icuSources/data/translit/InterIndic_Latin.txt

   1 # © 2016 and later: Unicode, Inc. and others.
   2 # License & terms of use: http://www.unicode.org/copyright.html#License
   3 #
   4 # File: InterIndic_Latin.txt
   5 # Generated from CLDR
   6 #
   7
   8 # InterIndic-Latin
   9 #\u0E00 reserved
  10 #consonants
  11 $chandrabindu=\uE001;
  12 $anusvara=\uE002;
  13 $visarga=\uE003;
  14 #\u0E004 reserved
  15 # w←vowel→ represents the stand-alone form
  16 $wa=\uE005;
  17 $waa=\uE006;
  18 $wi=\uE007;
  19 $wii=\uE008;
  20 $wu=\uE009;
  21 $wuu=\uE00A;
  22 $wr=\uE00B;
  23 $wl=\uE00C;
  24 $wce=\uE00D; # LETTER CANDRA E
  25 $wse=\uE00E; # LETTER SHORT E
  26 $we=\uE00F;  # ए LETTER E
  27 $wai=\uE010;
  28 $wco=\uE011; # LETTER CANDRA O
  29 $wso=\uE012; # LETTER SHORT O
  30 $wo=\uE013;  # ओ LETTER O
  31 $wau=\uE014;
  32 $ka=\uE015;
  33 $kha=\uE016;
  34 $ga=\uE017;
  35 $gha=\uE018;
  36 $nga=\uE019;
  37 $ca=\uE01A;
  38 $cha=\uE01B;
  39 $ja=\uE01C;
  40 $jha=\uE01D;
  41 $nya=\uE01E;
  42 $tta=\uE01F;
  43 $ttha=\uE020;
  44 $dda=\uE021;
  45 $ddha=\uE022;
  46 $nna=\uE023;
  47 $ta=\uE024;
  48 $tha=\uE025;
  49 $da=\uE026;
  50 $dha=\uE027;
  51 $na=\uE028;
  52 $ena=\uE029; #compatibility
  53 $pa=\uE02A;
  54 $pha=\uE02B;
  55 $ba=\uE02C;
  56 $bha=\uE02D;
  57 $ma=\uE02E;
  58 $ya=\uE02F;
  59 $ra=\uE030;
  60 $vva=\uE081;
  61 $rra=\uE031;
  62 $la=\uE032;
  63 $lla=\uE033;
  64 $ela=\uE034; #compatibility
  65 $va=\uE035;
  66 $sha=\uE036;
  67 $ssa=\uE037;
  68 $sa=\uE038;
  69 $ha=\uE039;
  70 #\u093A Reserved
  71 #\u093B Reserved
  72 $nukta=\uE03C;
  73 $avagraha=\uE03D; # SIGN AVAGRAHA
  74 # ←vowel→ represents the dependent form
  75 $aa=\uE03E;
  76 $i=\uE03F;
  77 $ii=\uE040;
  78 $u=\uE041;
  79 $uu=\uE042;
  80 $rh=\uE043;
  81 $rrh=\uE044;
  82 $ce=\uE045; #VOWEL SIGN CANDRA E
  83 $se=\uE046; #VOWEL SIGN SHORT E
  84 $e=\uE047;
  85 $ai=\uE048;
  86 $co=\uE049; # VOWEL SIGN CANDRA O
  87 $so=\uE04A; # VOWEL SIGN SHORT O
  88 $o=\uE04B;  # ो
  89 $au=\uE04C;
  90 $virama=\uE04D;
  91 # \u094E Reserved
  92 # \u094F Reserved
  93 $om=\uE050; # OM
  94 \uE051→;        # UNMAPPED STRESS SIGN UDATTA
  95 \uE052→;        # UNMAPPED STRESS SIGN ANUDATTA
  96 \uE053→;        # UNMAPPED GRAVE ACCENT
  97 \uE054→;        # UNMAPPED ACUTE ACCENT
  98 $lm = \uE055;#  Telugu Length Mark
  99 $ailm=\uE056;#  AI Length Mark
 100 $aulm=\uE057;#  AU Length Mark
 101 #urdu compatibity forms
 102 $uka=\uE058;
 103 $ukha=\uE059;
 104 $ugha=\uE05A;
 105 $ujha=\uE05B;
 106 $uddha=\uE05C;
 107 $udha=\uE05D;
 108 $ufa=\uE05E;
 109 $uya=\uE05F;
 110 $wrr=\uE060;
 111 $wll=\uE061;
 112 $lh=\uE062;
 113 $llh=\uE063;
 114 $danda=\uE064;
