icuSources/data/translit/ps_ps_Latn_BGN.txt

   1 # ***************************************************************************
   2 # *
   3 # *  Copyright (C) 2004-2016, International Business Machines
   4 # *  Corporation; Unicode, Inc.; and others.  All Rights Reserved.
   5 # *
   6 # ***************************************************************************
   7 # File: ps_ps_Latn_BGN.txt
   8 # Generated from CLDR
   9 #
  10
  11 #
  12 ########################################################################
  13 # BGN/PCGN 1968 System
  14 #
  15 # This system was adopted in 1968 for the romanization of Pashto
  16 # geographic names in Afghanistan. Persian names in Afghanistan are
  17 # romanized in accordance with the Romanization System for Persian
  18 # (BGN/PCGN 1958 System), shown on pages 87-92).
  19 #
  20 # Originally prepared by Michael Everson <everson@evertype.com>
  21 ########################################################################
  22 #
  23 # MINIMAL FILTER: Pashto-Latin
  24 #
  25 :: [ءآابةتثجحخدذرزسشصضطظعغفقكلمنهوىي\u064E\u064F\u0650\u0651\u0652\u0654٠١٢٣٤٥٦٧٨٩ټپځڅچډړږژښگڰڼیۍې] ;
  26 :: NFD (NFC) ;
  27 #
  28 #
  29 ########################################################################
  30 #
  31 ########################################################################
  32 #
  33 # Define All Transformation Variables
  34 #
  35 ########################################################################
  36 #
  37 $alef = ’;
  38 $ayin = ‘;
  39 $disambig =  \u0331 ;
  40 #
  41 #
  42 # Use this $wordBoundary until bug 2034 is fixed in ICU:
  43 # http://bugs.icu-project.org/cgi-bin/icu-bugs/transliterate?id=2034;expression=boundary;user=guest
  44 #
  45 $wordBoundary =  [^[:L:][:M:][:N:]] ;
  46 #
  47 #
  48 ########################################################################
  49 [:Nd:]{٫}[:Nd:] ↔ [:Nd:]{','}[:Nd:] ; # ARABIC DECIMAL SEPARATOR
  50 [:Nd:]{٬}[:Nd:] ↔ [:Nd:]{'.'}[:Nd:] ; # ARABIC THOUSANDS SEPARATOR
  51 ٫ ↔ ',' $disambig ; # ARABIC DECIMAL SEPARATOR
  52 ٬ ↔ '.' $disambig ; # ARABIC THOUSANDS SEPARATOR
  53 #  ٭ ↔ ; # ARABIC FIVE POINTED STAR // no need to transliterate
  54 ، ↔ ',' ; # ARABIC COMMA
  55 ؛ ↔ ';' ; # ARABIC SEMICOLON
  56 ؟ ↔ '?' ; # ARABIC QUESTION MARK
  57 ٪ ↔ '%' ; # ARABIC PERCENT SIGN
  58 ٠ ↔ 0 $disambig ; # ARABIC-INDIC DIGIT ZERO
  59 ١ ↔ 1 $disambig ; # ARABIC-INDIC DIGIT ONE
  60 ٢ ↔ 2 $disambig ; # ARABIC-INDIC DIGIT TWO
  61 ٣ ↔ 3 $disambig ; # ARABIC-INDIC DIGIT THREE
  62 ٤ ↔ 4 $disambig ; # ARABIC-INDIC DIGIT FOUR
  63 ٥ ↔ 5 $disambig ; # ARABIC-INDIC DIGIT FIVE
  64 ٦ ↔ 6 $disambig ; # ARABIC-INDIC DIGIT SIX
  65 ٧ ↔ 7 $disambig ; # ARABIC-INDIC DIGIT SEVEN
  66 ٨ ↔ 8 $disambig ; # ARABIC-INDIC DIGIT EIGHT
  67 ٩ ↔ 9 $disambig ; # ARABIC-INDIC DIGIT NINE
  68 ۰ ↔ 0 ; # EXTENDED ARABIC-INDIC DIGIT ZERO
  69 ۱ ↔ 1 ; # EXTENDED ARABIC-INDIC DIGIT ONE
  70 ۲ ↔ 2 ; # EXTENDED ARABIC-INDIC DIGIT TWO
  71 ۳ ↔ 3 ; # EXTENDED ARABIC-INDIC DIGIT THREE
  72 ۴ ↔ 4 ; # EXTENDED ARABIC-INDIC DIGIT FOUR
  73 ۵ ↔ 5 ; # EXTENDED ARABIC-INDIC DIGIT FIVE
  74 ۶ ↔ 6 ; # EXTENDED ARABIC-INDIC DIGIT SIX
  75 ۷ ↔ 7 ; # EXTENDED ARABIC-INDIC DIGIT SEVEN
  76 ۸ ↔ 8 ; # EXTENDED ARABIC-INDIC DIGIT EIGHT
  77 ۹ ↔ 9 ; # EXTENDED ARABIC-INDIC DIGIT NINE
  78 #
  79 ########################################################################
  80 #
  81 # Rules moved to front to avoid masking
  82 #
  83 ########################################################################
  84 #
  85 ########################################################################
  86 #
  87 # BGN Page 89 Rule 4
  88 #
  89 # The character sequences كه , زه , سه , and گه may be romanized k·h, z·h,
  90 # s·h, and g·h in order to differentiate those romanizations from the
  91 # digraphs kh, zh, sh, and gh.
