icuSources/data/translit/InterIndic_Latin.txt

   1 #--------------------------------------------------------------------
   2 # Copyright (c) 1999-2004, International Business Machines
   3 # Corporation and others. All Rights Reserved.
   4 #--------------------------------------------------------------------
   5
   6 # InterIndic-Latin
   7    #\u0e00 reserved
   8     #consonants
   9     $chandrabindu=\ue001;
  10     $anusvara=\ue002;
  11     $visarga=\ue003;
  12     #\u0e004 reserved
  13     # w<vowel> represents the stand-alone form
  14     $wa=\ue005;
  15     $waa=\ue006;
  16     $wi=\ue007;
  17     $wii=\ue008;
  18     $wu=\ue009;
  19     $wuu=\ue00a;
  20     $wr=\ue00b;
  21     $wl=\ue00c;
  22     $wce=\ue00d; # LETTER CANDRA E
  23     $wse=\ue00e; # LETTER SHORT E
  24     $we=\ue00f;  # \u090f LETTER E
  25     $wai=\ue010;
  26     $wco=\ue011; # LETTER CANDRA O
  27     $wso=\ue012; # LETTER SHORT O
  28     $wo=\ue013;  # \u0913 LETTER O
  29     $wau=\ue014;
  30     $ka=\ue015;
  31     $kha=\ue016;
  32     $ga=\ue017;
  33     $gha=\ue018;
  34     $nga=\ue019;
  35     $ca=\ue01a;
  36     $cha=\ue01b;
  37     $ja=\ue01c;
  38     $jha=\ue01d;
  39     $nya=\ue01e;
  40     $tta=\ue01f;
  41     $ttha=\ue020;
  42     $dda=\ue021;
  43     $ddha=\ue022;
  44     $nna=\ue023;
  45     $ta=\ue024;
  46     $tha=\ue025;
  47     $da=\ue026;
  48     $dha=\ue027;
  49     $na=\ue028;
  50     $ena=\ue029; #compatibility
  51     $pa=\ue02a;
  52     $pha=\ue02b;
  53     $ba=\ue02c;
  54     $bha=\ue02d;
  55     $ma=\ue02e;
  56     $ya=\ue02f;
  57     $ra=\ue030;
  58     $vva=\ue081;
  59     $rra=\ue031;
  60     $la=\ue032;
  61     $lla=\ue033;
  62     $ela=\ue034; #compatibility
  63     $va=\ue035;
  64     $sha=\ue036;
  65     $ssa=\ue037;
  66     $sa=\ue038;
  67     $ha=\ue039;
  68 #\u093a Reserved
  69 #\u093b Reserved
  70     $nukta=\ue03c;
  71     $avagraha=\ue03d; # SIGN AVAGRAHA
  72     # <vowel> represents the dependent form
  73     $aa=\ue03e;
  74     $i=\ue03f;
  75     $ii=\ue040;
  76     $u=\ue041;
  77     $uu=\ue042;
  78     $rh=\ue043;
  79     $lh=\ue044;
  80     $ce=\ue045; #VOWEL SIGN CANDRA E
  81     $se=\ue046; #VOWEL SIGN SHORT E
  82     $e=\ue047;
  83     $ai=\ue048;
  84     $co=\ue049; # VOWEL SIGN CANDRA O
  85     $so=\ue04a; # VOWEL SIGN SHORT O
  86     $o=\ue04b;  # \u094b
  87     $au=\ue04c;
  88     $virama=\ue04d;
  89 # \u094e Reserved
  90 # \u094f Reserved
  91     $om=\ue050; # OM
  92     \ue051>;        # UNMAPPED STRESS SIGN UDATTA
  93     \ue052>;        # UNMAPPED STRESS SIGN ANUDATTA
  94     \ue053>;        # UNMAPPED GRAVE ACCENT
  95     \ue054>;        # UNMAPPED ACUTE ACCENT
  96     $lm = \ue055;#  Telugu Length Mark
  97     $ailm=\ue056;#  AI Length Mark
  98     $aulm=\ue057;#  AU Length Mark
  99      #urdu compatibity forms
 100     $uka=\ue058;
 101     $ukha=\ue059;
 102     $ugha=\ue05a;
 103     $ujha=\ue05b;
 104     $uddha=\ue05c;
 105     $udha=\ue05d;
 106     $ufa=\ue05e;
 107     $uya=\ue05f;
 108     $wrr=\ue060;
 109     $wll=\ue061;
 110     $rrh=\ue062;
 111     $llh=\ue063;
 112     $danda=\ue064;
 113     $doubleDanda=\ue065;