 115 $doubleDanda=\uE065;
 116 $zero=\uE066;     # DIGIT ZERO
 117 $one=\uE067;      # DIGIT ONE
 118 $two=\uE068;      # DIGIT TWO
 119 $three=\uE069;    # DIGIT THREE
 120 $four=\uE06A;     # DIGIT FOUR
 121 $five=\uE06B;     # DIGIT FIVE
 122 $six=\uE06C;      # DIGIT SIX
 123 $seven=\uE06D;    # DIGIT SEVEN
 124 $eight=\uE06E;    # DIGIT EIGHT
 125 $nine=\uE06F;     # DIGIT NINE
 126 # Glottal stop
 127 $dgs=\uE082;
 128 #Khanda-ta
 129 $kta=\uE083;
 130 $depVowelAbove=[\uE03E-\uE040\uE045-\uE04C];
 131 $depVowelBelow=[\uE041-\uE044];
 132 # $x was originally called '§'; $z was '%'
 133 $x=[$aa$ai$au$ii$i$uu$u$rrh$rh$lh$llh$e$o$se$ce$so$co];
 134 $z=[bcdfghjklmnpqrstvwxyz];
 135 $vowels=[aeiour\u0304\u0325\u0306];
 136 $forceIndependentMatra = [^[[:L:][\u0300-\u034C]]];
 137 ######################################################################
 138 # convert from Native letters to Latin letters
 139 ######################################################################
 140 #transliterations for anusvara
 141 $anusvara} [$ka$kha$ga$gha$nga]             → n\u0307;
 142 $anusvara} [$ca$cha$ja$jha$nya]             → n\u0304;
 143 $anusvara} [$tta$ttha$dda$ddha$nna]         → n\u0323;
 144 $anusvara} [$ta$tha$da$dha$na]              → n;
 145 $anusvara} [$pa$pha$ba$bha$ma]              → m;
 146 $anusvara} [$ya$ra$lla$la$va$ssa$sha$sa$ha] → n;
 147 $anusvara→ m\u0307;
 148 # Urdu compatibility
 149 $ya$nukta}$x        → y\u0307;
 150 $ya$nukta$virama    → y\u0307;
 151 $ya$nukta           → y\u0307a;
 152 $la$nukta }$x       → l\u0331;
 153 $la$nukta$virama    → l\u0331;
 154 $la$nukta           → l\u0331a;
 155 $na$nukta }$x       → n\u0331;
 156 $na$nukta$virama    → n\u0331;
 157 $na$nukta           → n\u0331a;
 158 $ena }$x            → n\u0331;
 159 $ena$virama         → n\u0331;
 160 $ena                → n\u0331a;
 161 $uka                → qa;
 162 $ka$nukta }$x       → q;
 163 $ka$nukta$virama    → q;
 164 $ka$nukta           → qa;
 165 $kha$nukta }$x      → k\u0331h\u0331;
 166 $kha$nukta$virama   → k\u0331h\u0331;
 167 $kha$nukta          → k\u0331h\u0331a;
 168 $ukha$virama        → k\u0331h\u0331;
 169 $ukha               → k\u0331h\u0331a;
 170 $ugha               → g\u0307a;
 171 $ga$nukta }$x       → g\u0307;
 172 $ga$nukta$virama    → g\u0307;
 173 $ga$nukta           → g\u0307a;
 174 $ujha               → za;
 175 $ja$nukta }$x       → z;
 176 $ja$nukta$virama    → z;
 177 $ja$nukta           → za;
 178 $ddha$nukta}$x      → r\u0323h;
 179 $ddha$nukta$virama  → r\u0323h;
 180 $ddha$nukta         → r\u0323ha;
 181 $uddha}$x           → r\u0323;
 182 $uddha$virama       → r\u0323;
 183 $uddha              → r\u0323a;
