icuSources/data/translit/ar_ar_Latn_BGN.txt

   1 # © 2016 and later: Unicode, Inc. and others.
   2 # License & terms of use: http://www.unicode.org/copyright.html#License
   3 #
   4 # File: ar_ar_Latn_BGN.txt
   5 # Generated from CLDR
   6 #
   7
   8 #
   9 ########################################################################
  10 # BGN/PCGN 1956 System
  11 #
  12 # This system was adopted by the BGN in 1946 and by the PCGN
  13 # in 1956 and has been applied in the systematic romanization
  14 # of geographic names in Bahrain, Egypt, Iraq, Jordan,
  15 # Kuwait, Lebanon, Libya, Oman, Qatar, Saudi Arabia, Sudan,
  16 # Syria, Tunisia, the United Arab Emirates, and Yemen, all
  17 # of which has been covered by published BGN engineers.
  18 #
  19 # Originally prepared by Michael Everson <everson@evertype.com>
  20 ########################################################################
  21 #
  22 # MINIMAL FILTER: Arabic-Latin
  23 #
  24 :: [[:arabic:][:block=ARABIC:][ءآابةتثجحخدذرزسشصضطظعغفقكلمنهوىي\u064B\u064C\u064D\u064E\u064F\u0650\u0651\u0652٠١٢٣٤٥٦٧٨٩ٱ]] ;
  25 :: NFKD (NFC) ;
  26 #
  27 #
  28 ########################################################################
  29 #
  30 ########################################################################
  31 #
  32 # Define All Transformation Variables
  33 #
  34 ########################################################################
  35 #
  36 $alef = ’;
  37 $ayin = ‘;
  38 $disambig =  \u0331 ;
  39 #
  40 #
  41 # Use this $wordBoundary until bug 2034 is fixed in ICU:
  42 # http://bugs.icu-project.org/cgi-bin/icu-bugs/transliterate?id=2034;expression=boundary;user=guest
  43 #
  44 $wordBoundary =  [^[:L:][:M:][:N:]] ;
  45 #
  46 #
  47 ########################################################################
  48 # non-letters
  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 ۰ ↔ 0 $disambig ; # EXTENDED ARABIC-INDIC DIGIT ZERO
  59 ۱ ↔ 1 $disambig ; # EXTENDED ARABIC-INDIC DIGIT ONE
  60 ۲ ↔ 2 $disambig ; # EXTENDED ARABIC-INDIC DIGIT TWO
  61 ۳ ↔ 3 $disambig ; # EXTENDED ARABIC-INDIC DIGIT THREE
  62 ۴ ↔ 4 $disambig ; # EXTENDED ARABIC-INDIC DIGIT FOUR
  63 ۵ ↔ 5 $disambig ; # EXTENDED ARABIC-INDIC DIGIT FIVE
  64 ۶ ↔ 6 $disambig ; # EXTENDED ARABIC-INDIC DIGIT SIX
  65 ۷ ↔ 7 $disambig ; # EXTENDED ARABIC-INDIC DIGIT SEVEN
  66 ۸ ↔ 8 $disambig ; # EXTENDED ARABIC-INDIC DIGIT EIGHT
  67 ۹ ↔ 9 $disambig ; # EXTENDED ARABIC-INDIC DIGIT NINE
  68 ٠ ↔ 0 ; # ARABIC-INDIC DIGIT ZERO
  69 ١ ↔ 1 ; # ARABIC-INDIC DIGIT ONE
  70 ٢ ↔ 2 ; # ARABIC-INDIC DIGIT TWO
  71 ٣ ↔ 3 ; # ARABIC-INDIC DIGIT THREE
  72 ٤ ↔ 4 ; # ARABIC-INDIC DIGIT FOUR
  73 ٥ ↔ 5 ; # ARABIC-INDIC DIGIT FIVE
  74 ٦ ↔ 6 ; # ARABIC-INDIC DIGIT SIX
  75 ٧ ↔ 7 ; # ARABIC-INDIC DIGIT SEVEN
  76 ٨ ↔ 8 ; # ARABIC-INDIC DIGIT EIGHT
  77 ٩ ↔ 9 ; # ARABIC-INDIC DIGIT NINE
  78 #
  79 ########################################################################
  80 #
  81 # Rules moved to front to avoid masking
  82 #
  83 ########################################################################
  84 #
  85 ########################################################################
  86 #
  87 # BGN Page 8 Rule 5
  88 #
  89 # The character sequences ت , كه , ته , and سه may be romanized t·h, k·h,
  90 # d·h, and s·h in order to differentiate those romanizations from the
  91 # digraphs th, kh, dh, and sh.
