• Home
  • Line#
  • Scopes#
  • Navigate#
  • Raw
  • Download
1# © 2016 and later: Unicode, Inc. and others.
2# License & terms of use: http://www.unicode.org/copyright.html
3# Generated using tools/cldr/cldr-to-icu/build-icu-data.xml
4#
5# File: Latin_InterIndic.txt
6# Generated from CLDR
7#
8
9# Latin-InterIndic
10#:: NFD;
11#\u0E00 reserved
12#consonants
13$chandrabindu=\uE001;
14$anusvara=\uE002;
15$visarga=\uE003;
16#\u0E004 reserved
17# w←vowel→ represents the stand-alone form
18$wa=\uE005;
19$waa=\uE006;
20$wi=\uE007;
21$wii=\uE008;
22$wu=\uE009;
23$wuu=\uE00A;
24$wr=\uE00B;
25$wl=\uE00C;
26$wce=\uE00D; # LETTER CANDRA E
27$wse=\uE00E; # LETTER SHORT E
28$we=\uE00F;  # ए LETTER E
29$wai=\uE010;
30$wco=\uE011; # LETTER CANDRA O
31$wso=\uE012; # LETTER SHORT O
32$wo=\uE013;  # ओ LETTER O
33$wau=\uE014;
34$ka=\uE015;
35$kha=\uE016;
36$ga=\uE017;
37$gha=\uE018;
38$nga=\uE019;
39$ca=\uE01A;
40$cha=\uE01B;
41$ja=\uE01C;
42$jha=\uE01D;
43$nya=\uE01E;
44$tta=\uE01F;
45$ttha=\uE020;
46$dda=\uE021;
47$ddha=\uE022;
48$nna=\uE023;
49$ta=\uE024;
50$tha=\uE025;
51$da=\uE026;
52$dha=\uE027;
53$na=\uE028;
54$ena=\uE029; #compatibility
55$pa=\uE02A;
56$pha=\uE02B;
57$ba=\uE02C;
58$bha=\uE02D;
59$ma=\uE02E;
60$ya=\uE02F;
61$ra=\uE030;
62$rra=\uE031;
63$la=\uE032;
64$lla=\uE033;
65$ela=\uE034; #compatibility
66$va=\uE035;
67$vva=\uE081;
68$sha=\uE036;
69$ssa=\uE037;
70$sa=\uE038;
71$ha=\uE039;
72#\u093A Reserved
73#\u093B Reserved
74$nukta=\uE03C;
75$avagraha=\uE03D; # SIGN AVAGRAHA
76# ←vowel→ represents the dependent form
77$aa=\uE03E;
78$i=\uE03F;
79$ii=\uE040;
80$u=\uE041;
81$uu=\uE042;
82$rh=\uE043;
83$rrh=\uE044;
84$ce=\uE045; #VOWEL SIGN CANDRA E
85$se=\uE046; #VOWEL SIGN SHORT E
86$e=\uE047;
87$ai=\uE048;
88$co=\uE049; # VOWEL SIGN CANDRA O
89$so=\uE04A; # VOWEL SIGN SHORT O
90$o=\uE04B;  # ो
91$au=\uE04C;
92$virama=\uE04D;
93# \u094E Reserved
94# \u094F Reserved
95$om = \uE050; # OM
96# \u0951→;        # UNMAPPED STRESS SIGN UDATTA
97# \u0952→;        # UNMAPPED STRESS SIGN ANUDATTA
98# \u0953→;        # UNMAPPED GRAVE ACCENT
99# \u0954→;        # UNMAPPED ACUTE ACCENT
100$lm = \uE055;#  Telugu Length Mark
101$ailm=\uE056;#  AI Length Mark
102$aulm=\uE057;#  AU Length Mark
103#urdu compatibility forms
104$uka=\uE058;
105$ukha=\uE059;
106$ugha=\uE05A;
107$ujha=\uE05B;
108$uddha=\uE05C;
109$udha=\uE05D;
110$ufa=\uE05E;
111$uya=\uE05F;
112$wrr=\uE060;
113$wll=\uE061;
114$lh=\uE062;
115$llh=\uE063;
116$danda=\uE064;
117$doubleDanda=\uE065;
118$zero=\uE066;     # DIGIT ZERO
119$one=\uE067;      # DIGIT ONE
120$two=\uE068;      # DIGIT TWO
121$three=\uE069;    # DIGIT THREE
122$four=\uE06A;     # DIGIT FOUR
123$five=\uE06B;     # DIGIT FIVE
124$six=\uE06C;      # DIGIT SIX
125$seven=\uE06D;    # DIGIT SEVEN
126$eight=\uE06E;    # DIGIT EIGHT
127$nine=\uE06F;     # DIGIT NINE
128$dgs=\uE082;
129# For all other scripts
130$ecp0=\uE070;
131$ecp1=\uE071;
132$ecp2=\uE072;