 184 $udha               → r\u0323a;
 185 $dda$nukta}$x       → r\u0323;
 186 $dda$nukta$virama   → r\u0323;
 187 $dda$nukta          → r\u0323a;
 188 $pha$nukta }$x      → f;
 189 $pha$nukta$virama   → f;
 190 $pha$nukta          → fa;
 191 $ufa }$x            → f;
 192 $ufa$virama         → f;
 193 $ufa                → fa;
 194 $ra$nukta}$x        → r\u0331;
 195 $ra$nukta$virama    → r\u0331;
 196 $ra$nukta           → r\u0331a;
 197 $lla$nukta}$x       → l\u0331;
 198 $lla$nukta$virama   → l\u0331;
 199 $lla$nukta          → l\u0331a;
 200 $ela}$x             → l\u0331;
 201 $ela$virama         → l\u0331;
 202 $ela                → l\u0331a;
 203 $uya}$x             → y\u0307;
 204 $uya$virama         → y\u0307;
 205 $uya                → y\u0307a;
 206 # normal consonants
 207 $ka$virama}$ha→k'';
 208 $ka}$x→k;
 209 $ka$virama→k;
 210 $ka→ka;
 211 $kha}$x→kh;
 212 $kha$virama→kh;
 213 $kha→kha;
 214 $ga$virama}$ha→g'';
 215 $ga}$x→g;
 216 $ga$virama→g;
 217 $ga→ga;
 218 $gha}$x→gh;
 219 $gha$virama→gh;
 220 $gha→gha;
 221 $nga}$x→n\u0307;
 222 $nga$virama→n\u0307;
 223 $nga→n\u0307a;
 224 $ca$virama}$ha→c'';
 225 $ca}$x→c;
 226 $ca$virama→c;
 227 $ca→ca;
 228 $cha}$x→ch;
 229 $cha$virama→ch;
 230 $cha→cha;
 231 $ja$virama}$ha→j'';
 232 $ja}$x→j;
 233 $ja$virama→j;
 234 $ja→ja;
 235 $jha}$x→jh;
 236 $jha$virama→jh;
 237 $jha→jha;
 238 $nya }$x→n\u0303;
 239 $nya$virama→n\u0303;
 240 $nya → n\u0303a;
 241 $tta$virama}$ha→t\u0323'';
 242 $tta}$x→t\u0323;
 243 $tta$virama→t\u0323;
 244 $tta→t\u0323a;
 245 $ttha}$x→t\u0323h;
 246 $ttha$virama→t\u0323h;
 247 $ttha→t\u0323ha;
 248 $dda}$x$ha→d\u0323'';
 249 $dda}$x→d\u0323;
 250 $dda$virama→d\u0323;
 251 $dda→d\u0323a;
 252 $ddha}$x→d\u0323h;
 253 $ddha$virama→d\u0323h;
 254 $ddha→d\u0323ha;
 255 $nna}$x→n\u0323;
 256 $nna$virama→n\u0323;
 257 $nna→n\u0323a;
 258 $ta$virama}$ha→t'';
 259 $ta$virama}$ttha→t'';
 260 $ta$virama}$tta→t'';
 261 $ta$virama}$tha→t'';
 262 $ta}$x→t;
 263 $ta$virama→t;
 264 $ta→ta;
 265 $tha}$x→th;
 266 $tha$virama→th;
 267 $tha→tha;
 268 $da$virama}$ha→d'';
 269 $da$virama}$ddha→d'';
 270 $da$virama}$dda→d'';
 271 $da$virama}$dha→d'';
 272 $da}$x→d;
 273 $da$virama→d;
 274 $da→da;
 275 $dha}$x→dh;
 276 $dha$virama→dh;
 277 $dha→dha;
 278 $na$virama}$ga→n'';
 279 $na$virama}$ya→n'';
 280 $na}$x→n;
 281 $na$virama→n;
 282 $na→na;
 283 $pa$virama}$ha→p'';
 284 $pa}$x→p;
 285 $pa$virama→p;
 286 $pa→pa;
 287 $pha}$x→ph;
 288 $pha$virama→ph;
 289 $pha→pha;
 290 $ba$virama}$ha→b'';
 291 $ba}$x→b;
 292 $ba$virama→b;
 293 $ba→ba;
 294 $bha}$x→bh;
 295 $bha$virama→bh;
 296 $bha→bha;
 297 $ma$virama}$ma→m'';
 298 $ma}$x→m;
 299 $ma$virama→m;
 300 $ma→ma;
 301 $ya}$x→y;
 302 $ya$virama→y;