  92 #
  93 ########################################################################
  94 #
  95 كه → k·h ; # ARABIC LETTER KAF + HEH
  96 زه → z·h ; # ARABIC LETTER ZAIN + HEH
  97 سه → s·h ; # ARABIC LETTER SEEN + HEH
  98 گه → g·h ; # ARABIC LETTER GAF + HEH
  99 #
 100 #
 101 ########################################################################
 102 #
 103 # End Rule 4
 104 #
 105 ########################################################################
 106 #
 107 ########################################################################
 108 #
 109 # BGN Page 91 Rule 7
 110 #
 111 # Doubles consonant sounds are represented in Arabic script by
 112 # placing a shaddah ( \u0651 ) over a consonant character. In romanization
 113 # the letter should be doubled. [The remainder of this rule deals with
 114 # the definite article and is lexical.]
 115 #
 116 ########################################################################
 117 #
 118 ب\u0651 → bb ; # ARABIC LETTER BEH + SHADDA
 119 پ\u0651 → pp ; # ARABIC LETTER PEH + SHADDA
 120 ت\u0651 → tt ; # ARABIC LETTER TEH + SHADDA
 121 ټ\u0651 → ṯṯ ; # ARABIC LETTER TEH WITH RING + SHADDA
 122 ث\u0651 → s\u0304s\u0304 ; # ARABIC LETTER THEH + SHADDA
 123 ج\u0651 → jj ; # ARABIC LETTER JEEM + SHADDA
 124 چ\u0651 → chch ; # ARABIC LETTER TCHEH + SHADDA
 125 \u0651څ → tsts ; # ARABIC LETTER HAH WITH THREE DOTS ABOVE + SHADDA
 126 \u0651ځ → dz ; # ARABIC LETTER HAH WITH HAMZA ABOVE + SHADDA
 127 ح\u0651 → ḥḥ ; # ARABIC LETTER HAH + SHADDA
 128 خ\u0651 → khkh ; # ARABIC LETTER KHAH + SHADDA
 129 د\u0651 → dd ; # ARABIC LETTER DAL + SHADDA
 130 \u0651ډ → ḏḏ ; # ARABIC LETTER DAL WITH RING + SHADDA
 131 ذ\u0651 → z\u0304z\u0304 ; # ARABIC LETTER THAL + SHADDA
 132 ر\u0651 → rr ; # ARABIC LETTER REH + SHADDA
 133 \u0651ړ → ṟṟ ; # ARABIC LETTER REH WITH RING + SHADDA
 134 ز\u0651 → zz ; # ARABIC LETTER ZAIN + SHADDA
 135 ژ\u0651 → zhzh ; # ARABIC LETTER JEH + SHADDA
 136 \u0651ژ → z\u035Fhz\u035Fh ; # ARABIC LETTER REH WITH DOT BELOW AND DOT ABOVE + SHADDA
 137 س\u0651 → ss ; # ARABIC LETTER SEEN + SHADDA
 138 ش\u0651 → shsh ; # ARABIC LETTER SHEEN + SHADDA
 139 \u0651ښ → s\u035Fhs\u035Fh ; # ARABIC LETTER SEEN WITH DOT BELOW AND DOT ABOVE
 140 ص\u0651 → ṣṣ ; # ARABIC LETTER SAD + SHADDA
 141 ض\u0651 → ḍḍ ; # ARABIC LETTER DAD + SHADDA
 142 ط\u0651 → ṭṭ ; # ARABIC LETTER TAH + SHADDA
 143 ظ\u0651 → ẓẓ ; # ARABIC LETTER ZAH + SHADDA
 144 ع\u0651 → $ayin $ayin ; # ARABIC LETTER AIN + SHADDA
 145 غ\u0651 → ghgh ; # ARABIC LETTER GHAIN + SHADDA
 146 ف\u0651 → ff ; # ARABIC LETTER FEH + SHADDA
 147 ق\u0651 → qq ; # ARABIC LETTER QAF + SHADDA
 148 ك\u0651 → kk ; # ARABIC LETTER KAF + SHADDA
 149 \u0651گ → gg ; # ARABIC LETTER GAF + SHADDA
 150 \u0651ڰ → gg ; # ARABIC LETTER GAF WITH RING + SHADDA
 151 ل\u0651 → ll ; # ARABIC LETTER LAM + SHADDA
 152 م\u0651 → mm ; # ARABIC LETTER MEEM + SHADDA
 153 ن\u0651 → nn ; # ARABIC LETTER NOON + SHADDA
 154 \u0651ڼ → ṉṉ ; # ARABIC LETTER NOON WITH RING + SHADDA
 155 ه\u0651 → hh ; # ARABIC LETTER HEH + SHADDA
 156 و\u0651 → ww ; # ARABIC LETTER WAW + SHADDA
 157 \u0651ی → yy ; # ARABIC LETTER FARSI YEH + SHADDA
