icuSources/data/translit/Latin_InterIndic.txt

   1 # ***************************************************************************
   2 # *
   3 # *  Copyright (C) 2004-2016, International Business Machines
   4 # *  Corporation; Unicode, Inc.; and others.  All Rights Reserved.
   5 # *
   6 # ***************************************************************************
   7 # File: Latin_InterIndic.txt
   8 # Generated from CLDR
   9 #
  10
  11 # Latin-InterIndic
  12 #:: NFD;
  13 #\u0E00 reserved
  14 #consonants
  15 $chandrabindu=\uE001;
  16 $anusvara=\uE002;
  17 $visarga=\uE003;
  18 #\u0E004 reserved
  19 # w←vowel→ represents the stand-alone form
  20 $wa=\uE005;
  21 $waa=\uE006;
  22 $wi=\uE007;
  23 $wii=\uE008;
  24 $wu=\uE009;
  25 $wuu=\uE00A;
  26 $wr=\uE00B;
  27 $wl=\uE00C;
  28 $wce=\uE00D; # LETTER CANDRA E
  29 $wse=\uE00E; # LETTER SHORT E
  30 $we=\uE00F;  # ए LETTER E
  31 $wai=\uE010;
  32 $wco=\uE011; # LETTER CANDRA O
  33 $wso=\uE012; # LETTER SHORT O
  34 $wo=\uE013;  # ओ LETTER O
  35 $wau=\uE014;
  36 $ka=\uE015;
  37 $kha=\uE016;
  38 $ga=\uE017;
  39 $gha=\uE018;
  40 $nga=\uE019;
  41 $ca=\uE01A;
  42 $cha=\uE01B;
  43 $ja=\uE01C;
  44 $jha=\uE01D;
  45 $nya=\uE01E;
  46 $tta=\uE01F;
  47 $ttha=\uE020;
  48 $dda=\uE021;
  49 $ddha=\uE022;
  50 $nna=\uE023;
  51 $ta=\uE024;
  52 $tha=\uE025;
  53 $da=\uE026;
  54 $dha=\uE027;
  55 $na=\uE028;
  56 $ena=\uE029; #compatibility
  57 $pa=\uE02A;
  58 $pha=\uE02B;
  59 $ba=\uE02C;
  60 $bha=\uE02D;
  61 $ma=\uE02E;
  62 $ya=\uE02F;
  63 $ra=\uE030;
  64 $rra=\uE031;
  65 $la=\uE032;
  66 $lla=\uE033;
  67 $ela=\uE034; #compatibility
  68 $va=\uE035;
  69 $vva=\uE081;
  70 $sha=\uE036;
  71 $ssa=\uE037;
  72 $sa=\uE038;
  73 $ha=\uE039;
  74 #\u093A Reserved
  75 #\u093B Reserved
  76 $nukta=\uE03C;
  77 $avagraha=\uE03D; # SIGN AVAGRAHA
  78 # ←vowel→ represents the dependent form
  79 $aa=\uE03E;
  80 $i=\uE03F;
  81 $ii=\uE040;
  82 $u=\uE041;
  83 $uu=\uE042;
  84 $rh=\uE043;
  85 $rrh=\uE044;
  86 $ce=\uE045; #VOWEL SIGN CANDRA E
  87 $se=\uE046; #VOWEL SIGN SHORT E
  88 $e=\uE047;
  89 $ai=\uE048;
  90 $co=\uE049; # VOWEL SIGN CANDRA O
  91 $so=\uE04A; # VOWEL SIGN SHORT O
  92 $o=\uE04B;  # ो
  93 $au=\uE04C;
  94 $virama=\uE04D;
  95 # \u094E Reserved
  96 # \u094F Reserved
  97 $om = \uE050; # OM
  98 # \u0951→;        # UNMAPPED STRESS SIGN UDATTA
  99 # \u0952→;        # UNMAPPED STRESS SIGN ANUDATTA
