git.saurik.com Git - apple/icu.git/blob - icuSources/data/translit/bg_bg_Latn

2 # License & terms of use: http://www.unicode.org/copyright.html#License

3 #

4 # File: bg_bg_Latn_BGN.txt

5 # Generated from CLDR

6 #

8 #

9 ########################################################################

10 # BGN/PCGN 1952 System

11 #

12 # This system was adopted by the BGN in 1949 and by the PCGN in 1952.

13 # It reflects the much simplified Bulgarian orthography as officially

14 # revised in February 1945. The Bulgarian alphabet contains all of

15 # the characters present in the Russian alphabet with the exception

16 # of Ёё, Ыы, and Ээ. Two obsolete letters Ѫѫ and Ѣѣ are also given.

17 #

18 # The Bulgarian Alphabet as defined by the BGN (Page 15):

19 #

20 # АБВГДЕЖЗИЙКЛМНОПРСТУФХЦЧШЩЪЬЮЯѪѢ

21 # абвгдежзийклмнопрстуфхцчшщъьюяѫѣ

22 #

23 # Originally prepared by Michael Everson <everson@evertype.com>

24 ########################################################################

25 #

26 # MINIMAL FILTER: Bulgarian-Latin

27 #

28 :: [АБВГДЕЖЗИЙКЛМНОПРСТУФХЦЧШЩЪЬЮЯѪѢабвгдежзийклмнопрстуфхцчшщъьюяѫѣ] ;

29 :: NFD (NFC) ;

30 #

31 #

32 ########################################################################

33 #

34 ########################################################################

35 #

36 # Define All Transformation Variables

37 #

38 ########################################################################

39 #

40 $upperConsonants = [БВГДЖЗЙКЛМНПРСТФХЦЧШЩЬ] ;

41 $lowerConsonants = [бвгджзйклмнпрстфхцчшщь] ;

42 $consonants = [$upperConsonants $lowerConsonants] ;

43 $upperVowels = [АЕИОУЪЮЯѪѢ] ;

44 $lowerVowels = [аеиоуъюяѫѣ] ;

45 $vowels = [$upperVowels $lowerVowels] ;

46 $lower = [$lowerConsonants $lowerVowels] ;

47 $bulgarian = [ $lower $upperConsonants $upperVowels ] ;

48 #

49 #

50 # Use this $wordBoundary until bug 2034 is fixed in ICU:

51 # http://bugs.icu-project.org/cgi-bin/icu-bugs/transliterate?id=2034;expression=boundary;user=guest

52 #

53 $wordBoundary = [^[:L:][:M:][:N:]] ;