 114     $zero=\ue066;     # DIGIT ZERO
 115     $one=\ue067;      # DIGIT ONE
 116     $two=\ue068;      # DIGIT TWO
 117     $three=\ue069;    # DIGIT THREE
 118     $four=\ue06a;     # DIGIT FOUR
 119     $five=\ue06b;     # DIGIT FIVE
 120     $six=\ue06c;      # DIGIT SIX
 121     $seven=\ue06d;    # DIGIT SEVEN
 122     $eight=\ue06e;    # DIGIT EIGHT
 123     $nine=\ue06f;     # DIGIT NINE
 124
 125 # \u0970>; # UNMAPPED ABBREVIATION SIGN
 126     $depVowelAbove=[\ue03e-\ue040\ue045-\ue04c];
 127     $depVowelBelow=[\ue041-\ue044];
 128     # $x was originally called '&'; $z was '%'
 129     $x=[$aa$ai$au$ii$i$uu$u$rrh$rh$lh$llh$e$o$se$ce$so$co];
 130     $z=[bcdfghjklmnpqrstvwxyz];
 131     $vowels=[aeiour\u0304\u0325\u0306];
 132     $forceIndependentMatra = [^[[:L:][\u0300-\u034c]]];
 133     ######################################################################
 134     # convert from Native letters to Latin letters
 135     ######################################################################
 136      #transliterations for anusvara
 137     $anusvara} [$ka$kha$ga$gha$nga]             > n\u0307;
 138     $anusvara} [$ca$cha$ja$jha$nya]             > n\u0304;
 139     $anusvara} [$tta$ttha$dda$ddha$nna]         > n\u0323;
 140     $anusvara} [$ta$tha$da$dha$na]              > n      ;
 141     $anusvara} [$pa$pha$ba$bha$ma]              > m      ;
 142     $anusvara} [$ya$ra$lla$la$va$ssa$sha$sa$ha] > n      ;
 143     $anusvara> m\u0307;
 144
 145     # Urdu compatibility
 146     $ya$nukta}$x        > y\u0307  ;
 147     $ya$nukta$virama    > y\u0307  ;
 148     $ya$nukta           > y\u0307a ;
 149
 150     $la$nukta }$x       > l\u0331  ;
 151     $la$nukta$virama    > l\u0331  ;
 152     $la$nukta           > l\u0331a ;
 153
 154     $na$nukta }$x       > n\u0331  ;
 155     $na$nukta$virama    > n\u0331  ;
 156     $na$nukta           > n\u0331a ;
 157
 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
 175     $ujha                > za       ;
 176     $ja$nukta }$x       > z        ;
 177     $ja$nukta$virama    > z        ;
 178     $ja$nukta           > za       ;
 179     $ddha$nukta}$x      > r\u0323h ;
 180     $ddha$nukta$virama  > r\u0323h ;
 181     $ddha$nukta         > r\u0323ha;
 182
 183     $uddha}$x           > r\u0323  ;
 184     $uddha$virama       > r\u0323 ;
 185     $uddha              > r\u0323a;
 186
 187     $udha               > r\u0323a ;
 188     $dda$nukta}$x       > r\u0323  ;
 189     $dda$nukta$virama   > r\u0323  ;
 190     $dda$nukta          > r\u0323a ;
 191     $pha$nukta }$x      > f  ;
 192     $pha$nukta$virama   > f  ;
 193     $pha$nukta          > fa ;
 194     $ufa }$x            > f  ;
 195     $ufa$virama         > f  ;
 196     $ufa                > fa ;
 197
 198     $ra$nukta}$x        > r\u0331;
 199     $ra$nukta$virama    > r\u0331;
 200     $ra$nukta           > r\u0331a;
 201     $lla$nukta}$x       > l\u0331;
 202     $lla$nukta$virama   > l\u0331;
 203     $lla$nukta          > l\u0331a;
 204
 205     $ela}$x             > l\u0331;
 206     $ela$virama         > l\u0331;
 207     $ela                > l\u0331a;
 208
 209     $uya}$x             > y\u0307;
 210     $uya$virama         > y\u0307;