 100 # \u0953→;        # UNMAPPED GRAVE ACCENT
 101 # \u0954→;        # UNMAPPED ACUTE ACCENT
 102 $lm = \uE055;#  Telugu Length Mark
 103 $ailm=\uE056;#  AI Length Mark
 104 $aulm=\uE057;#  AU Length Mark
 105 #urdu compatibity forms
 106 $uka=\uE058;
 107 $ukha=\uE059;
 108 $ugha=\uE05A;
 109 $ujha=\uE05B;
 110 $uddha=\uE05C;
 111 $udha=\uE05D;
 112 $ufa=\uE05E;
 113 $uya=\uE05F;
 114 $wrr=\uE060;
 115 $wll=\uE061;
 116 $lh=\uE062;
 117 $llh=\uE063;
 118 $danda=\uE064;
 119 $doubleDanda=\uE065;
 120 $zero=\uE066;     # DIGIT ZERO
 121 $one=\uE067;      # DIGIT ONE
 122 $two=\uE068;      # DIGIT TWO
 123 $three=\uE069;    # DIGIT THREE
 124 $four=\uE06A;     # DIGIT FOUR
 125 $five=\uE06B;     # DIGIT FIVE
 126 $six=\uE06C;      # DIGIT SIX
 127 $seven=\uE06D;    # DIGIT SEVEN
 128 $eight=\uE06E;    # DIGIT EIGHT
 129 $nine=\uE06F;     # DIGIT NINE
 130 $dgs=\uE082;
 131 # For all other scripts
 132 $ecp0=\uE070;
 133 $ecp1=\uE071;
 134 $ecp2=\uE072;
 135 $ecp3=\uE073;
 136 $ecp4=\uE074;
 137 $ecp5=\uE075;
 138 $ecp6=\uE076;
 139 $ecp7=\uE077;
 140 $ecp8=\uE078;
 141 $ecp9=\uE079;
 142 $ecpA=\uE07A;
 143 $ecpB=\uE07B;
 144 $ecpC=\uE07C;
 145 $ecpD=\uE07D;
 146 $ecpE=\uE07E;
 147 $ecpF=\uE07F;
 148 # Khanda-ta
 149 $kta=\uE083;
 150 # ॰→; # nothing in Latin maps to InterIndic ABBREVIATION SIGN
 151 $depVowelAbove=[\uE03E-\uE040\uE045-\uE04C];
 152 $depVowelBelow=[\uE041-\uE044];
 153 $endThing=[$danda$doubleDanda];
 154 # $x was originally called '§'; $z was '%'
 155 $x=[$virama$aa$ai$au$ii$i$uu$u$rrh$rh$lh$e$o$se$ce$so$co];
 156 $z=[bcdfghjklmnpqrstvwxyz];
 157 $consonants=[[$ka-$ha]$z[क-ह][ক-হ][ਕ-ਹ][ક-હ][କ-ହ][க-ஹ][క-హ][ಕ-ಹ][ക-ഹ]];
 158 \u0315 → $avagraha;
 159 \u0303→$chandrabindu$anusvara;
 160 m\u0310→$chandrabindu;
 161 h\u0323→$visarga;
 162 x→$ka$virama$sa;
 163 # convert to independent forms at start of word or syllable:
 164 # dependent forms for roundtrip
 165 \u0314a\u0304→$aa;
 166 \u0314ai→$ai;
 167 \u0314au→$au;
 168 \u0314ii→$ii;
 169 \u0314i\u0304→$ii;
 170 \u0314i→$i;
 171 \u0314u\u0304→$uu;
 172 \u0314u→$u;
 173 \u0314r\u0325\u0304→$rrh;
 174 \u0314r\u0325→$rh;
 175 \u0314l\u0325\u0304→$llh;
 176 \u0314lh→$lh;
 177 \u0314l\u0325→$lh;
 178 \u0314e\u0304→$e;
 179 \u0314o\u0304→$o;
 180 \u0314a→;
 181 \u0314e\u0306→$ce;
 182 \u0314o\u0306→$co;
 183 \u0314e→$se;
 184 \u0314o→$so;
 185 # preceeded by consonants
 186 $consonants{ a\u0304→$aa;
 187 $consonants{ ai→$ai;
 188 $consonants{ au→$au;
