icuSources/data/translit/t_Latn_InterIndic.txt

   1  // -*- Coding: utf-8; -*-
   2 //--------------------------------------------------------------------
   3 // Copyright (c) 1999-2002, International Business Machines
   4 // Corporation and others.  All Rights Reserved.
   5 //--------------------------------------------------------------------
   6 // THIS IS A MACHINE-GENERATED FILE
   7 // Tool: dumpicurules.bat
   8 // Source: ../../../impl/data/Transliterator_Latin_InterIndic.txt
   9 // Date: Sat Jul 27 10:31:07 2002
  10 //--------------------------------------------------------------------
  11
  12 // Latin_InterIndic
  13
  14 t_Latn_InterIndic {
  15   Rule {
  16 //--------------------------------------------------------------------
  17 //--------------------------------------------------------------------
  18 //--------------------------------------------------------------------
  19
  20 // Latin-InterIndic
  21     //:: NFD;
  22     //\u0e00 reserved
  23     //consonants
  24     "$chandrabindu=\ue001;"
  25     "$anusvara=\ue002;"
  26     "$visarga=\ue003;"
  27     //\u0e004 reserved
  28     // w<vowel> represents the stand-alone form
  29     "$wa=\ue005;"
  30     "$waa=\ue006;"
  31     "$wi=\ue007;"
  32     "$wii=\ue008;"
  33     "$wu=\ue009;"
  34     "$wuu=\ue00a;"
  35     "$wr=\ue00b;"
  36     "$wl=\ue00c;"
  37     "$wce=\ue00d;" // LETTER CANDRA E
  38     "$wse=\ue00e;" // LETTER SHORT E
  39     "$we=\ue00f;"  // \u090f LETTER E
  40     "$wai=\ue010;"
  41     "$wco=\ue011;" // LETTER CANDRA O
  42     "$wso=\ue012;" // LETTER SHORT O
  43     "$wo=\ue013;"  // \u0913 LETTER O
  44     "$wau=\ue014;"
  45     "$ka=\ue015;"
  46     "$kha=\ue016;"
  47     "$ga=\ue017;"
  48     "$gha=\ue018;"
  49     "$nga=\ue019;"
  50     "$ca=\ue01a;"
  51     "$cha=\ue01b;"
  52     "$ja=\ue01c;"
  53     "$jha=\ue01d;"
  54     "$nya=\ue01e;"
  55     "$tta=\ue01f;"
  56     "$ttha=\ue020;"
  57     "$dda=\ue021;"
  58     "$ddha=\ue022;"
  59     "$nna=\ue023;"
  60     "$ta=\ue024;"
  61     "$tha=\ue025;"
  62     "$da=\ue026;"
  63     "$dha=\ue027;"
  64     "$na=\ue028;"
  65     "$ena=\ue029;" //compatibility
  66     "$pa=\ue02a;"
  67     "$pha=\ue02b;"
  68     "$ba=\ue02c;"
  69     "$bha=\ue02d;"
  70     "$ma=\ue02e;"
  71     "$ya=\ue02f;"
  72     "$ra=\ue030;"
  73     "$rra=\ue031;"
  74     "$la=\ue032;"
  75     "$lla=\ue033;"
  76     "$ela=\ue034;" //compatibility
  77     "$va=\ue035;"
  78     "$sha=\ue036;"
  79     "$ssa=\ue037;"
  80     "$sa=\ue038;"
  81     "$ha=\ue039;"
  82 //\u093a Reserved
  83 //\u093b Reserved
  84     "$nukta=\ue03c;"
  85     "$avagraha=\ue03d;" // SIGN AVAGRAHA
  86     // <vowel> represents the dependent form
  87     "$aa=\ue03e;"
  88     "$i=\ue03f;"
  89     "$ii=\ue040;"
  90     "$u=\ue041;"
  91     "$uu=\ue042;"
  92     "$rh=\ue043;"
  93     "$lh=\ue044;"
  94     "$ce=\ue045;" //VOWEL SIGN CANDRA E
  95     "$se=\ue046;" //VOWEL SIGN SHORT E
  96     "$e=\ue047;"
  97     "$ai=\ue048;"
  98     "$co=\ue049;" // VOWEL SIGN CANDRA O