 211     $uya                > y\u0307a;
 212
 213
 214     # normal consonants
 215     $ka$virama}$ha>k'';
 216     $ka}$x>k;
 217     $ka$virama>k;
 218     $ka>ka;
 219     $kha}$x>kh;
 220     $kha$virama>kh;
 221     $kha>kha;
 222     $ga$virama}$ha>g'';
 223     $ga}$x>g;
 224     $ga$virama>g;
 225     $ga>ga;
 226
 227     $gha}$x>gh;
 228     $gha$virama>gh;
 229     $gha>gha;
 230
 231     $nga}$x>n\u0307;
 232     $nga$virama>n\u0307;
 233     $nga>n\u0307a  ;
 234     $ca$virama}$ha>c'';
 235     $ca}$x>c;
 236     $ca$virama>c;
 237     $ca>ca;
 238
 239     $cha}$x>ch;
 240     $cha$virama>ch;
 241     $cha>cha;
 242     $ja$virama}$ha>j'';
 243     $ja}$x>j;
 244     $ja$virama>j;
 245     $ja>ja;
 246
 247     $jha}$x>jh;
 248     $jha$virama>jh;
 249     $jha>jha;
 250
 251     $nya }$x>n\u0303 ;
 252     $nya$virama>n\u0303;
 253     $nya > n\u0303a  ;
 254
 255
 256     $tta$virama}$ha>t\u0323'';
 257     $tta}$x>t\u0323;
 258     $tta$virama>t\u0323;
 259     $tta>t\u0323a;
 260
 261     $ttha}$x>t\u0323h;
 262     $ttha$virama>t\u0323h;
 263     $ttha>t\u0323ha;
 264     $dda}$x$ha>d\u0323'';
 265     $dda}$x>d\u0323;
 266     $dda$virama>d\u0323;
 267     $dda>d\u0323a;
 268
 269     $ddha}$x>d\u0323h;
 270     $ddha$virama>d\u0323h;
 271     $ddha>d\u0323ha;
 272
 273     $nna}$x>n\u0323  ;
 274     $nna$virama>n\u0323;
 275     $nna>n\u0323a   ;
 276
 277
 278     $ta$virama}$ha>t'';
 279     $ta$virama}$ttha>t'';
 280     $ta$virama}$tta>t'';
 281     $ta$virama}$tha>t'';
 282     $ta}$x>t;
 283     $ta$virama>t;
 284     $ta>ta;
 285     $tha}$x>th;
 286     $tha$virama>th;
 287     $tha>tha;
 288
 289     $da$virama}$ha>d'';
 290     $da$virama}$ddha>d'';
 291     $da$virama}$dda>d'';
 292     $da$virama}$dha>d'';
 293     $da}$x>d;
 294     $da$virama>d;
 295     $da>da;
 296     $dha}$x>dh;
 297     $dha$virama>dh;
 298     $dha>dha;
 299     $na$virama}$ga>n'';
 300     $na$virama}$ya>n'';
 301     $na}$x>n;
 302     $na$virama>n;
 303     $na>na;
 304
 305
 306     $pa$virama}$ha>p'';
 307     $pa}$x>p;
 308     $pa$virama>p;
 309     $pa>pa;
 310     $pha}$x>ph;
 311     $pha$virama>ph;
 312     $pha>pha;
 313     $ba$virama}$ha>b'';
 314     $ba}$x>b;
 315     $ba$virama>b;
 316     $ba>ba;
 317
 318     $bha}$x>bh;
 319     $bha$virama>bh;
 320     $bha>bha;
 321
 322     $ma$virama}$ma>m'';
 323     $ma}$x>m;
 324     $ma$virama>m;
 325     $ma>ma;
 326
 327     $ya}$x>y;
 328     $ya$virama>y;
 329     $ya>ya;
 330     $ra$virama}$ha>r'';
 331     $ra}$x>r;
 332     $ra$virama>r;
 333     $ra>ra;
 334     $vva$virama}$ha>w\u0307'';
 335     $vva}$x>w\u0307;
 336     $vva$virama>w\u0307;
 337     $vva>w\u0307a;
 338     $rra$virama}$ha>r\u0331'';
 339     $rra}$x>r\u0331;
 340     $rra$virama>r\u0331;
 341     $rra>r\u0331a;
 342     $la$virama}$ha>l'';
 343     $la}$x>l;
 344     $la$virama>l;
 345     $la>la;
 346     $lla$virama}$ha>l\u0323'';
 347     $lla}$x>l\u0323;
 348     $lla$virama>l\u0323;
 349     $lla>l\u0323a;
 350     $va}$x>v;
 351     $va$virama>v;
 352     $va>va;
 353     $sa$virama}$ha>s'';
 354     $sa$virama}$sha>s'';
 355     $sa$virama}$ssa>s'';
 356     $sa$virama}$sa>s'';
 357     $sa}$x>s;