133$ecp3=\uE073;
134$ecp4=\uE074;
135$ecp5=\uE075;
136$ecp6=\uE076;
137$ecp7=\uE077;
138$ecp8=\uE078;
139$ecp9=\uE079;
140$ecpA=\uE07A;
141$ecpB=\uE07B;
142$ecpC=\uE07C;
143$ecpD=\uE07D;
144$ecpE=\uE07E;
145$ecpF=\uE07F;
146# Khanda-ta
147$kta=\uE083;
148# ॰→; # nothing in Latin maps to InterIndic ABBREVIATION SIGN
149$depVowelAbove=[\uE03E-\uE040\uE045-\uE04C];
150$depVowelBelow=[\uE041-\uE044];
151$endThing=[$danda$doubleDanda];
152# $x was originally called '§'; $z was '%'
153$x=[$virama$aa$ai$au$ii$i$uu$u$rrh$rh$lh$e$o$se$ce$so$co];
154$z=[bcdfghjklmnpqrstvwxyz];
155$consonants=[[$ka-$ha]$z[क-ह][ক-হ][ਕ-ਹ][ક-હ][କ-ହ][க-ஹ][క-హ][ಕ-ಹ][ക-ഹ]];
156\u0315 → $avagraha;
157\u0303→$chandrabindu$anusvara;
158m\u0310→$chandrabindu;
159h\u0323→$visarga;
160x→$ka$virama$sa;
161# convert to independent forms at start of word or syllable:
162# dependent forms for roundtrip
163\u0314a\u0304→$aa;
164\u0314ai→$ai;
165\u0314au→$au;
166\u0314ii→$ii;
167\u0314i\u0304→$ii;
168\u0314i→$i;
169\u0314u\u0304→$uu;
170\u0314u→$u;
171\u0314r\u0325\u0304→$rrh;
172\u0314r\u0325→$rh;
173\u0314l\u0325\u0304→$llh;
174\u0314lh→$lh;
175\u0314l\u0325→$lh;
176\u0314e\u0304→$e;
177\u0314o\u0304→$o;
178\u0314a→;
179\u0314e\u0306→$ce;
180\u0314o\u0306→$co;
181\u0314e→$se;
182\u0314o→$so;
183# preceded by consonants
184$consonants{ a\u0304→$aa;
185$consonants{ ai→$ai;
186$consonants{ au→$au;
187$consonants{ ii→$ii;
188$consonants{ i\u0304→$ii;
189$consonants{ i→$i;
190$consonants{ u\u0304→$uu;
191$consonants{ u→$u;
192$consonants{ r\u0325\u0304→$rrh;
193$consonants{ r\u0325a→$rh;
194$consonants{ r\u0325→$rh;
195$consonants{ l\u0325\u0304→$llh;
196$consonants{ lh→$lh;
197$consonants{ l\u0325→$lh;
198$consonants{ e\u0304→$e;
199$consonants{ o\u0304→$o;
200$consonants{ e\u0306→$ce;
201$consonants{ o\u0306→$co;
202$consonants{ e→$se;
203$consonants{ o→$so;
204# e.g. keai -→ {ka}{e}{wai}; k'ai -→ {ka}{wai}; (ai) -→ ({wai})
205a\u0304→$waa;
206ai→$wai;
207au→$wau;
208i\u0304→$wii;
209i→$wi;
210u\u0304→$wuu;
211u→$wu;
212r\u0325\u0304→$wrr;
213r\u0325→$wr;
214l\u0325\u0304→$wll;
215lh→$wl;
216l\u0325→$wl;
217e\u0304→$we;
218o\u0304→$wo;
219a→$wa;
220e\u0306→$wce;
221o\u0306→$wco;
222e→$wse;
223''om→$om;
224o→$wso;
225# rules for anusvara
226n}r\u0325           → $na|$virama;
227n}l\u0325           → $na|$virama;
228n}na                → $na|$virama;
229n\u0307}[kg]        → $anusvara;
230n\u0307}n\u0307     → $anusvara;
231n\u0304}[cj]        → $anusvara;
232n\u0304}n\u0303     → $anusvara;
233n\u0323}[tdn]\u0323 → $anusvara;
234n}[tdn]             → $anusvara;
235m}[pbm]             → $anusvara;
236n}[ylvshr]          → $anusvara;
237m\u0307             → $anusvara;
238#urdu compatibility
239q→$uka|$virama;
240k\u0331h\u0331→$ukha |$virama;
241g\u0307→ $ugha | $virama;
242z → $ujha |$virama;
243f → $ufa|$virama;
244t\u0331→$kta;
245# dev
246y\u0307→$uya|$virama;
247l\u0331→$ela|$virama;
248n\u0331→$ena|$virama;