  99     "$so=\ue04a;" // VOWEL SIGN SHORT O
 100     "$o=\ue04b;"  // \u094b
 101     "$au=\ue04c;"
 102     "$virama=\ue04d;"
 103 // \u094e Reserved
 104 // \u094f Reserved
 105    "$om = \ue050;" // OM
 106 // \u0951>;        # UNMAPPED STRESS SIGN UDATTA
 107 // \u0952>;        # UNMAPPED STRESS SIGN ANUDATTA
 108 // \u0953>;        # UNMAPPED GRAVE ACCENT
 109 // \u0954>;        # UNMAPPED ACUTE ACCENT
 110     "$lm = \ue055;"//  Telugu Length Mark
 111     "$ailm=\ue056;"//  AI Length Mark
 112     "$aulm=\ue057;"//  AU Length Mark
 113      //urdu compatibity forms
 114     "$uka=\ue058;"
 115     "$ukha=\ue059;"
 116     "$ugha=\ue05a;"
 117     "$ujha=\ue05b;"
 118     "$uddha=\ue05c;"
 119     "$udha=\ue05d;"
 120     "$ufa=\ue05e;"
 121     "$uya=\ue05f;"
 122     "$wrr=\ue060;"
 123     "$wll=\ue061;"
 124     "$rrh=\ue062;"
 125     "$llh=\ue063;"
 126     "$danda=\ue064;"
 127     "$doubleDanda=\ue065;"
 128     "$zero=\ue066;"     // DIGIT ZERO
 129     "$one=\ue067;"      // DIGIT ONE
 130     "$two=\ue068;"      // DIGIT TWO
 131     "$three=\ue069;"    // DIGIT THREE
 132     "$four=\ue06a;"     // DIGIT FOUR
 133     "$five=\ue06b;"     // DIGIT FIVE
 134     "$six=\ue06c;"      // DIGIT SIX
 135     "$seven=\ue06d;"    // DIGIT SEVEN
 136     "$eight=\ue06e;"    // DIGIT EIGHT
 137     "$nine=\ue06f;"     // DIGIT NINE
 138     // For all other scripts
 139     "$ecp0=\ue070;"
 140     "$ecp1=\ue071;"
 141     "$ecp2=\ue072;"
 142     "$ecp3=\ue073;"
 143     "$ecp4=\ue074;"
 144     "$ecp5=\ue075;"
 145     "$ecp6=\ue076;"
 146     "$ecp7=\ue077;"
 147     "$ecp8=\ue078;"
 148     "$ecp9=\ue079;"
 149     "$ecpA=\ue07a;"
 150     "$ecpB=\ue07b;"
 151     "$ecpC=\ue07c;"
 152     "$ecpD=\ue07d;"
 153     "$ecpE=\ue07e;"
 154     "$ecpF=\ue07f;"
 155 // \u0970>; # UNMAPPED ABBREVIATION SIGN
 156     "$depVowelAbove=[\ue03e-\ue040\ue045-\ue04c];"
 157     "$depVowelBelow=[\ue041-\ue044];"
 158     "$endThing=[$danda$doubleDanda];"
 159     // $x was originally called '&'; $z was '%'
 160     "$x=[$virama$aa$ai$au$ii$i$uu$u$rrh$rh$lh$e$o$se$ce$so$co];"
 161     "$z=[bcdfghjklmnpqrstvwxyz];"
 162     "$consonants=[[$ka-$ha]$z[\u0915-\u0939][\u0995-\u09b9][\u0a15-\u0a39][\u0a95-\u0ab9][\u0b15-\u0b39][\u0b95-\u0bb9][\u0c15-\u0c39][\u0c95-\u0cb9][\u0d15-\u0d39]];"
 163     "\u0315 > $avagraha;"
 164     "\u0303>$chandrabindu$anusvara;"
 165     "m\u0310>$chandrabindu;"
 166     "h\u0323>$visarga;"
 167     "x>$ka$virama$sa;"
 168 // convert to independent forms at start of word or syllable:
 169 // dependent forms for roundtrip
 170     "\u0314a\u0304>$aa;"
 171     "\u0314ai>$ai;"
 172     "\u0314au>$au;"
 173     "\u0314ii>$ii;"
 174     "\u0314i\u0304>$ii;"
 175     "\u0314i>$i;"
 176     "\u0314u\u0304>$uu;"
 177     "\u0314u>$u;"
 178     "\u0314r\u0325\u0304>$rrh;"
 179     "\u0314r\u0325>$rh;"
 180     "\u0314l\u0325\u0304>$llh;"
 181     "\u0314lh>$lh;"
 182     "\u0314l\u0325>$lh;"
 183     "\u0314e\u0304>$e;"
 184     "\u0314o\u0304>$o;"
 185     "\u0314a>;"
 186     "\u0314e\u0306>$ce;"