 358     $sa$virama>s;
 359
 360     #for gurmukhi
 361     $sa$nukta}$x>s\u0301;
 362     $sa$nukta$virama>s\u0301;
 363     $sa$nukta>s\u0301a;
 364     $sa>sa;
 365
 366     $sha}$x>s\u0301;
 367     $sha$virama>s\u0301;
 368     $sha>s\u0301a;
 369
 370     $ssa}$x>s\u0323;
 371     $ssa$virama>s\u0323;
 372     $ssa>s\u0323a;
 373     $ha}$x>h;
 374     $ha$virama>h;
 375     $ha>ha;
 376
 377     # dependent vowels (should never occur except following consonants)
 378     $forceIndependentMatra{$aa  > \u0314a\u0304       ;
 379     $forceIndependentMatra{$ai  > \u0314ai            ;
 380     $forceIndependentMatra{$au  > \u0314au            ;
 381     $forceIndependentMatra{$ii  > \u0314i\u0304       ;
 382     $forceIndependentMatra{$i   > \u0314i             ;
 383     $forceIndependentMatra{$uu  > \u0314u\u0304       ;
 384     $forceIndependentMatra{$u   > \u0314u             ;
 385     $forceIndependentMatra{$rrh > \u0314r\u0325\u0304 ;
 386     $forceIndependentMatra{$rh  > \u0314r\u0325       ;
 387     $forceIndependentMatra{$llh > \u0314l\u0325\u0304 ;
 388     $forceIndependentMatra{$lh  > \u0314l\u0325       ;
 389     $forceIndependentMatra{$e   > \u0314e\u0304       ;
 390     $forceIndependentMatra{$o   > \u0314o\u0304       ;
 391     #extra vowels
 392     $forceIndependentMatra{$ce  > \u0314e\u0306       ;
 393     $forceIndependentMatra{$co  > \u0314o\u0306       ;
 394     $forceIndependentMatra{$se  > \u0314e             ;
 395     $forceIndependentMatra{$so  > \u0314o             ;
 396     $forceIndependentMatra{$nukta  >; # Nukta cannot appear independently or as first character
 397     $forceIndependentMatra{$virama >; # Virama cannot appear independently or as first character
 398     $aa  > a\u0304       ;
 399     $ai  > ai            ;
 400     $au  > au            ;
 401     $ii  > i\u0304       ;
 402     $i   > i             ;
 403     $uu  > u\u0304       ;
 404     $u   > u             ;
 405     $rrh > r\u0325\u0304 ;
 406     $rh  > r\u0325       ;
 407     $llh > l\u0325\u0304 ;
 408     $lh  > l\u0325       ;
 409     $e   > e\u0304       ;
 410     $o   > o\u0304       ;
 411     #extra vowels
 412     $ce  > e\u0306       ;
 413     $co  > o\u0306       ;
 414     $se  > e             ;
 415     $so  > o             ;
 416     #dependent vowels when following independent vowels. Generally Illegal only for roundtripping
 417     $waa} $x > a\u0304\u0314        ;
 418     $wai} $x > ai\u0314             ;
 419     $wau} $x > au\u0314             ;
 420     $wii} $x > i\u0304\u0314        ;
 421     $wi } $x > i\u0314              ;
 422     $wuu} $x > u\u0304\u0314        ;
 423     $wu } $x > u\u0314              ;
 424     $wrr} $x > r\u0325\u0304\u0314  ;
 425     $wr } $x > r\u0325\u0314        ;
 426     $wll} $x > l\u0325\u0304\u0314  ;
 427     $wl } $x > l\u0325\u0314        ;
 428     $we } $x > e\u0304\u0314        ;
 429     $wo } $x > o\u0304\u0314        ;
 430     $wa } $x > a\u0314              ;
 431     #extra vowels
 432     $wce} $x > e\u0306\u0314        ;
 433     $wco} $x > o\u0306\u0314        ;
 434     $wse} $x > e\u0314              ;
 435     $wso} $x > o\u0314              ;