 189 $consonants{ ii→$ii;
 190 $consonants{ i\u0304→$ii;
 191 $consonants{ i→$i;
 192 $consonants{ u\u0304→$uu;
 193 $consonants{ u→$u;
 194 $consonants{ r\u0325\u0304→$rrh;
 195 $consonants{ r\u0325a→$rh;
 196 $consonants{ r\u0325→$rh;
 197 $consonants{ l\u0325\u0304→$llh;
 198 $consonants{ lh→$lh;
 199 $consonants{ l\u0325→$lh;
 200 $consonants{ e\u0304→$e;
 201 $consonants{ o\u0304→$o;
 202 $consonants{ e\u0306→$ce;
 203 $consonants{ o\u0306→$co;
 204 $consonants{ e→$se;
 205 $consonants{ o→$so;
 206 # e.g. keai -→ {ka}{e}{wai}; k'ai -→ {ka}{wai}; (ai) -→ ({wai})
 207 a\u0304→$waa;
 208 ai→$wai;
 209 au→$wau;
 210 i\u0304→$wii;
 211 i→$wi;
 212 u\u0304→$wuu;
 213 u→$wu;
 214 r\u0325\u0304→$wrr;
 215 r\u0325→$wr;
 216 l\u0325\u0304→$wll;
 217 lh→$wl;
 218 l\u0325→$wl;
 219 e\u0304→$we;
 220 o\u0304→$wo;
 221 a→$wa;
 222 e\u0306→$wce;
 223 o\u0306→$wco;
 224 e→$wse;
 225 ''om→$om;
 226 o→$wso;
 227 # rules for anusvara
 228 n}r\u0325           → $na|$virama;
 229 n}l\u0325           → $na|$virama;
 230 n}na                → $na|$virama;
 231 n\u0307}[kg]        → $anusvara;
 232 n\u0307}n\u0307     → $anusvara;
 233 n\u0304}[cj]        → $anusvara;
 234 n\u0304}n\u0303     → $anusvara;
 235 n\u0323}[tdn]\u0323 → $anusvara;
 236 n}[tdn]             → $anusvara;
 237 m}[pbm]             → $anusvara;
 238 n}[ylvshr]          → $anusvara;
 239 m\u0307             → $anusvara;
 240 #urdu compatibility
 241 q→$uka|$virama;
 242 k\u0331h\u0331→$ukha |$virama;
 243 g\u0307→ $ugha | $virama;
 244 z → $ujha |$virama;
 245 f → $ufa|$virama;
 246 t\u0331→$kta;
 247 # dev
 248 y\u0307→$uya|$virama;
 249 l\u0331→$ela|$virama;
 250 n\u0331→$ena|$virama;
 251 n\u0307→$nga|$virama;
 252 n\u0303→$nya|$virama;
 253 n\u0323→$nna|$virama;
 254 t\u0323h→$ttha|$virama;
 255 t\u0323→$tta|$virama;
 256 r\u0323h→$udha|$virama;
 257 r\u0323→$uddha|$virama;
 258 d\u0323h→$ddha|$virama;
 259 d\u0323→$dda|$virama;
 260 kh→$kha|$virama;
 261 k→$ka|$virama;
 262 gh→$gha|$virama;
 263 g→$ga|$virama;
 264 ch→$cha|$virama;
 265 c→$ca|$virama;
 266 jh→$jha|$virama;
 267 j→$ja|$virama;
 268 ny→$nya|$virama;
 269 tth→$ttha|$virama;
 270 ddh→$ddha|$virama;
 271 th→$tha|$virama;
 272 t→$ta|$virama;
 273 dh→$dha|$virama;
 274 d→$da|$virama;
 275 n→$na|$virama;
 276 ph→$pha|$virama;
 277 p→$pa|$virama;
 278 bh→$bha|$virama;
 279 b→$ba|$virama;
 280 m→$ma|$virama;
 281 y→$ya|$virama;
 282 r\u0331→$rra|$virama;
 283 r→$ra|$virama;
