• 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: bg_bg_Latn_BGN.txt
6# Generated from CLDR
7#
8
9#
10########################################################################
11# BGN/PCGN 1952 System
12#
13# This system was adopted by the BGN in 1949 and by the PCGN in 1952.
14# It reflects the much simplified Bulgarian orthography as officially
15# revised in February 1945. The Bulgarian alphabet contains all of
16# the characters present in the Russian alphabet with the exception
17# of Ёё, Ыы, and Ээ. Two obsolete letters Ѫѫ and Ѣѣ are also given.
18#
19# The Bulgarian Alphabet as defined by the BGN (Page 15):
20#
21#   АБВГДЕЖЗИЙКЛМНОПРСТУФХЦЧШЩЪЬЮЯѪѢ
22#   абвгдежзийклмнопрстуфхцчшщъьюяѫѣ
23#
24# Originally prepared by Michael Everson <everson@evertype.com>
25########################################################################
26#
27# MINIMAL FILTER: Bulgarian-Latin
28#
29:: [АБВГДЕЖЗИЙКЛМНОПРСТУФХЦЧШЩЪЬЮЯѪѢабвгдежзийклмнопрстуфхцчшщъьюяѫѣ] ;
30:: NFD (NFC) ;
31#
32#
33########################################################################
34#
35########################################################################
36#
37# Define All Transformation Variables
38#
39########################################################################
40#
41$upperConsonants = [БВГДЖЗЙКЛМНПРСТФХЦЧШЩЬ] ;
42$lowerConsonants = [бвгджзйклмнпрстфхцчшщь] ;
43$consonants = [$upperConsonants $lowerConsonants] ;
44$upperVowels = [АЕИОУЪЮЯѪѢ] ;
45$lowerVowels = [аеиоуъюяѫѣ] ;
46$vowels = [$upperVowels $lowerVowels] ;
47$lower = [$lowerConsonants $lowerVowels] ;
48$bulgarian = [ $lower $upperConsonants $upperVowels ] ;
49#
50#
51# Use this $wordBoundary until bug 2034 is fixed in ICU:
52# http://bugs.icu-project.org/cgi-bin/icu-bugs/transliterate?id=2034;expression=boundary;user=guest
53#
54$wordBoundary =  [^[:L:][:M:][:N:]] ;
55#
56#
57########################################################################
58#
59########################################################################
60#
61# Start of Alphabetic Transformations
62#
63########################################################################
64#
65А → A ; # CYRILLIC CAPITAL LETTER A
66а → a ; # CYRILLIC SMALL LETTER A
67Б → B ; # CYRILLIC CAPITAL LETTER BE
68б → b ; # CYRILLIC SMALL LETTER BE
69В → V ; # CYRILLIC CAPITAL LETTER VE
70в → v ; # CYRILLIC SMALL LETTER VE
71Г → G ; # CYRILLIC CAPITAL LETTER GHE
72г → g ; # CYRILLIC SMALL LETTER GHE
73Д → D ; # CYRILLIC CAPITAL LETTER DE
74д → d ; # CYRILLIC SMALL LETTER DE
75Е → E ; # CYRILLIC CAPITAL LETTER DE
76е → e ; # CYRILLIC SMALL LETTER DE
77Ж} $lower → Zh ; # CYRILLIC CAPITAL LETTER ZHE
78Ж → ZH ; # CYRILLIC CAPITAL LETTER ZHE
79ж → zh ; # CYRILLIC SMALL LETTER ZHE
80З → Z ; # CYRILLIC CAPITAL LETTER ZE
81з → z ; # CYRILLIC SMALL LETTER ZE
82И → I ; # CYRILLIC CAPITAL LETTER I
83и → i ; # CYRILLIC SMALL LETTER I
84Й → Y ; # CYRILLIC CAPITAL LETTER I
85й → y ; # CYRILLIC SMALL LETTER I
86К → K ; # CYRILLIC CAPITAL LETTER KA
87к → k ; # CYRILLIC SMALL LETTER KA
88Л → L ; # CYRILLIC CAPITAL LETTER EL
89л → l ; # CYRILLIC SMALL LETTER EL
90М → M ; # CYRILLIC CAPITAL LETTER EM
91м → m ; # CYRILLIC SMALL LETTER EM
92Н → N ; # CYRILLIC CAPITAL LETTER EN
93н → n ; # CYRILLIC SMALL LETTER EN
94О → O ; # CYRILLIC CAPITAL LETTER O
95о → o ; # CYRILLIC SMALL LETTER O
96П → P ; # CYRILLIC CAPITAL LETTER PE
97п → p ; # CYRILLIC SMALL LETTER PE
98Р → R ; # CYRILLIC CAPITAL LETTER ER
99р → r ; # CYRILLIC SMALL LETTER ER
100С → S ; # CYRILLIC CAPITAL LETTER ES
101с → s ; # CYRILLIC SMALL LETTER ES
102#
103#
104########################################################################
105#
106# BGN Page 16 Note 4
107#
108# тс becomes t·s
109#
110########################################################################
111#
112ТС → T·S ; # CYRILLIC CAPITAL LETTER TE
113Тс → T·s ; # CYRILLIC CAPITAL LETTER TE
114тс → t·s ; # CYRILLIC SMALL LETTER TE
115Т → T ; # CYRILLIC CAPITAL LETTER TE
116т → t ; # CYRILLIC SMALL LETTER TE
117#
118#
119########################################################################
120#
121# End Note 4
122#
123########################################################################
124У → U ; # CYRILLIC CAPITAL LETTER U
125у → u ; # CYRILLIC SMALL LETTER U
126Ф → F ; # CYRILLIC CAPITAL LETTER EF
127ф → f ; # CYRILLIC SMALL LETTER EF
128Х} $lower → Kh ; # CYRILLIC CAPITAL LETTER HA
129Х → KH ; # CYRILLIC CAPITAL LETTER HA
130х → kh ; # CYRILLIC SMALL LETTER HA
131Ц} $lower → Ts ; # CYRILLIC CAPITAL LETTER TSE
132Ц → TS ; # CYRILLIC CAPITAL LETTER TSE
133ц → ts ; # CYRILLIC SMALL LETTER TSE
134Ч} $lower → Ch ; # CYRILLIC CAPITAL LETTER CHE
135Ч → CH ; # CYRILLIC CAPITAL LETTER CHE
136ч → ch ; # CYRILLIC SMALL LETTER CHE
137#
138########################################################################
139#
140# Implied rule from BGN Russian-Latin transliteration (Page 94 Note 3.6).