 303 $ya→ya;
 304 $ra$virama}$ha→r'';
 305 $ra}$x→r;
 306 $ra$virama→r;
 307 $ra→ra;
 308 $vva$virama}$ha→w\u0307'';
 309 $vva}$x→w\u0307;
 310 $vva$virama→w\u0307;
 311 $vva→w\u0307a;
 312 $rra$virama}$ha→r\u0331'';
 313 $rra}$x→r\u0331;
 314 $rra$virama→r\u0331;
 315 $rra→r\u0331a;
 316 $la$virama}$ha→l'';
 317 $la}$x→l;
 318 $la$virama→l;
 319 $la→la;
 320 $lla$virama}$ha→l\u0323'';
 321 $lla}$x→l\u0323;
 322 $lla$virama→l\u0323;
 323 $lla→l\u0323a;
 324 $va}$x→v;
 325 $va$virama→v;
 326 $va→va;
 327 $sa$virama}$ha→s'';
 328 $sa$virama}$sha→s'';
 329 $sa$virama}$ssa→s'';
 330 $sa$virama}$sa→s'';
 331 $sa}$x→s;
 332 $sa$virama→s;
 333 #for gurmukhi
 334 $sa$nukta}$x→s\u0301;
 335 $sa$nukta$virama→s\u0301;
 336 $sa$nukta→s\u0301a;
 337 $sa→sa;
 338 $sha}$x→s\u0301;
 339 $sha$virama→s\u0301;
 340 $sha→s\u0301a;
 341 $ssa}$x→s\u0323;
 342 $ssa$virama→s\u0323;
 343 $ssa→s\u0323a;
 344 $ha}$x→h;
 345 $ha$virama→h;
 346 $ha→ha;
 347 # dependent vowels (should never occur except following consonants)
 348 $forceIndependentMatra{$aa  → \u0314a\u0304;
 349 $forceIndependentMatra{$ai  → \u0314ai;
 350 $forceIndependentMatra{$au  → \u0314au;
 351 $forceIndependentMatra{$ii  → \u0314i\u0304;
 352 $forceIndependentMatra{$i   → \u0314i;
 353 $forceIndependentMatra{$uu  → \u0314u\u0304;
 354 $forceIndependentMatra{$u   → \u0314u;
 355 $forceIndependentMatra{$rrh → \u0314r\u0325\u0304;
 356 $forceIndependentMatra{$rh  → \u0314r\u0325;
 357 $forceIndependentMatra{$llh → \u0314l\u0325\u0304;
 358 $forceIndependentMatra{$lh  → \u0314l\u0325;
 359 $forceIndependentMatra{$e   → \u0314e\u0304;
 360 $forceIndependentMatra{$o   → \u0314o\u0304;
 361 #extra vowels
 362 $forceIndependentMatra{$ce  → \u0314e\u0306;
 363 $forceIndependentMatra{$co  → \u0314o\u0306;
 364 $forceIndependentMatra{$se  → \u0314e;
 365 $forceIndependentMatra{$so  → \u0314o;
 366 $forceIndependentMatra{$nukta  →; # Nukta cannot appear independently or as first character
 367 $forceIndependentMatra{$virama →; # Virama cannot appear independently or as first character
 368 $aa  → a\u0304;
 369 $ai  → ai;
 370 $au  → au;
 371 $ii  → i\u0304;
 372 $i   → i;
 373 $uu  → u\u0304;
 374 $u   → u;
 375 $rrh → r\u0325\u0304;
 376 $rh  → r\u0325;
 377 $llh → l\u0325\u0304;
 378 $lh  → l\u0325;
 379 $e   → e\u0304;
 380 $o   → o\u0304;
 381 #extra vowels
 382 $ce  → e\u0306;
 383 $co  → o\u0306;
 384 $se  → e;
 385 $so  → o;
 386 #dependent vowels when following independent vowels. Generally Illegal only for roundtripping
 387 $waa} $x → a\u0304\u0314;
 388 $wai} $x → ai\u0314;
 389 $wau} $x → au\u0314;
 390 $wii} $x → i\u0304\u0314;
 391 $wi } $x → i\u0314;
 392 $wuu} $x → u\u0304\u0314;