54 #

55 #

56 ########################################################################

57 #

58 ########################################################################

59 #

60 # Start of Alphabetic Transformations

61 #

62 ########################################################################

63 #

64 А → A ; # CYRILLIC CAPITAL LETTER A

65 а → a ; # CYRILLIC SMALL LETTER A

66 Б → B ; # CYRILLIC CAPITAL LETTER BE

67 б → b ; # CYRILLIC SMALL LETTER BE

68 В → V ; # CYRILLIC CAPITAL LETTER VE

69 в → v ; # CYRILLIC SMALL LETTER VE

70 Г → G ; # CYRILLIC CAPITAL LETTER GHE

71 г → g ; # CYRILLIC SMALL LETTER GHE

72 Д → D ; # CYRILLIC CAPITAL LETTER DE

73 д → d ; # CYRILLIC SMALL LETTER DE

74 Е → E ; # CYRILLIC CAPITAL LETTER DE

75 е → e ; # CYRILLIC SMALL LETTER DE

76 Ж} $lower → Zh ; # CYRILLIC CAPITAL LETTER ZHE

77 Ж → ZH ; # CYRILLIC CAPITAL LETTER ZHE

78 ж → zh ; # CYRILLIC SMALL LETTER ZHE

79 З → Z ; # CYRILLIC CAPITAL LETTER ZE

80 з → z ; # CYRILLIC SMALL LETTER ZE

81 И → I ; # CYRILLIC CAPITAL LETTER I

82 и → i ; # CYRILLIC SMALL LETTER I

83 Й → Y ; # CYRILLIC CAPITAL LETTER I

84 й → y ; # CYRILLIC SMALL LETTER I

85 К → K ; # CYRILLIC CAPITAL LETTER KA

86 к → k ; # CYRILLIC SMALL LETTER KA

87 Л → L ; # CYRILLIC CAPITAL LETTER EL

88 л → l ; # CYRILLIC SMALL LETTER EL

89 М → M ; # CYRILLIC CAPITAL LETTER EM

90 м → m ; # CYRILLIC SMALL LETTER EM

91 Н → N ; # CYRILLIC CAPITAL LETTER EN

92 н → n ; # CYRILLIC SMALL LETTER EN

93 О → O ; # CYRILLIC CAPITAL LETTER O

94 о → o ; # CYRILLIC SMALL LETTER O

95 П → P ; # CYRILLIC CAPITAL LETTER PE

96 п → p ; # CYRILLIC SMALL LETTER PE

97 Р → R ; # CYRILLIC CAPITAL LETTER ER

98 р → r ; # CYRILLIC SMALL LETTER ER

99 С → S ; # CYRILLIC CAPITAL LETTER ES

100 с → s ; # CYRILLIC SMALL LETTER ES

101 #

102 #

103 ########################################################################

104 #

105 # BGN Page 16 Note 4

106 #

107 # тс becomes t·s

108 #

109 ########################################################################

110 #

111 ТС → T·S ; # CYRILLIC CAPITAL LETTER TE

112 Тс → T·s ; # CYRILLIC CAPITAL LETTER TE

113 тс → t·s ; # CYRILLIC SMALL LETTER TE

114 Т → T ; # CYRILLIC CAPITAL LETTER TE

115 т → t ; # CYRILLIC SMALL LETTER TE

116 #

117 #

118 ########################################################################

119 #

120 # End Note 4

121 #

122 ########################################################################

123 У → U ; # CYRILLIC CAPITAL LETTER U

124 у → u ; # CYRILLIC SMALL LETTER U

125 Ф → F ; # CYRILLIC CAPITAL LETTER EF

126 ф → f ; # CYRILLIC SMALL LETTER EF

127 Х} $lower → Kh ; # CYRILLIC CAPITAL LETTER HA

128 Х → KH ; # CYRILLIC CAPITAL LETTER HA

129 х → kh ; # CYRILLIC SMALL LETTER HA

130 Ц} $lower → Ts ; # CYRILLIC CAPITAL LETTER TSE

131 Ц → TS ; # CYRILLIC CAPITAL LETTER TSE

132 ц → ts ; # CYRILLIC SMALL LETTER TSE

133 Ч} $lower → Ch ; # CYRILLIC CAPITAL LETTER CHE

134 Ч → CH ; # CYRILLIC CAPITAL LETTER CHE

135 ч → ch ; # CYRILLIC SMALL LETTER CHE

136 #

137 ########################################################################

138 #

139 # Implied rule from BGN Russian-Latin transliteration (Page 94 Note 3.6).

140 #

141 # шт becomes sh·t

142 #

143 ########################################################################

144 #

145 ШТ → SH·T ; # CYRILLIC CAPITAL LETTER SHA

146 Шт → Sh·t ; # CYRILLIC CAPITAL LETTER SHA

147 шт → sh·t ; # CYRILLIC SMALL LETTER SHA

148 Ш} $lower → Sh ; # CYRILLIC CAPITAL LETTER SHA

149 Ш → SH ; # CYRILLIC CAPITAL LETTER SHA

150 ш → sh ; # CYRILLIC SMALL LETTER SHA

151 Щ} $lower → Sht ; # CYRILLIC CAPITAL LETTER SHCHA

152 Щ → SHT ; # CYRILLIC CAPITAL LETTER SHCHA

153 щ → sht ; # CYRILLIC SMALL LETTER SHCHA

154 #

155 #

156 ########################################################################

157 #

158 # End Implied rule

159 #

160 ########################################################################

161 Ъ → Ŭ ; # CYRILLIC CAPITAL LETTER HARD SIGN

162 ъ → ŭ ; # CYRILLIC SMALL LETTER HARD SIGN

163 #

164 ########################################################################

165 #

166 # BGN Page 16 Note 1

167 #

168 # In modern Bulgarian orthography, the character ъ does not occur in

169 # word-final position. It should be omitted in romanization when found

170 # on older sources.

171 #

172 # The following rule removes all Ъъ at the end of a word. It is assumed

173 # that when the condition is met, the text must be from an older source.

174 # Comment out with a '#' at the start of a line to disable.

175 #

176 #

177 ########################################################################

178 #

179 $bulgarian { [Ъъ] } $wordBoundary > ;

180 #

181 #

182 ########################################################################

183 #

184 # End BGN Page 16 Note 1

185 #

186 ########################################################################

187 Ь → ’ ; # CYRILLIC CAPITAL LETTER SOFT SIGN

188 ь → ’ ; # CYRILLIC SMALL LETTER SOFT SIGN

189 Ю} $lower → Yu ; # CYRILLIC CAPITAL LETTER YU

190 Ю → YU ; # CYRILLIC CAPITAL LETTER YU

191 ю → yu ; # CYRILLIC SMALL LETTER YU

192 Я} $lower → Ya ; # CYRILLIC CAPITAL LETTER YA

193 Я → YA ; # CYRILLIC CAPITAL LETTER YA

194 я → ya ; # CYRILLIC SMALL LETTER YA

195 #

196 ########################################################################

197 #

198 # BGN Page 16 Note 2

199 #

200 # The obsolete character Ѫ, which was replaced by Ъ in 1945, should be

201 # romanized Ŭ.

202 #

203 ########################################################################

204 #

205 Ѫ → Ŭ ; # CYRILLIC CAPITAL LETTER BIG YUS

206 ѫ → ŭ ; # CYRILLIC SMALL LETTER BIG YUS

207 #

208 #

209 ########################################################################

210 #

211 # End BGN Page 16 Note 2

212 #

213 ########################################################################

214 #

215 ########################################################################

216 #

217 # BGN Page 16 Note 3

218 #

219 # The obsolete character Ѣ, replaced in 1945 by Я or Е according to local

220 # pronunciation, should be romanized as e or ya, accordingly, if the

221 # pronunciation is known; otherwise as ye.

222 #

223 ########################################################################

224 #

225 Ѣ} $lower → Ye ; # CYRILLIC CAPITAL LETTER YAT

226 Ѣ → YE ; # CYRILLIC CAPITAL LETTER YAT

227 ѣ → ye ; # CYRILLIC SMALL LETTER YAT

228 #

229 #

230 # Alternative rule where appropriate for local pronounciation. To apply

231 # uncomment the following by removing the '#' mark at the start of the

232 # line and insert before the three rule lines above.

233 #

234 # Ѣ} $lower → e ; # CYRILLIC CAPITAL LETTER YAT

235 # Ѣ → E ; # CYRILLIC CAPITAL LETTER YAT

236 # ѣ → e ; # CYRILLIC SMALL LETTER YAT

237 #

238 ########################################################################

239 #

240 # End BGN Page 16 Note 3

241 #

242 ########################################################################

243