249n\u0307→$nga|$virama;
250n\u0303→$nya|$virama;
251n\u0323→$nna|$virama;
252t\u0323h→$ttha|$virama;
253t\u0323→$tta|$virama;
254r\u0323h→$udha|$virama;
255r\u0323→$uddha|$virama;
256d\u0323h→$ddha|$virama;
257d\u0323→$dda|$virama;
258kh→$kha|$virama;
259k→$ka|$virama;
260gh→$gha|$virama;
261g→$ga|$virama;
262ch→$cha|$virama;
263c→$ca|$virama;
264jh→$jha|$virama;
265j→$ja|$virama;
266ny→$nya|$virama;
267tth→$ttha|$virama;
268ddh→$ddha|$virama;
269th→$tha|$virama;
270t→$ta|$virama;
271dh→$dha|$virama;
272d→$da|$virama;
273n→$na|$virama;
274ph→$pha|$virama;
275p→$pa|$virama;
276bh→$bha|$virama;
277b→$ba|$virama;
278m→$ma|$virama;
279y→$ya|$virama;
280r\u0331→$rra|$virama;
281r→$ra|$virama;
282l\u0323→$lla|$virama;
283l→$la|$virama;
284v→$va|$virama;
285w\u0307→$vva|$virama;
286w→$va|$virama;
287sh→$sha|$virama;
288ss→$ssa|$virama;
289s\u0323→$ssa|$virama;
290s\u0301→$sha|$virama;
291s→$sa|$virama;
292h→$ha|$virama;
293'.'→$danda;
294$danda'.'→$doubleDanda;
295$depVowelAbove{'~'→$anusvara;
296$depVowelBelow{'~'→$chandrabindu;
297# convert to dependent forms after consonant with no vowel:
298# e.g. kai -→ {ka}{virama}ai -→ {ka}{ai}
299#$virama aa→$aa;
300$virama a\u0304→$aa;
301$virama ai→$ai;
302$virama au→$au;
303$virama ii→$ii;
304$virama i\u0304→$ii;
305$virama i→$i;
306#$virama uu→$uu;
307$virama u\u0304→$uu;
308$virama u→$u;
309#$virama rrh→$rrh;
310$virama r\u0325\u0304→$rrh;
311#$virama rh→$rh;
312$virama r\u0325a→$rh;
313$virama r\u0325→$rh;
314$virama l\u0325\u0304→$llh;
315$virama lh→$lh;
316$virama l\u0325→$lh;
317$virama e\u0304→$e;
318$virama o\u0304→$o;
319$virama a→;
320$virama e\u0306→$ce;
321$virama o\u0306→$co;
322$virama e→$se;
323$virama o→$so;
324# otherwise convert independent forms when separated by ': k'ai -→ {ka}{virama}{wai}
325#$virama''aa→$waa;
326$virama''a\u0304→$waa;
327$virama''ai→$wai;
328$virama''au→$wau;
329#$virama''ii→$wii;
330$virama''i\u0304→$wii;
331$virama''i→$wi;
332#$virama''uu→$wuu;
333$virama''u\u0304→$wuu;
334$virama''u→$wu;
335#$virama''rrh→$wrr;
336$virama''r\u0325\u0304→$wrr;
337#$virama''rh→$wr;
338$virama''r\u0325→$wr;
339$virama''l\u0325\u0304→$wll;
340#$virama''lh→$wl;
341$virama''l\u0325→$wl;
342$virama''e\u0304→$we;
343$virama''o\u0304→$wo;
344$virama''a→$wa;
345$virama''e\u0306→$wce;
346$virama''o\u0306→$wco;
347$virama''e→$wse;
348$virama''o→$wso;
349# no virama
350''a\u0304→$waa;
351''ai→$wai;
352''au→$wau;
353''i\u0304→$wii;
354''i→$wi;
355''u\u0304→$wuu;
356''u→$wu;
357''r\u0325\u0304→$wrr;
358''r\u0325→$wr;
359''l\u0325\u0304→$wll;
360''l\u0325→$wl;
361''e\u0304→$we;
362''o\u0304→$wo;
363''a→$wa;
364''e\u0306→$wce;
365''o\u0306→$wco;
366''e→$wse;
367''o→$wso;
368$virama } [$z] → $virama;
369$virama } ' ' → $virama ;
370$virama}$endThing→;
371ʔ→$dgs; # Glottal Stop
3720→$zero;
3731→$one;
3742→$two;
3753→$three;
3764→$four;
3775→$five;
3786→$six;
3797→$seven;
3808→$eight;
3819→$nine;
382''→;
383#:: NFC (NFD) ;
384
385