[apple/icu.git] / icuSources / data / translit / Latin_InterIndic.txt

# © 2016 and later: Unicode, Inc. and others.
# License & terms of use: http://www.unicode.org/copyright.html#License
#
# File: Latin_InterIndic.txt
# Generated from CLDR
#

# Latin-InterIndic
#:: NFD;
#\u0E00 reserved
#consonants
$chandrabindu=\uE001;
$anusvara=\uE002;
$visarga=\uE003;
#\u0E004 reserved
# w←vowel→ represents the stand-alone form
$wa=\uE005;
$waa=\uE006;
$wi=\uE007;
$wii=\uE008;
$wu=\uE009;
$wuu=\uE00A;
$wr=\uE00B;
$wl=\uE00C;
$wce=\uE00D; # LETTER CANDRA E
$wse=\uE00E; # LETTER SHORT E
$we=\uE00F;  # ए LETTER E
$wai=\uE010;
$wco=\uE011; # LETTER CANDRA O
$wso=\uE012; # LETTER SHORT O
$wo=\uE013;  # ओ LETTER O
$wau=\uE014;
$ka=\uE015;
$kha=\uE016;
$ga=\uE017;
$gha=\uE018;
$nga=\uE019;
$ca=\uE01A;
$cha=\uE01B;
$ja=\uE01C;
$jha=\uE01D;
$nya=\uE01E;
$tta=\uE01F;
$ttha=\uE020;
$dda=\uE021;
$ddha=\uE022;
$nna=\uE023;
$ta=\uE024;
$tha=\uE025;
$da=\uE026;
$dha=\uE027;
$na=\uE028;
$ena=\uE029; #compatibility
$pa=\uE02A;
$pha=\uE02B;
$ba=\uE02C;
$bha=\uE02D;
$ma=\uE02E;
$ya=\uE02F;
$ra=\uE030;
$rra=\uE031;
$la=\uE032;
$lla=\uE033;
$ela=\uE034; #compatibility
$va=\uE035;
$vva=\uE081;
$sha=\uE036;
$ssa=\uE037;
$sa=\uE038;
$ha=\uE039;
#\u093A Reserved
#\u093B Reserved
$nukta=\uE03C;
$avagraha=\uE03D; # SIGN AVAGRAHA
# ←vowel→ represents the dependent form
$aa=\uE03E;
$i=\uE03F;
$ii=\uE040;
$u=\uE041;
$uu=\uE042;
$rh=\uE043;
$rrh=\uE044;
$ce=\uE045; #VOWEL SIGN CANDRA E
$se=\uE046; #VOWEL SIGN SHORT E
$e=\uE047;
$ai=\uE048;
$co=\uE049; # VOWEL SIGN CANDRA O
$so=\uE04A; # VOWEL SIGN SHORT O
$o=\uE04B;  # ो
$au=\uE04C;
$virama=\uE04D;
# \u094E Reserved
# \u094F Reserved
$om = \uE050; # OM
# \u0951→;        # UNMAPPED STRESS SIGN UDATTA
# \u0952→;        # UNMAPPED STRESS SIGN ANUDATTA
# \u0953→;        # UNMAPPED GRAVE ACCENT
# \u0954→;        # UNMAPPED ACUTE ACCENT
$lm = \uE055;#  Telugu Length Mark
$ailm=\uE056;#  AI Length Mark
$aulm=\uE057;#  AU Length Mark
#urdu compatibity forms
$uka=\uE058;
$ukha=\uE059;
$ugha=\uE05A;
$ujha=\uE05B;
$uddha=\uE05C;
$udha=\uE05D;
$ufa=\uE05E;
$uya=\uE05F;
$wrr=\uE060;
$wll=\uE061;
$lh=\uE062;
$llh=\uE063;
$danda=\uE064;
$doubleDanda=\uE065;
$zero=\uE066;     # DIGIT ZERO
$one=\uE067;      # DIGIT ONE
$two=\uE068;      # DIGIT TWO
$three=\uE069;    # DIGIT THREE
$four=\uE06A;     # DIGIT FOUR
$five=\uE06B;     # DIGIT FIVE
$six=\uE06C;      # DIGIT SIX
$seven=\uE06D;    # DIGIT SEVEN
$eight=\uE06E;    # DIGIT EIGHT
$nine=\uE06F;     # DIGIT NINE
$dgs=\uE082;
# For all other scripts
$ecp0=\uE070;
$ecp1=\uE071;
