1 <?xml version="1.0" encoding="UTF-8" ?> 2 <!DOCTYPE supplementalData SYSTEM "../../common/dtd/ldmlSupplemental.dtd"> 3 <!-- 4 Copyright 1991-2013 Unicode, Inc. 5 CLDR data files are interpreted according to the LDML specification (http://unicode.org/reports/tr35/) 6 For terms of use, see http://www.unicode.org/copyright.html 7 --> 8 <supplementalData> 9 <version number="$Revision: 13787 $"/> 10 <transforms> 11 <transform source="bg" target="bg_Latn" variant="BGN" direction="forward" draft="contributed" alias="Bulgarian-Latin/BGN bg-Latn-t-bg-m0-bgn"> 12 <tRule><![CDATA[ 13 # 14 ######################################################################## 15 # BGN/PCGN 1952 System 16 # 17 # This system was adopted by the BGN in 1949 and by the PCGN in 1952. 18 # It reflects the much simplified Bulgarian orthography as officially 19 # revised in February 1945. The Bulgarian alphabet contains all of 20 # the characters present in the Russian alphabet with the exception 21 # of , , and . Two obsolete letters and are also given. 22 # 23 # The Bulgarian Alphabet as defined by the BGN (Page 15): 24 # 25 # 26 # 27 # 28 # Originally prepared by Michael Everson <everson (a] evertype.com> 29 ######################################################################## 30 # 31 # MINIMAL FILTER: Bulgarian-Latin 32 # 33 34 :: [] ; 35 :: NFD (NFC) ; 36 # 37 # 38 ######################################################################## 39 40 41 # 42 ######################################################################## 43 # 44 # Define All Transformation Variables 45 # 46 ######################################################################## 47 # 48 49 $upperConsonants = [] ; 50 $lowerConsonants = [] ; 51 $consonants = [$upperConsonants $lowerConsonants] ; 52 53 $upperVowels = [] ; 54 $lowerVowels = [] ; 55 $vowels = [$upperVowels $lowerVowels] ; 56 57 $lower = [$lowerConsonants $lowerVowels] ; 58 59 $bulgarian = [ $lower $upperConsonants $upperVowels ] ; 60 # 61 # 62 # Use this $wordBoundary until bug 2034 is fixed in ICU: 63 # http://bugs.icu-project.org/cgi-bin/icu-bugs/transliterate?id=2034;expression=boundary;user=guest 64 # 65 66 $wordBoundary = [^[:L:][:M:][:N:]] ; 67 # 68 # 69 ######################################################################## 70 71 72 # 73 ######################################################################## 74 # 75 # Start of Alphabetic Transformations 76 # 77 ######################################################################## 78 # 79 80 A ; # CYRILLIC CAPITAL LETTER A 81 a ; # CYRILLIC SMALL LETTER A 82 B ; # CYRILLIC CAPITAL LETTER BE 83 b ; # CYRILLIC SMALL LETTER BE 84 V ; # CYRILLIC CAPITAL LETTER VE 85 v ; # CYRILLIC SMALL LETTER VE 86 G ; # CYRILLIC CAPITAL LETTER GHE 87 g ; # CYRILLIC SMALL LETTER GHE 88 D ; # CYRILLIC CAPITAL LETTER DE 89 d ; # CYRILLIC SMALL LETTER DE 90 E ; # CYRILLIC CAPITAL LETTER DE 91 e ; # CYRILLIC SMALL LETTER DE 92 } $lower Zh ; # CYRILLIC CAPITAL LETTER ZHE 93 ZH ; # CYRILLIC CAPITAL LETTER ZHE 94 zh ; # CYRILLIC SMALL LETTER ZHE 95 Z ; # CYRILLIC CAPITAL LETTER ZE 96 z ; # CYRILLIC SMALL LETTER ZE 97 I ; # CYRILLIC CAPITAL LETTER I 98 i ; # CYRILLIC SMALL LETTER I 99 Y ; # CYRILLIC CAPITAL LETTER I 100 y ; # CYRILLIC SMALL LETTER I 101 K ; # CYRILLIC CAPITAL LETTER KA 102 k ; # CYRILLIC SMALL LETTER KA 103 L ; # CYRILLIC CAPITAL LETTER EL 104 l ; # CYRILLIC SMALL LETTER EL 105 M ; # CYRILLIC CAPITAL LETTER EM 106 m ; # CYRILLIC SMALL LETTER EM 107 N ; # CYRILLIC CAPITAL LETTER EN 108 n ; # CYRILLIC SMALL LETTER EN 109 O ; # CYRILLIC CAPITAL LETTER O 110 o ; # CYRILLIC SMALL LETTER O 111 P ; # CYRILLIC CAPITAL LETTER PE 112 p ; # CYRILLIC SMALL LETTER PE 113 R ; # CYRILLIC CAPITAL LETTER ER 114 r ; # CYRILLIC SMALL LETTER ER 115 S ; # CYRILLIC CAPITAL LETTER ES 116 s ; # CYRILLIC SMALL LETTER ES 117 # 118 # 119 ######################################################################## 120 # 121 # BGN Page 16 Note 4 122 # 123 # becomes ts 124 # 125 ######################################################################## 126 # 127 128 TS ; # CYRILLIC CAPITAL LETTER TE 129 Ts ; # CYRILLIC CAPITAL LETTER TE 130 ts ; # CYRILLIC SMALL LETTER TE 131 T ; # CYRILLIC CAPITAL LETTER TE 132 t ; # CYRILLIC SMALL LETTER TE 133 # 134 # 135 ######################################################################## 136 # 137 # End Note 4 138 # 139 ######################################################################## 