 158 ى\u0651 → yy ; # ARABIC LETTER YEH + SHADDA
 159 #
 160 #
 161 ########################################################################
 162 #
 163 # End Rule 7
 164 #
 165 ########################################################################
 166 #
 167 ########################################################################
 168 #
 169 # Start of Transformations
 170 #
 171 ########################################################################
 172 #
 173 $wordBoundary{ء →  ; # ARABIC LETTER HAMZA
 174 ء → $alef ; # ARABIC LETTER HAMZA
 175 $wordBoundary{ا →  ; # ARABIC LETTER ALEF
 176 آ → $alef ā ; # ARABIC LETTER ALEF WITH MADDA ABOVE
 177 ب → b ; # ARABIC LETTER BEH
 178 پ → p ; # ARABIC LETTER PEH
 179 ت → t ; # ARABIC LETTER TEH
 180 ټ → ṯ ; # ARABIC LETTER TEH WITH RING
 181 ة → h ; # ARABIC LETTER TEH MARBUTA
 182 ث → s\u0304 ; # ARABIC LETTER THEH
 183 ج → j ; # ARABIC LETTER JEEM
 184 چ → ch ; # ARABIC LETTER TCHEH
 185 څ → ts ; # ARABIC LETTER HAH WITH THREE DOTS ABOVE
 186 ځ → dz ; # ARABIC LETTER HAH WITH HAMZA ABOVE
 187 ح → ḥ ; # ARABIC LETTER HAH
 188 خ → kh ; # ARABIC LETTER KHAH
 189 د → d ; # ARABIC LETTER DAL
 190 ډ → ḏ ; # ARABIC LETTER DAL WITH RING
 191 ذ → z\u0304 ; # ARABIC LETTER THAL
 192 ر → r ; # ARABIC LETTER REH
 193 ړ → ṟ ; # ARABIC LETTER REH WITH RING
 194 ز → z ; # ARABIC LETTER ZAIN
 195 ژ → zh ; # ARABIC LETTER JEH
 196 ږ → z\u035Fh ; # ARABIC LETTER REH WITH DOT BELOW AND DOT ABOVE
 197 س → s ; # ARABIC LETTER SEEN
 198 ش → sh ; # ARABIC LETTER SHEEN
 199 ښ → s\u035Fh ; # ARABIC LETTER SEEN WITH DOT BELOW AND DOT ABOVE
 200 ص → ṣ ; # ARABIC LETTER SAD
 201 ض → ẕ ; # ARABIC LETTER DAD
 202 ط → ṭ ; # ARABIC LETTER TAH
 203 ظ → ẓ ; # ARABIC LETTER ZAH
 204 ع → $ayin ; # ARABIC LETTER AIN
 205 غ → gh ; # ARABIC LETTER GHAIN
 206 ف → f ; # ARABIC LETTER FEH
 207 ق → q ; # ARABIC LETTER QAF
 208 ك → k ; # ARABIC LETTER KAF
 209 گ → g ; # ARABIC LETTER GAF
 210 ڰ → g ; # ARABIC LETTER GAF WITH RING
 211 ل → l ; # ARABIC LETTER LAM
 212 م → m ; # ARABIC LETTER MEEM
 213 ن → n ; # ARABIC LETTER NOON
 214 ڼ → ṉ ; # ARABIC LETTER NOON WITH RING
 215 و → w ; # ARABIC LETTER WAW
 216 ه → h ; # ARABIC LETTER HEH
 217 \u0654ی → ey ; # ARABIC LETTER FARSI YEH + HAMZA ABOVE
 218 ی → y ; # ARABIC LETTER FARSI YEH
 219 ى → y ; # ARABIC LETTER YEH
 220 ې → e ; # ARABIC LETTER E
 221 \u064Eا → ā ; # ARABIC FATHA + ALEF
 222 \u064Eى\u0652 → ay ; # ARABIC FATHA + FARSI YEH + SUKUN
 223 \u064Eى → á ; # ARABIC FATHA + ALEF MAKSURA
 224 \u064E\u0652ۍ → êy ; # ARABIC FATHA + SUKUN + YEH WITH TAIL
 225 \u064E\u0652 → ê ; # ARABIC FATHA + SUKUN
 226 \u064E → a ; # ARABIC FATHA
 227 \u0650\u0652ي → ey ; # ARABIC KASRA + FARSI YEH + SUKUN
 228 \u0650ي → ī ; # ARABIC KASRA + FARSI YEH
 229 \u0650ى → ī ; # ARABIC KASRA + YEH
 230 \u0650و → ew ; # ARABIC KASRA + WAW
 231 \u0650 → i ; # ARABIC KASRA
 232 \u064Fو\u0652 → ow ; # ARABIC DAMMA + WAW + SUKUN
 233 \u064Fوی → ūy ; # ARABIC DAMMA + WAW + FARSI YEH
 234 \u064Fو → ū ; # ARABIC DAMMA + WAW
 235 \u064F → u ; # ARABIC DAMMA
 236 \u0652 →  ; # ARABIC SUKUN
 237 #
 238 #
 239 ########################################################################
 240