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="fa" target="fa_Latn" variant="BGN" direction="forward" draft="contributed" alias="Persian-Latin/BGN fa-Latn-t-fa-m0-bgn"> 12 <tRule><![CDATA[ 13 # 14 ######################################################################## 15 # BGN/PCGN 1956 System 16 # 17 # This system was adopted by the BGN in 1946 and by the PCGN in 1958. 18 # It is used for the romanization of geographic names in Iran and 19 # for Persian-language names in Afghanistan. 20 # 21 # Originally prepared by Michael Everson <everson (a] evertype.com> 22 ######################################################################## 23 # 24 # MINIMAL FILTER: Persian-Latin 25 # 26 27 :: [[:arabic:][:block=ARABIC:][]] ; 28 :: NFKD (NFC) ; 29 # 30 # 31 ######################################################################## 32 33 34 # 35 ######################################################################## 36 # 37 # Define All Transformation Variables 38 # 39 ######################################################################## 40 # 41 42 $alef = ; 43 $ayin = ; 44 $disambig = ; 45 # 46 # 47 # Use this $wordBoundary until bug 2034 is fixed in ICU: 48 # http://bugs.icu-project.org/cgi-bin/icu-bugs/transliterate?id=2034;expression=boundary;user=guest 49 # 50 51 $wordBoundary = [^[:L:][:M:][:N:]] ; 52 # 53 # 54 ######################################################################## 55 56 # non-letters 57 [:Nd:]{}[:Nd:] [:Nd:]{','}[:Nd:] ; # ARABIC DECIMAL SEPARATOR 58 [:Nd:]{}[:Nd:] [:Nd:]{'.'}[:Nd:] ; # ARABIC THOUSANDS SEPARATOR 59 ',' $disambig ; # ARABIC DECIMAL SEPARATOR 60 '.' $disambig ; # ARABIC THOUSANDS SEPARATOR 61 # ; # ARABIC FIVE POINTED STAR // no need to transliterate 62 ',' ; # ARABIC COMMA 63 ';' ; # ARABIC SEMICOLON 64 '?' ; # ARABIC QUESTION MARK 65 '%' ; # ARABIC PERCENT SIGN 66 0 $disambig ; # ARABIC-INDIC DIGIT ZERO 67 1 $disambig ; # ARABIC-INDIC DIGIT ONE 68 2 $disambig ; # ARABIC-INDIC DIGIT TWO 69 3 $disambig ; # ARABIC-INDIC DIGIT THREE 70 4 $disambig ; # ARABIC-INDIC DIGIT FOUR 71 5 $disambig ; # ARABIC-INDIC DIGIT FIVE 72 6 $disambig ; # ARABIC-INDIC DIGIT SIX 73 7 $disambig ; # ARABIC-INDIC DIGIT SEVEN 74 8 $disambig ; # ARABIC-INDIC DIGIT EIGHT 75 9 $disambig ; # ARABIC-INDIC DIGIT NINE 76 0 ; # EXTENDED ARABIC-INDIC DIGIT ZERO 77 1 ; # EXTENDED ARABIC-INDIC DIGIT ONE 78 2 ; # EXTENDED ARABIC-INDIC DIGIT TWO 79 3 ; # EXTENDED ARABIC-INDIC DIGIT THREE 80 4 ; # EXTENDED ARABIC-INDIC DIGIT FOUR 81 5 ; # EXTENDED ARABIC-INDIC DIGIT FIVE 82 6 ; # EXTENDED ARABIC-INDIC DIGIT SIX 83 7 ; # EXTENDED ARABIC-INDIC DIGIT SEVEN 84 8 ; # EXTENDED ARABIC-INDIC DIGIT EIGHT 85 9 ; # EXTENDED ARABIC-INDIC DIGIT NINE 86 87 # 88 ######################################################################## 89 # 90 # Rules moved to front to avoid masking 91 # 92 ######################################################################## 93 94 95 # 96 ######################################################################## 97 # 98 # BGN Page 89 Rule 4 99 # 100 # The character sequences , , , and may be romanized kh, zh, 101 # sh, and gh in order to differentiate those romanizations from the 102 # digraphs kh, zh, sh, and gh. 103 # 104 ######################################################################## 105 # 106 107 kh ; # ARABIC LETTER KAF + HEH 108 zh ; # ARABIC LETTER ZAIN + HEH 109 sh ; # ARABIC LETTER SEEN + HEH 110 gh ; # ARABIC LETTER GAF + HEH 111 # 112 # 113 ######################################################################## 114 # 115 # End Rule 4 116 # 117 ######################################################################## 118 119 120 121 # 122 ######################################################################## 123 # 124 # BGN Page 91 Rule 7 125 # 126 # Doubles consonant sounds are represented in Arabic script by 127 # placing a shaddah ( ) over a consonant character. In romanization 128 # the letter should be doubled. [The remainder of this rule deals with 129 # the definite article and is lexical.] 