 393 $wu } $x → u\u0314;
 394 $wrr} $x → r\u0325\u0304\u0314;
 395 $wr } $x → r\u0325\u0314;
 396 $wll} $x → l\u0325\u0304\u0314;
 397 $wl } $x → l\u0325\u0314;
 398 $we } $x → e\u0304\u0314;
 399 $wo } $x → o\u0304\u0314;
 400 $wa } $x → a\u0314;
 401 #extra vowels
 402 $wce} $x → e\u0306\u0314;
 403 $wco} $x → o\u0306\u0314;
 404 $wse} $x → e\u0314;
 405 $wso} $x → o\u0314;
 406 $om} $x → ''om\u0314;
 407 # independent vowels when preceeded by vowels
 408 $vowels{$waa  → ''a\u0304;
 409 $vowels{$wai  → ''ai;
 410 $vowels{$wau  → ''au;
 411 $vowels{$wii  → ''i\u0304;
 412 $vowels{$wi   → ''i;
 413 $vowels{$wuu  → ''u\u0304;
 414 $vowels{$wu   → ''u;
 415 $vowels{$wrr  → ''r\u0325\u0304;
 416 $vowels{$wr   → ''r\u0325;
 417 $vowels{$wll  → ''l\u0325\u0304;
 418 $vowels{$wl   → ''l\u0325;
 419 $vowels{$we   → ''e\u0304;
 420 $vowels{$wo   → ''o\u0304;
 421 $vowels{$wa   → ''a;
 422 #extra vowels
 423 $vowels{$wce  → ''e\u0306;
 424 $vowels{$wco  → ''o\u0306;
 425 $vowels{$wse  → ''e;
 426 $vowels{$wso  → ''o;
 427 # independent vowels (otherwise)
 428 $waa → a\u0304;
 429 $wai → ai;
 430 $wau → au;
 431 $wii → i\u0304;
 432 $wi  → i;
 433 $wuu → u\u0304;
 434 $wu  → u;
 435 $wrr → r\u0325\u0304;
 436 $wr  → r\u0325;
 437 $wll → l\u0325\u0304;
 438 $wl  → l\u0325;
 439 $we  → e\u0304;
 440 $wo  → o\u0304;
 441 $wa  → a;
 442 #extra vowels
 443 $wce → e\u0306;
 444 $wco → o\u0306;
 445 $wse → e;
 446 $wso → o;
 447 $om → ''om;
 448 #stress marks
 449 $avagraha → \u0315;
 450 $chandrabindu$anusvara→\u0303;
 451 $chandrabindu → m\u0310;
 452 $visarga→h\u0323;
 453 #numbers
 454 $zero  → 0;
 455 $one   → 1;
 456 $two   → 2;
 457 $three → 3;
 458 $four  → 4;
 459 $five  → 5;
 460 $six   → 6;
 461 $seven → 7;
 462 $eight → 8;
 463 $nine  → 9;
 464 $lm   →;
 465 $ailm →;
 466 $aulm →;
 467 $dgs→ʔ;
 468 $kta→t\u0331;
 469 $danda→'.';
 470 $doubleDanda→'.';
 471 \uE070→;       # ABBREVIATION SIGN
 472 # LETTER RA WITH MIDDLE DIAGONAL
 473 \uE071}$x→ra;
 474 \uE071$virama→r;
 475 \uE071→ra;
 476 # LETTER RA WITH LOWER DIAGONAL
 477 \uE072}$x→ra;
 478 \uE072$virama→r;
 479 \uE072→ra;
 480 \uE073→;       # RUPEE MARK
 481 \uE074→;       # RUPEE SIGN
 482 \uE075→;       # CURRENCY NUMERATOR ONE
 483 \uE076→;       # CURRENCY NUMERATOR TWO
 484 \uE077→;       # CURRENCY NUMERATOR THREE
 485 \uE078→;       # CURRENCY NUMERATOR FOUR
 486 \uE079→;       # CURRENCY NUMERATOR ONE LESS THAN THE DENOMINATOR
 487 \uE07A→;       # CURRENCY DENOMINATOR SIXTEEN
 488 \uE07B→;       # ISSHAR
 489 \uE07C→;       # TIPPI
 490 \uE07D→;       # ADDAK
 491 \uE07E→;       # IRI
 492 \uE07F→;       # URA
 493 \uE080→;       # EK ONKAR
 494 \uE004→;       # DEVANAGARI VOWEL SIGN SHORT A
 495