 284 l\u0323→$lla|$virama;
 285 l→$la|$virama;
 286 v→$va|$virama;
 287 w\u0307→$vva|$virama;
 288 w→$va|$virama;
 289 sh→$sha|$virama;
 290 ss→$ssa|$virama;
 291 s\u0323→$ssa|$virama;
 292 s\u0301→$sha|$virama;
 293 s→$sa|$virama;
 294 h→$ha|$virama;
 295 '.'→$danda;
 296 $danda'.'→$doubleDanda;
 297 $depVowelAbove{'~'→$anusvara;
 298 $depVowelBelow{'~'→$chandrabindu;
 299 # convert to dependent forms after consonant with no vowel:
 300 # e.g. kai -→ {ka}{virama}ai -→ {ka}{ai}
 301 #$virama aa→$aa;
 302 $virama a\u0304→$aa;
 303 $virama ai→$ai;
 304 $virama au→$au;
 305 $virama ii→$ii;
 306 $virama i\u0304→$ii;
 307 $virama i→$i;
 308 #$virama uu→$uu;
 309 $virama u\u0304→$uu;
 310 $virama u→$u;
 311 #$virama rrh→$rrh;
 312 $virama r\u0325\u0304→$rrh;
 313 #$virama rh→$rh;
 314 $virama r\u0325a→$rh;
 315 $virama r\u0325→$rh;
 316 $virama l\u0325\u0304→$llh;
 317 $virama lh→$lh;
 318 $virama l\u0325→$lh;
 319 $virama e\u0304→$e;
 320 $virama o\u0304→$o;
 321 $virama a→;
 322 $virama e\u0306→$ce;
 323 $virama o\u0306→$co;
 324 $virama e→$se;
 325 $virama o→$so;
 326 # otherwise convert independent forms when separated by ': k'ai -→ {ka}{virama}{wai}
 327 #$virama''aa→$waa;
 328 $virama''a\u0304→$waa;
 329 $virama''ai→$wai;
 330 $virama''au→$wau;
 331 #$virama''ii→$wii;
 332 $virama''i\u0304→$wii;
 333 $virama''i→$wi;
 334 #$virama''uu→$wuu;
 335 $virama''u\u0304→$wuu;
 336 $virama''u→$wu;
 337 #$virama''rrh→$wrr;
 338 $virama''r\u0325\u0304→$wrr;
 339 #$virama''rh→$wr;
 340 $virama''r\u0325→$wr;
 341 $virama''l\u0325\u0304→$wll;
 342 #$virama''lh→$wl;
 343 $virama''l\u0325→$wl;
 344 $virama''e\u0304→$we;
 345 $virama''o\u0304→$wo;
 346 $virama''a→$wa;
 347 $virama''e\u0306→$wce;
 348 $virama''o\u0306→$wco;
 349 $virama''e→$wse;
 350 $virama''o→$wso;
 351 # no virama
 352 ''a\u0304→$waa;
 353 ''ai→$wai;
 354 ''au→$wau;
 355 ''i\u0304→$wii;
 356 ''i→$wi;
 357 ''u\u0304→$wuu;
 358 ''u→$wu;
 359 ''r\u0325\u0304→$wrr;
 360 ''r\u0325→$wr;
 361 ''l\u0325\u0304→$wll;
 362 ''l\u0325→$wl;
 363 ''e\u0304→$we;
 364 ''o\u0304→$wo;
 365 ''a→$wa;
 366 ''e\u0306→$wce;
 367 ''o\u0306→$wco;
 368 ''e→$wse;
 369 ''o→$wso;
 370 $virama } [$z] → $virama;
 371 $virama } ' ' → $virama ;
 372 $virama}$endThing→;
 373 ʔ→$dgs; # Glottal Stop
 374 0→$zero;
 375 1→$one;
 376 2→$two;
 377 3→$three;
 378 4→$four;
 379 5→$five;
 380 6→$six;
 381 7→$seven;
 382 8→$eight;
 383 9→$nine;
 384 ''→;
 385 #:: NFC (NFD) ;
 386