130 # 131 ######################################################################## 132 # 133 134 bb ; # ARABIC LETTER BEH + SHADDA 135 pp ; # ARABIC LETTER PEH + SHADDA 136 tt ; # ARABIC LETTER TEH + SHADDA 137 ss ; # ARABIC LETTER THEH + SHADDA 138 jj ; # ARABIC LETTER JEEM + SHADDA 139 chch ; # ARABIC LETTER TCHEH + SHADDA 140 ; # ARABIC LETTER HAH + SHADDA 141 khkh ; # ARABIC LETTER KHAH + SHADDA 142 dd ; # ARABIC LETTER DAL + SHADDA 143 zz ; # ARABIC LETTER THAL + SHADDA 144 rr ; # ARABIC LETTER REH + SHADDA 145 zz ; # ARABIC LETTER ZAIN + SHADDA 146 zhzh ; # ARABIC LETTER JEH + SHADDA 147 ss ; # ARABIC LETTER SEEN + SHADDA 148 shsh ; # ARABIC LETTER SHEEN + SHADDA 149 ; # ARABIC LETTER SAD + SHADDA 150 ; # ARABIC LETTER DAD + SHADDA 151 ; # ARABIC LETTER TAH + SHADDA 152 ; # ARABIC LETTER ZAH + SHADDA 153 $ayin $ayin ; # ARABIC LETTER AIN + SHADDA 154 ghgh ; # ARABIC LETTER GHAIN + SHADDA 155 ff ; # ARABIC LETTER FEH + SHADDA 156 qq ; # ARABIC LETTER QAF + SHADDA 157 kk ; # ARABIC LETTER KAF + SHADDA 158 ll ; # ARABIC LETTER LAM + SHADDA 159 mm ; # ARABIC LETTER MEEM + SHADDA 160 nn ; # ARABIC LETTER NOON + SHADDA 161 hh ; # ARABIC LETTER HEH + SHADDA 162 ww ; # ARABIC LETTER WAW + SHADDA 163 yy ; # ARABIC LETTER FARSI YEH + SHADDA 164 # 165 # 166 ######################################################################## 167 # 168 # End Rule 7 169 # 170 ######################################################################## 171 172 173 174 # 175 ######################################################################## 176 # 177 # Start of Transformations 178 # 179 ######################################################################## 180 # 181 182 $wordBoundary{ ; # ARABIC LETTER HAMZA 183 $alef ; # ARABIC LETTER HAMZA 184 $wordBoundary{ ; # ARABIC LETTER ALEF 185 $alef ; # ARABIC FATHA ALEF WITH MADDA ABOVE 186 b ; # ARABIC LETTER BEH 187 p ; # ARABIC LETTER PEH 188 t ; # ARABIC LETTER TEH 189 h ; # ARABIC LETTER TEH MARBUTA 190 s ; # ARABIC LETTER THEH 191 j ; # ARABIC LETTER JEEM 192 ch ; # ARABIC LETTER TCHEH 193 ; # ARABIC LETTER HAH 194 kh ; # ARABIC LETTER KHAH 195 d ; # ARABIC LETTER DAL 196 z ; # ARABIC LETTER THAL 197 r ; # ARABIC LETTER REH 198 z ; # ARABIC LETTER ZAIN 199 zh ; # ARABIC LETTER JEH 200 s ; # ARABIC LETTER SEEN 201 sh ; # ARABIC LETTER SHEEN 202 ; # ARABIC LETTER SAD 203 ; # ARABIC LETTER DAD 204 ; # ARABIC LETTER TAH 205 ; # ARABIC LETTER ZAH 206 $ayin ; # ARABIC LETTER AIN 207 gh ; # ARABIC LETTER GHAIN 208 f ; # ARABIC LETTER FEH 209 q ; # ARABIC LETTER QAF 210 k ; # ARABIC LETTER KEHEH 211 k $disambig ; # ARABIC LETTER KAF 212 g ; # ARABIC LETTER GAF 213 l ; # ARABIC LETTER LAM 214 m ; # ARABIC LETTER MEEM 215 n ; # ARABIC LETTER NOON 216 h ; # ARABIC LETTER HEH 217 v ; # ARABIC LETTER WAW 218 y ; # ARABIC LETTER FARSI YEH 219 220 ; # ARABIC FATHA + ALEF 221 ; # ARABIC FATHA + FARSI YEH 222 ow ; # ARABIC FATHA + WAW + SUKUN 223 a ; # ARABIC FATHA 224 225 ; # ARABIC KASRA + YEH 226 e ; # ARABIC KASRA 227 228 ; # ARABIC DAMMA + WAW 229 o ; # ARABIC DAMMA 230 231 ; # ARABIC SUKUN 232 ::NFC (NFD) ; 233 234 # 235 # 236 ######################################################################## 237 238 ]]></tRule> 239 </transform> 240 </transforms> 241 </supplementalData> 242