 187     "\u0314o\u0306>$co;"
 188     "\u0314e>$se;"
 189     "\u0314o>$so;"
 190
 191 // preceeded by consonants
 192     "$consonants{ a\u0304>$aa;"
 193     "$consonants{ ai>$ai;"
 194     "$consonants{ au>$au;"
 195     "$consonants{ ii>$ii;"
 196     "$consonants{ i\u0304>$ii;"
 197     "$consonants{ i>$i;"
 198     "$consonants{ u\u0304>$uu;"
 199     "$consonants{ u>$u;"
 200     "$consonants{ r\u0325\u0304>$rrh;"
 201     "$consonants{ r\u0325a>$rh;"
 202     "$consonants{ r\u0325>$rh;"
 203     "$consonants{ l\u0325\u0304>$llh;"
 204     "$consonants{ lh>$lh;"
 205     "$consonants{ l\u0325>$lh;"
 206     "$consonants{ e\u0304>$e;"
 207     "$consonants{ o\u0304>$o;"
 208     "$consonants{ e\u0306>$ce;"
 209     "$consonants{ o\u0306>$co;"
 210     "$consonants{ e>$se;"
 211     "$consonants{ o>$so;"
 212
 213 // e.g. keai -> {ka}{e}{wai}; k'ai -> {ka}{wai}; (ai) -> ({wai})
 214     "a\u0304>$waa;"
 215     "ai>$wai;"
 216     "au>$wau;"
 217     "i\u0304>$wii;"
 218     "i>$wi;"
 219     "u\u0304>$wuu;"
 220     "u>$wu;"
 221     "r\u0325\u0304>$wrr;"
 222     "r\u0325>$wr;"
 223     "l\u0325\u0304>$wll;"
 224     "lh>$wl;"
 225     "l\u0325>$wl;"
 226     "e\u0304>$we;"
 227     "o\u0304>$wo;"
 228     "a>$wa;"
 229     "e\u0306>$wce;"
 230     "o\u0306>$wco;"
 231     "e>$wse;"
 232     "''om>$om;"
 233     "o>$wso;"
 234
 235     // rules for anusvara
 236     "n}r\u0325           > $na|$virama;"
 237     "n}l\u0325           > $na|$virama;"
 238     "n}na                > $na|$virama;"
 239     "n\u0307}[kg]        > $anusvara;"
 240     "n\u0307}n\u0307     > $anusvara;"
 241     "n\u0304}[cj]        > $anusvara;"
 242     "n\u0304}n\u0303     > $anusvara;"
 243     "n\u0323}[tdn]\u0323 > $anusvara;"
 244     "n}[tdn]             > $anusvara;"
 245     "m}[pbm]             > $anusvara;"
 246     "n}[ylvshr]          > $anusvara;"
 247     "m\u0307             > $anusvara;"
 248
 249     //urdu compatibility
 250     "q>$uka|$virama;"
 251     "k\u0331h\u0331>$ukha |$virama;"
 252     "g\u0307> $ugha | $virama;"
 253     "z > $ujha |$virama;"
 254     "f > $ufa|$virama;"
 255
 256     // dev
 257     "y\u0307>$uya|$virama;"
 258     "l\u0331>$ela|$virama;"
 259     "n\u0331>$ena|$virama;"
 260     "n\u0307>$nga|$virama;"
 261     "n\u0303>$nya|$virama;"
 262     "n\u0323>$nna|$virama;"
 263     "t\u0323h>$ttha|$virama;"
 264     "t\u0323>$tta|$virama;"
 265     "r\u0323h>$udha|$virama;"
 266     "r\u0323>$uddha|$virama;"
 267     "d\u0323h>$ddha|$virama;"
 268     "d\u0323>$dda|$virama;"
 269     "kh>$kha|$virama;"
 270     "k>$ka|$virama;"
 271     "gh>$gha|$virama;"
 272     "g>$ga|$virama;"
 273     "ch>$cha|$virama;"
 274     "c>$ca|$virama;"
 275     "jh>$jha|$virama;"
 276     "j>$ja|$virama;"
 277     "ny>$nya|$virama;"
 278     "tth>$ttha|$virama;"
 279     "ddh>$ddha|$virama;"
 280     "th>$tha|$virama;"
 281     "t>$ta|$virama;"
 282     "dh>$dha|$virama;"
 283     "d>$da|$virama;"
 284     "n>$na|$virama;"
 285     "ph>$pha|$virama;"
 286     "p>$pa|$virama;"
 287     "bh>$bha|$virama;"
 288     "b>$ba|$virama;"