  92 #
  93 ########################################################################
  94 #
  95 ته → t·h ; # ARABIC LETTER TEH + HEH
  96 كه → k·h ; # ARABIC LETTER KAF + HEH
  97 ده → d·h ; # ARABIC LETTER DAL + HEH
  98 سه → s·h ; # ARABIC LETTER SEEN + HEH
  99 #
 100 #
 101 ########################################################################
 102 #
 103 # End Rule 5
 104 #
 105 ########################################################################
 106 ########################################################################
 107 #
 108 #
 109 # BGN Page 8 Rule 9
 110 #
 111 # Doubles consonant sounds are represented in Arabic script by placing
 112 # a shaddah ( \u0651 ) over a consonant character. In romanization the letter
 113 # should be doubled. [The remainder of this rule deals with the definite
 114 # article and is lexical.]
 115 #
 116 ########################################################################
 117 #
 118 ب\u0651 → bb ; # ARABIC LETTER BEH + SHADDA
 119 ت\u0651 → tt ; # ARABIC LETTER TEH + SHADDA
 120 ث\u0651 → thth ; # ARABIC LETTER THEH + SHADDA
 121 ج\u0651 → jj ; # ARABIC LETTER JEEM + SHADDA
 122 ح\u0651 → ḥḥ ; # ARABIC LETTER HAH + SHADDA
 123 خ\u0651 → khkh ; # ARABIC LETTER KHAH + SHADDA
 124 د\u0651 → dd ; # ARABIC LETTER DAL + SHADDA
 125 ذ\u0651 → dhdh ; # ARABIC LETTER THAL + SHADDA
 126 ر\u0651 → rr ; # ARABIC LETTER REH + SHADDA
 127 ز\u0651 → zz ; # ARABIC LETTER ZAIN + SHADDA
 128 س\u0651 → ss ; # ARABIC LETTER SEEN + SHADDA
 129 ش\u0651 → shsh ; # ARABIC LETTER SHEEN + SHADDA
 130 ص\u0651 → ṣṣ ; # ARABIC LETTER SAD + SHADDA
 131 ض\u0651 → ḍḍ ; # ARABIC LETTER DAD + SHADDA
 132 ط\u0651 → ṭṭ ; # ARABIC LETTER TAH + SHADDA
 133 ظ\u0651 → ẓẓ ; # ARABIC LETTER ZAH + SHADDA
 134 ع\u0651 → $ayin $ayin ; # ARABIC LETTER AIN + SHADDA
 135 غ\u0651 → ghgh ; # ARABIC LETTER GHAIN + SHADDA
 136 ف\u0651 → ff ; # ARABIC LETTER FEH + SHADDA
 137 ق\u0651 → qq ; # ARABIC LETTER QAF + SHADDA
 138 ك\u0651 → kk ; # ARABIC LETTER KAF + SHADDA
 139 ل\u0651 → ll ; # ARABIC LETTER LAM + SHADDA
 140 م\u0651 → mm ; # ARABIC LETTER MEEM + SHADDA
 141 ن\u0651 → nn ; # ARABIC LETTER NOON + SHADDA
 142 ه\u0651 → hh ; # ARABIC LETTER HEH + SHADDA
 143 و\u0651 → ww ; # ARABIC LETTER WAW + SHADDA
 144 ى\u0651 → yy ; # ARABIC LETTER YEH + SHADDA
 145 #
 146 #