141#
142# шт becomes sh·t
143#
144########################################################################
145#
146ШТ → SH·T ; # CYRILLIC CAPITAL LETTER SHA
147Шт → Sh·t ; # CYRILLIC CAPITAL LETTER SHA
148шт → sh·t ; # CYRILLIC SMALL LETTER SHA
149Ш} $lower → Sh ; # CYRILLIC CAPITAL LETTER SHA
150Ш → SH ; # CYRILLIC CAPITAL LETTER SHA
151ш → sh ; # CYRILLIC SMALL LETTER SHA
152Щ} $lower → Sht ; # CYRILLIC CAPITAL LETTER SHCHA
153Щ → SHT ; # CYRILLIC CAPITAL LETTER SHCHA
154щ → sht ; # CYRILLIC SMALL LETTER SHCHA
155#
156#
157########################################################################
158#
159# End Implied rule
160#
161########################################################################
162Ъ → Ŭ ; # CYRILLIC CAPITAL LETTER HARD SIGN
163ъ → ŭ ; # CYRILLIC SMALL LETTER HARD SIGN
164#
165########################################################################
166#
167# BGN Page 16 Note 1
168#
169# In modern Bulgarian orthography, the character ъ does not occur in
170# word-final position.  It should be omitted in romanization when found
171# on older sources.
172#
173# The following rule removes all Ъъ at the end of a word. It is assumed
174# that when the condition is met, the text must be from an older source.
175# Comment out with a '#' at the start of a line to disable.
176#
177#
178########################################################################
179#
180$bulgarian { [Ъъ]  } $wordBoundary > ;
181#
182#
183########################################################################
184#
185# End BGN Page 16 Note 1
186#
187########################################################################
188Ь → ’ ; # CYRILLIC CAPITAL LETTER SOFT SIGN
189ь → ’ ; # CYRILLIC SMALL LETTER SOFT SIGN
190Ю} $lower → Yu ; # CYRILLIC CAPITAL LETTER YU
191Ю → YU ; # CYRILLIC CAPITAL LETTER YU
192ю → yu ; # CYRILLIC SMALL LETTER YU
193Я} $lower → Ya ; # CYRILLIC CAPITAL LETTER YA
194Я → YA ; # CYRILLIC CAPITAL LETTER YA
195я → ya ; # CYRILLIC SMALL LETTER YA
196#
197########################################################################
198#
199# BGN Page 16 Note 2
200#
201# The obsolete character Ѫ, which was replaced by Ъ in 1945, should be
202# romanized Ŭ.
203#
204########################################################################
205#
206Ѫ → Ŭ ; # CYRILLIC CAPITAL LETTER BIG YUS
207ѫ → ŭ ; # CYRILLIC SMALL LETTER BIG YUS
208#
209#
210########################################################################
211#
212# End BGN Page 16 Note 2
213#
214########################################################################
215#
216########################################################################
217#
218# BGN Page 16 Note 3
219#
220# The obsolete character Ѣ, replaced in 1945 by Я or Е according to local
221# pronunciation, should be romanized as e or ya, accordingly, if the
222# pronunciation is known; otherwise as ye.
223#
224########################################################################
225#
226Ѣ} $lower → Ye ; # CYRILLIC CAPITAL LETTER YAT
227Ѣ → YE ; # CYRILLIC CAPITAL LETTER YAT
228ѣ → ye ; # CYRILLIC SMALL LETTER YAT
229#
230#
231# Alternative rule where appropriate for local pronounciation. To apply
232# uncomment the following by removing the '#' mark at the start of the
233# line and insert before the three rule lines above.
234#
235# Ѣ} $lower → e ; # CYRILLIC CAPITAL LETTER YAT
236# Ѣ → E ; # CYRILLIC CAPITAL LETTER YAT
237# ѣ → e ; # CYRILLIC SMALL LETTER YAT
238#
239########################################################################
240#
241# End BGN Page 16 Note 3
242#
243########################################################################
244
245