 289     "m>$ma|$virama;"
 290     "y>$ya|$virama;"
 291     "r\u0331>$rra|$virama;"
 292     "r>$ra|$virama;"
 293     "l\u0323>$lla|$virama;"
 294     "l>$la|$virama;"
 295     "v>$va|$virama;"
 296     "w>$va|$virama;"
 297     "sh>$sha|$virama;"
 298     "ss>$ssa|$virama;"
 299     "s\u0323>$ssa|$virama;"
 300     "s\u0301>$sha|$virama;"
 301     "s>$sa|$virama;"
 302     "h>$ha|$virama;"
 303     "'.'>$danda;"
 304     "$danda'.'>$doubleDanda;"
 305     "$depVowelAbove{'~'>$anusvara;"
 306     "$depVowelBelow{'~'>$chandrabindu;"
 307 // convert to dependent forms after consonant with no vowel:
 308 // e.g. kai -> {ka}{virama}ai -> {ka}{ai}
 309     //$virama aa>$aa;
 310     "$virama a\u0304>$aa;"
 311     "$virama ai>$ai;"
 312     "$virama au>$au;"
 313     "$virama ii>$ii;"
 314     "$virama i\u0304>$ii;"
 315     "$virama i>$i;"
 316     //$virama uu>$uu;
 317     "$virama u\u0304>$uu;"
 318     "$virama u>$u;"
 319     //$virama rrh>$rrh;
 320     "$virama r\u0325\u0304>$rrh;"
 321     //$virama rh>$rh;
 322     "$virama r\u0325a>$rh;"
 323     "$virama r\u0325>$rh;"
 324     "$virama l\u0325\u0304>$llh;"
 325     "$virama lh>$lh;"
 326     "$virama l\u0325>$lh;"
 327     "$virama e\u0304>$e;"
 328     "$virama o\u0304>$o;"
 329     "$virama a>;"
 330     "$virama e\u0306>$ce;"
 331     "$virama o\u0306>$co;"
 332     "$virama e>$se;"
 333     "$virama o>$so;"
 334
 335
 336 // otherwise convert independent forms when separated by ': k'ai -> {ka}{virama}{wai}
 337     //$virama''aa>$waa;
 338     "$virama''a\u0304>$waa;"
 339     "$virama''ai>$wai;"
 340     "$virama''au>$wau;"
 341     //$virama''ii>$wii;
 342     "$virama''i\u0304>$wii;"
 343     "$virama''i>$wi;"
 344     //$virama''uu>$wuu;
 345     "$virama''u\u0304>$wuu;"
 346     "$virama''u>$wu;"
 347     //$virama''rrh>$wrr;
 348     "$virama''r\u0325\u0304>$wrr;"
 349     //$virama''rh>$wr;
 350     "$virama''r\u0325>$wr;"
 351     "$virama''l\u0325\u0304>$wll;"
 352     //$virama''lh>$wl;
 353     "$virama''l\u0325>$wl;"
 354     "$virama''e\u0304>$we;"
 355     "$virama''o\u0304>$wo;"
 356     "$virama''a>$wa;"
 357     "$virama''e\u0306>$wce;"
 358     "$virama''o\u0306>$wco;"
 359     "$virama''e>$wse;"
 360     "$virama''o>$wso;"
 361 // no virama
 362     "''a\u0304>$waa;"
 363     "''ai>$wai;"
 364     "''au>$wau;"
 365     "''i\u0304>$wii;"
 366     "''i>$wi;"
 367     "''u\u0304>$wuu;"
 368     "''u>$wu;"
 369     "''r\u0325\u0304>$wrr;"
 370     "''r\u0325>$wr;"
 371     "''l\u0325\u0304>$wll;"
 372     "''l\u0325>$wl;"
 373     "''e\u0304>$we;"
 374     "''o\u0304>$wo;"
 375     "''a>$wa;"
 376     "''e\u0306>$wce;"
 377     "''o\u0306>$wco;"
 378     "''e>$wse;"
 379     "''o>$wso;"
 380
 381     "$virama } [$z] > $virama;"
 382     "$virama } ' ' > $virama ;"
 383     "$virama}$endThing>;"
 384     "0>$zero;"
 385     "1>$one;"
 386     "2>$two;"
 387     "3>$three;"
 388     "4>$four;"
 389     "5>$five;"
 390     "6>$six;"
 391     "7>$seven;"
 392     "8>$eight;"
 393     "9>$nine;"
 394     "''>;"
 395     //:: NFC (NFD) ;
 396   }
 397 }