 436     $om} $x > ''om\u0314            ;
 437
 438     # independent vowels when preceeded by vowels
 439     $vowels{$waa    > ''a\u0304         ;
 440     $vowels{$wai    > ''ai              ;
 441     $vowels{$wau    > ''au              ;
 442     $vowels{$wii     > ''i\u0304        ;
 443     $vowels{$wi     > ''i               ;
 444     $vowels{$wuu    > ''u\u0304         ;
 445     $vowels{$wu     > ''u               ;
 446     $vowels{$wrr    > ''r\u0325\u0304   ;
 447     $vowels{$wr     > ''r\u0325         ;
 448     $vowels{$wll    > ''l\u0325\u0304   ;
 449     $vowels{$wl     > ''l\u0325         ;
 450     $vowels{$we     > ''e\u0304         ;
 451     $vowels{$wo     > ''o\u0304         ;
 452     $vowels{$wa     > ''a               ;
 453     #extra vowels
 454     $vowels{$wce    > ''e\u0306         ;
 455     $vowels{$wco    > ''o\u0306         ;
 456     $vowels{$wse    > ''e               ;
 457     $vowels{$wso    > ''o               ;
 458
 459     # independent vowels (otherwise)
 460     $waa > a\u0304          ;
 461     $wai > ai               ;
 462     $wau > au               ;
 463     $wii > i\u0304          ;
 464     $wi  > i                ;
 465     $wuu > u\u0304          ;
 466     $wu  > u                ;
 467     $wrr > r\u0325\u0304    ;
 468     $wr  > r\u0325          ;
 469     $wll > l\u0325\u0304    ;
 470     $wl  > l\u0325          ;
 471     $we  > e\u0304          ;
 472     $wo  > o\u0304          ;
 473     $wa  > a                ;
 474     #extra vowels
 475     $wce > e\u0306          ;
 476     $wco > o\u0306          ;
 477     $wse > e                ;
 478     $wso > o                ;
 479     $om > ''om              ;
 480
 481     #stress marks
 482     $avagraha > \u0315;
 483     $chandrabindu$anusvara>\u0303;
 484     $chandrabindu > m\u0310;
 485     $visarga>h\u0323;
 486     #numbers
 487     $zero  > 0;
 488     $one   > 1;
 489     $two   > 2;
 490     $three > 3;
 491     $four  > 4;
 492     $five  > 5;
 493     $six   > 6;
 494     $seven > 7;
 495     $eight > 8;
 496     $nine  > 9;
 497     $lm   >;
 498     $ailm >;
 499     $aulm >;
 500
 501     $danda>'.';
 502     $doubleDanda>'.';
 503
 504     \ue070>;       # ABBREVIATION SIGN
 505     # LETTER RA WITH MIDDLE DIAGONAL
 506         \ue071}$x>ra;
 507         \ue071$virama>r;
 508         \ue071>ra;
 509         # LETTER RA WITH LOWER DIAGONAL
 510         \ue072}$x>ra;
 511         \ue072$virama>r;
 512         \ue072>ra;
 513
 514         \ue073>;       # RUPEE MARK
 515         \ue074>;       # RUPEE SIGN
 516         \ue075>;       # CURRENCY NUMERATOR ONE
 517         \ue076>;       # CURRENCY NUMERATOR TWO
 518         \ue077>;       # CURRENCY NUMERATOR THREE
 519         \ue078>;       # CURRENCY NUMERATOR FOUR
 520         \ue079>;       # CURRENCY NUMERATOR ONE LESS THAN THE DENOMINATOR
 521         \ue07A>;       # CURRENCY DENOMINATOR SIXTEEN
 522         \ue07B>;       # ISSHAR
 523         \uE07C>;       # TIPPI
 524         \uE07D>;       # ADDAK
 525         \uE07E>;       # IRI
 526         \uE07F>;       # URA
 527         \uE080>;       # EK ONKAR
 528         \uE004>;       # DEVANAGARI VOWEL SIGN SHORT A
 529