icuSources/data/translit/dv_dv_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: dv_dv_Latn_BGN.txt
   5 # Generated from CLDR
   6 #
   7
   8 # BGN/PCGN 1988 Agreement, with modifications 2009
   9 #
  10 # This romanization system supersedes the one which was approved by
  11 # the BGN and the PCGN in 1972. This official system was submitted
  12 # to the PCGN by the Maldivian government in 1987 and approved by BGN
  13 # and PCGN in 1988. The system presented here reflects the 1988 Agreement
  14 # with minor modifications introduced by the government of the Maldives
  15 # in 2009.
  16 #
  17 # In our rules, we also convert Arabic punctuation characters to Latin.
  18 # These appears to be used in Maldivian text, for example in the Universal
  19 # Declaration of Human Rights.
  20 ::[[:block=thaana:][،؛؟٪٫٬]\uFDF2] ;
  21 ::NFD;
  22 $wordBoundary = [^[:L:][:M:][:N:]] ;
  23 $vowel = [\u07A6-\u07AF] ;
  24 $sukun = \u07B0 ;
  25 $sign = [$sukun $vowel] ;
  26 $rule4 = [އށ] $sukun ; # see note 4
  27 ### Consonants
  28 # HAA
  29 $rule4 ހ → hh ;
  30 ހ → h ;
  31 # NOONU
  32 # See note 5: "romanized n’ when appearing without any vowel or auxiliary sign"
  33 $rule4 ނ } $sign → nn ;
  34 $rule4 ނ → nn\' ;
  35 ނ } $sign → n ;
  36 ނ → n\' ;
  37 # RAA
  38 $rule4 ރ → rr ;
  39 ރ → r ;
  40 # BAA
  41 $rule4 ބ → bb ;
  42 ބ → b ;
  43 # LHAVIYANI
  44 $rule4 ޅ → hlh ;
  45 ޅ → lh;
  46 # KAAFU
  47 $rule4 ކ → kk ;
  48 ކ → k ;
  49 # VAAVU
  50 $rule4 ވ → vv ;
  51 ވ → v ;
  52 # MEEMU
  53 $rule4 މ → mm ;
  54 މ → m ;
  55 # FAAFU
  56 $rule4 ފ → ff ;
  57 ފ → f;
  58 # DHAALU
  59 $rule4 ދ → hdh ;
  60 ދ → dh;
  61 # THAA
  62 # See note 6: "romanized iy when appearing in combination with a supercircle"
  63 $rule4 ތ $sukun → hiy ;
  64 $rule4 ތ → hth ;
  65 \u07A8 ތ $sukun → iy ;
  66 ތ $sukun → iy ;
  67 ތ → th ;
  68 # LAAMU
  69 $rule4 ލ → ll ;
  70 ލ → l ;
  71 # GAAFU
  72 $rule4 ގ → gg ;
  73 ގ → g ;
  74 # GNAVIYANI
  75 $rule4 ޏ → hgn ;
  76 ޏ → gn ;
  77 # SEENU
  78 $rule4 ސ → ss ;
  79 ސ → s ;
  80 # DAVIYANI
  81 $rule4 ޑ → dd ;
  82 ޑ → d ;
  83 # ZAVIYANI
  84 $rule4 ޒ → zz ;
  85 ޒ → z ;
  86 # TAVIYANI
  87 $rule4 ޓ → tt ;
  88 ޓ → t ;
  89 # YAA
  90 $rule4 ޔ → yy ;
  91 ޔ → y ;
  92 # PAVIYANI
  93 $rule4 ޕ → pp ;
  94 ޕ → p ;
  95 # JAVIYANI
  96 $rule4 ޖ → jj ;
  97 ޖ → j ;
  98 # CHAVIYANI
  99 $rule4 ޗ → hch ;
 100 ޗ → ch ;
 101 ### Borrowed Consonants (See Rule 7)
 102 # SAADHU
 103 $rule4 ޞ → şş ;
 104 ޞ → ş ;
 105 # SHEENU
 106 $rule4 ޝ → hsh ;
 107 ޝ → sh ;
 108 # ZAA
 109 $rule4 ޜ → zz ;
 110 ޜ → z;
 111 # KHAA
 112 $rule4 ޚ → hkh ;
 113 ޚ → kh;
 114 # HHAA
 115 $rule4 ޙ → ḩḩ ;
 116 ޙ → ḩ ;
 117 # THAALU
 118 $rule4 ޛ → hdh ;
 119 ޛ → dh ;
 120 # TTAA
 121 $rule4 ޘ → hth ;
 122 ޘ → th ;
 123 # WAAVU
 124 $rule4 ޥ → ww ;
 125 ޥ → w ;
 126 # QAAFU
 127 $rule4 ޤ → qq ;
 128 ޤ → q ;
 129 # GHAINU
 130 $rule4 ޣ → hgh ;
 131 ޣ → gh ;
 132 # AINU
 133 $rule4 ޢ → \'\' ;
 134 ޢ → \' ;
 135 # ZO
 136 $rule4 ޡ → z\u0327z\u0327 ;
 137 ޡ → z\u0327 ;
 138 # TO
 139 $rule4 ޠ → ţţ ;
 140 ޠ → ţ ;
 141 # DAADHU
 142 $rule4 ޟ → ḑḑ ;
 143 ޟ → ḑ ;
 144 # NOTE: not in Maldivian BGN system, but for completeness of Thaana block
 145 # NAA
 146 $rule4 ޱ → n\u0332n\u0332 ;
 147 ޱ → n\u0332 ;
 148 # Rule 4 in word-final position
 149 $rule4 } $wordBoundary → h;
 150 # SHAVIYANI (placed last to avoid masking)
 151 $rule4 ށ → hsh;
 152 ށ → sh;
 153 # Otherwise, these signs are not romanized elsewhere
 154 $rule4 → ;
 155 \u07B0 → ;
 156 އ → ;
 157 # NOTE: not in Maldivian BGN system, but common in names (e.g. Abdullah)
 158 ($vowel) \uFDF2 → | $1 llāh ;
 159 \uFDF2 → allāh;
 160 ### Vowels
 161 \u07A6 → a; # ABAFILI
 162 \u07A7 → aa; # AABAAFILI
 163 \u07AC → e; # EBEFILI
 164 \u07AD → ey; # EYBEYFILI
 165 \u07A8 → i; # IBIFILI
 166 \u07A9 → ee; # EEBEEFILI
 167 \u07AE → o; # OBOFILI
 168 \u07AF → oa; # OABOAFILI
 169 \u07AA → u; # UBUFILI
 170 \u07AB → oo; # OOBOOFILI
 171 ، → ',';  # U+060C ARABIC COMMA
 172 ؛ → ';';  # U+061B ARABIC SEMICOLON
 173 ؟ → '?';  # U+061F ARABIC QUESTION MARK
 174 ٪ → '%';  # U+066A ARABIC PERCENT SIGN
 175 ٫ → '.';  # U+066B ARABIC DECIMAL SEPARATOR
 176 ٬ → ',';  # U+066C ARABIC THOUSANDS SEPARATOR
 177