kh ; # CYRILLIC CAPITAL LETTER HA
188Х → KH ; # CYRILLIC CAPITAL LETTER HA
189х → kh ; # CYRILLIC SMALL LETTER HA
190Ц} $lower → Ts ; # CYRILLIC CAPITAL LETTER TSE
191Ц → TS ; # CYRILLIC CAPITAL LETTER TSE
192ц → ts ; # CYRILLIC SMALL LETTER TSE
193Ч} $lower → Ch ; # CYRILLIC CAPITAL LETTER CHE
194Ч → CH ; # CYRILLIC CAPITAL LETTER CHE
195ч → ch ; # CYRILLIC SMALL LETTER CHE
196########################################################################
197# BGN Page 94 Rule 3.6
198# шч becomes sh·ch
199########################################################################
200ШЧ → SH·CH ; # CYRILLIC CAPITAL LETTER SHA
201Шч → Sh·ch ; # CYRILLIC CAPITAL LETTER SHA
202шч → sh·ch ; # CYRILLIC SMALL LETTER SHA
203Ш} $lower → Sh ; # CYRILLIC CAPITAL LETTER SHA
204Ш → SH ; # CYRILLIC CAPITAL LETTER SHA
205ш → sh ; # CYRILLIC SMALL LETTER SHA
206Щ} $lower → Shch ; # CYRILLIC CAPITAL LETTER SHCHA
207Щ → SHCH ; # CYRILLIC CAPITAL LETTER SHCHA
208щ → shch ; # CYRILLIC SMALL LETTER SHCHA
209########################################################################
210# End Rule 3.6
211########################################################################
212Ъ → $doublePrime ; # CYRILLIC CAPITAL LETTER HARD SIGN
213ъ → $doublePrime ; # CYRILLIC SMALL LETTER HARD SIGN
214########################################################################
215# BGN Page 94 Rule 3.2
216# ы before а, у, ы, or э becomes у·
217#
218# BGN Page 94 Rule 3.3
219# ы after any vowel character becomes ·у
220########################################################################
221#
222# BUG(ftang) the following line said the vowels will change
223# $vowels}Ы → ·Y ; # CYRILLIC CAPITAL LETTER I
224# $vowels}ы → ·y ; # CYRILLIC CAPITAL LETTER I
225Ы}[АаУуЫыЭэ] → Y· ; # CYRILLIC CAPITAL LETTER YERU
226ы}[ауыэ] → y· ; # CYRILLIC SMALL LETTER YERU
227Ы → Y ; # CYRILLIC CAPITAL LETTER YERU
228ы → y ; # CYRILLIC SMALL LETTER YERU
229########################################################################
230# End Rule 3.2 and 3.3
231########################################################################
232Ь → $prime ; # CYRILLIC CAPITAL LETTER SOFT SIGN
233ь → $prime ; # CYRILLIC SMALL LETTER SOFT SIGN
234Э → E ; # CYRILLIC CAPITAL LETTER E
235э → e ; # CYRILLIC SMALL LETTER E
236Ю} $lower → Yu ; # CYRILLIC CAPITAL LETTER YU
237Ю → YU ; # CYRILLIC CAPITAL LETTER YU
238ю → yu ; # CYRILLIC SMALL LETTER YU
239Я} $lower → Ya ; # CYRILLIC CAPITAL LETTER YA
240Я → YA ; # CYRILLIC CAPITAL LETTER YA
241я → ya ; # CYRILLIC SMALL LETTER YA
242
243