140 141 142 U ; # CYRILLIC CAPITAL LETTER U 143 u ; # CYRILLIC SMALL LETTER U 144 F ; # CYRILLIC CAPITAL LETTER EF 145 f ; # CYRILLIC SMALL LETTER EF 146 } $lower Kh ; # CYRILLIC CAPITAL LETTER HA 147 KH ; # CYRILLIC CAPITAL LETTER HA 148 kh ; # CYRILLIC SMALL LETTER HA 149 } $lower Ts ; # CYRILLIC CAPITAL LETTER TSE 150 TS ; # CYRILLIC CAPITAL LETTER TSE 151 ts ; # CYRILLIC SMALL LETTER TSE 152 } $lower Ch ; # CYRILLIC CAPITAL LETTER CHE 153 CH ; # CYRILLIC CAPITAL LETTER CHE 154 ch ; # CYRILLIC SMALL LETTER CHE 155 156 # 157 ######################################################################## 158 # 159 # Implied rule from BGN Russian-Latin transliteration (Page 94 Note 3.6). 160 # 161 # becomes sht 162 # 163 ######################################################################## 164 # 165 166 SHT ; # CYRILLIC CAPITAL LETTER SHA 167 Sht ; # CYRILLIC CAPITAL LETTER SHA 168 sht ; # CYRILLIC SMALL LETTER SHA 169 } $lower Sh ; # CYRILLIC CAPITAL LETTER SHA 170 SH ; # CYRILLIC CAPITAL LETTER SHA 171 sh ; # CYRILLIC SMALL LETTER SHA 172 } $lower Sht ; # CYRILLIC CAPITAL LETTER SHCHA 173 SHT ; # CYRILLIC CAPITAL LETTER SHCHA 174 sht ; # CYRILLIC SMALL LETTER SHCHA 175 # 176 # 177 ######################################################################## 178 # 179 # End Implied rule 180 # 181 ######################################################################## 182 183 184 ; # CYRILLIC CAPITAL LETTER HARD SIGN 185 ; # CYRILLIC SMALL LETTER HARD SIGN 186 187 # 188 ######################################################################## 189 # 190 # BGN Page 16 Note 1 191 # 192 # In modern Bulgarian orthography, the character does not occur in 193 # word-final position. It should be omitted in romanization when found 194 # on older sources. 195 # 196 # The following rule removes all at the end of a word. It is assumed 197 # that when the condition is met, the text must be from an older source. 198 # Comment out with a '#' at the start of a line to disable. 199 # 200 # 201 ######################################################################## 202 # 203 204 $bulgarian { [] } $wordBoundary > ; 205 # 206 # 207 ######################################################################## 208 # 209 # End BGN Page 16 Note 1 210 # 211 ######################################################################## 212 213 214 ; # CYRILLIC CAPITAL LETTER SOFT SIGN 215 ; # CYRILLIC SMALL LETTER SOFT SIGN 216 } $lower Yu ; # CYRILLIC CAPITAL LETTER YU 217 YU ; # CYRILLIC CAPITAL LETTER YU 218 yu ; # CYRILLIC SMALL LETTER YU 219 } $lower Ya ; # CYRILLIC CAPITAL LETTER YA 220 YA ; # CYRILLIC CAPITAL LETTER YA 221 ya ; # CYRILLIC SMALL LETTER YA 222 223 # 224 ######################################################################## 225 # 226 # BGN Page 16 Note 2 227 # 228 # The obsolete character , which was replaced by in 1945, should be 229 # romanized . 230 # 231 ######################################################################## 232 # 233 234 ; # CYRILLIC CAPITAL LETTER BIG YUS 235 ; # CYRILLIC SMALL LETTER BIG YUS 236 # 237 # 238 ######################################################################## 239 # 240 # End BGN Page 16 Note 2 241 # 242 ######################################################################## 243 244 245 # 246 ######################################################################## 247 # 248 # BGN Page 16 Note 3 249 # 250 # The obsolete character , replaced in 1945 by or according to local 251 # pronunciation, should be romanized as e or ya, accordingly, if the 252 # pronunciation is known; otherwise as ye. 253 # 254 ######################################################################## 255 # 256 257 } $lower Ye ; # CYRILLIC CAPITAL LETTER YAT 258 YE ; # CYRILLIC CAPITAL LETTER YAT 259 ye ; # CYRILLIC SMALL LETTER YAT 260 # 261 # 262 # Alternative rule where appropriate for local pronounciation. To apply 263 # uncomment the following by removing the '#' mark at the start of the 264 # line and insert before the three rule lines above. 265 # 266 # } $lower e ; # CYRILLIC CAPITAL LETTER YAT 267 # E ; # CYRILLIC CAPITAL LETTER YAT 268 # e ; # CYRILLIC SMALL LETTER YAT 269 # 270 ######################################################################## 271 # 272 # End BGN Page 16 Note 3 273 # 274 ######################################################################## 275 276 ]]></tRule> 277 </transform> 278 </transforms> 279 </supplementalData> 280