 147 ########################################################################
 148 #
 149 # End Rule 9
 150 #
 151 ########################################################################
 152 #
 153 ########################################################################
 154 #
 155 # Start of Transformations
 156 #
 157 ########################################################################
 158 #
 159 $wordBoundary{ء →  ; # ARABIC LETTER HAMZA
 160 ء → $alef ; # ARABIC LETTER HAMZA
 161 $wordBoundary{ا →  ; # ARABIC LETTER ALEF
 162 ٱ → $alef ; # ARABIC LETTER ALEF WASLA
 163 $wordBoundary{آ → ā ; # ARABIC LETTER ALEF WITH MADDA ABOVE
 164 آ → $alef ā ; # ARABIC LETTER ALEF WITH MADDA ABOVE
 165 ب → b ; # ARABIC LETTER BEH
 166 ت → t ; # ARABIC LETTER TEH
 167 ة → h ; # ARABIC LETTER TEH MARBUTA
 168 ث → th ; # ARABIC LETTER THEH
 169 ج → j ; # ARABIC LETTER JEEM
 170 ح → ḩ ; # ARABIC LETTER HAH
 171 خ → kh ; # ARABIC LETTER KHAH
 172 د → d ; # ARABIC LETTER DAL
 173 ذ → dh ; # ARABIC LETTER THAL
 174 ر → r ; # ARABIC LETTER REH
 175 ز → z ; # ARABIC LETTER ZAIN
 176 س → s ; # ARABIC LETTER SEEN
 177 ش → sh ; # ARABIC LETTER SHEEN
 178 ص → ş ; # ARABIC LETTER SAD
 179 ض → ḑ ; # ARABIC LETTER DAD
 180 ط → ţ ; # ARABIC LETTER TAH
 181 ظ → z\u0327 ; # ARABIC LETTER ZAH
 182 ع → $ayin ; # ARABIC LETTER AIN
 183 غ → gh ; # ARABIC LETTER GHAIN
 184 ف → f ; # ARABIC LETTER FEH
 185 ق → q ; # ARABIC LETTER QAF
 186 ک ↔ k $disambig ; # ARABIC LETTER KEHEH
 187 ك ↔ k ; # ARABIC LETTER KAF
 188 ل → l ; # ARABIC LETTER LAM
 189 م → m ; # ARABIC LETTER MEEM
 190 ن → n ; # ARABIC LETTER NOON
 191 ه → h ; # ARABIC LETTER HEH
 192 و → w ; # ARABIC LETTER WAW
 193 ى → y ; # ARABIC LETTER YEH
 194 \u064Eا → ā ; # ARABIC FATHA + ALEF
 195 \u064Eى → á ; # ARABIC FATHA + ALEF MAKSURA
 196 \u064Eي\u0652 → ay ; # ARABIC FATHA + YEH + SUKUN
 197 \u064Eو\u0652 → aw ; # ARABIC FATHA + WAW + SUKUN
 198 \u064E → a ; # ARABIC FATHA
 199 \u0650ي → ī ; # ARABIC KASRA + YEH
 200 \u0650 → i ; # ARABIC KASRA
 201 \u064Fو → ū ; # ARABIC DAMMA + WAW
 202 \u064F → u ; # ARABIC DAMMA
 203 \u0652 →  ; # ARABIC SUKUN
 204 \u064B → aⁿ ; # ARABIC FATHATAN
 205 \u064D → iⁿ ; # ARABIC KASRATAN
 206 \u064C → uⁿ ; # ARABIC DAMMATAN
 207 ::NFC (NFD) ;
 208 #
 209 #
 210 ########################################################################
 211