$ecp2=\uE072;
$ecp3=\uE073;
$ecp4=\uE074;
$ecp5=\uE075;
$ecp6=\uE076;
$ecp7=\uE077;
$ecp8=\uE078;
$ecp9=\uE079;
$ecpA=\uE07A;
$ecpB=\uE07B;
$ecpC=\uE07C;
$ecpD=\uE07D;
$ecpE=\uE07E;
$ecpF=\uE07F;
# Khanda-ta
$kta=\uE083;
# ॰→; # nothing in Latin maps to InterIndic ABBREVIATION SIGN
$depVowelAbove=[\uE03E-\uE040\uE045-\uE04C];
$depVowelBelow=[\uE041-\uE044];
$endThing=[$danda$doubleDanda];
# $x was originally called '§'; $z was '%'
$x=[$virama$aa$ai$au$ii$i$uu$u$rrh$rh$lh$e$o$se$ce$so$co];
$z=[bcdfghjklmnpqrstvwxyz];
$consonants=[[$ka-$ha]$z[क-ह][ক-হ][ਕ-ਹ][ક-હ][କ-ହ][க-ஹ][క-హ][ಕ-ಹ][ക-ഹ]];
\u0315 → $avagraha;
\u0303→$chandrabindu$anusvara;
m\u0310→$chandrabindu;
h\u0323→$visarga;
x→$ka$virama$sa;
# convert to independent forms at start of word or syllable:
# dependent forms for roundtrip
\u0314a\u0304→$aa;
\u0314ai→$ai;
\u0314au→$au;
\u0314ii→$ii;
\u0314i\u0304→$ii;
\u0314i→$i;
\u0314u\u0304→$uu;
\u0314u→$u;
\u0314r\u0325\u0304→$rrh;
\u0314r\u0325→$rh;
\u0314l\u0325\u0304→$llh;
\u0314lh→$lh;
\u0314l\u0325→$lh;
\u0314e\u0304→$e;
\u0314o\u0304→$o;
\u0314a→;
\u0314e\u0306→$ce;
\u0314o\u0306→$co;
\u0314e→$se;
\u0314o→$so;
# preceeded by consonants
$consonants{ a\u0304→$aa;
$consonants{ ai→$ai;
$consonants{ au→$au;
$consonants{ ii→$ii;
$consonants{ i\u0304→$ii;
$consonants{ i→$i;
$consonants{ u\u0304→$uu;
$consonants{ u→$u;
$consonants{ r\u0325\u0304→$rrh;
$consonants{ r\u0325a→$rh;
$consonants{ r\u0325→$rh;
$consonants{ l\u0325\u0304→$llh;
$consonants{ lh→$lh;
$consonants{ l\u0325→$lh;
$consonants{ e\u0304→$e;
$consonants{ o\u0304→$o;
$consonants{ e\u0306→$ce;
$consonants{ o\u0306→$co;
$consonants{ e→$se;
$consonants{ o→$so;
# e.g. keai -→ {ka}{e}{wai}; k'ai -→ {ka}{wai}; (ai) -→ ({wai})
a\u0304→$waa;
ai→$wai;
au→$wau;
i\u0304→$wii;
i→$wi;
u\u0304→$wuu;
u→$wu;
r\u0325\u0304→$wrr;
r\u0325→$wr;
l\u0325\u0304→$wll;
lh→$wl;
l\u0325→$wl;
e\u0304→$we;
o\u0304→$wo;
a→$wa;
e\u0306→$wce;
o\u0306→$wco;
e→$wse;
''om→$om;
o→$wso;
# rules for anusvara
n}r\u0325           → $na|$virama;
n}l\u0325           → $na|$virama;
n}na                → $na|$virama;
n\u0307}[kg]        → $anusvara;
n\u0307}n\u0307     → $anusvara;
n\u0304}[cj]        → $anusvara;
n\u0304}n\u0303     → $anusvara;
n\u0323}[tdn]\u0323 → $anusvara;
n}[tdn]             → $anusvara;
m}[pbm]             → $anusvara;
n}[ylvshr]          → $anusvara;
m\u0307             → $anusvara;
#urdu compatibility
q→$uka|$virama;
k\u0331h\u0331→$ukha |$virama;
g\u0307→ $ugha | $virama;
z → $ujha |$virama;
f → $ufa|$virama;
t\u0331→$kta;
# dev
y\u0307→$uya|$virama;
l\u0331→$ela|$virama;
n\u0331→$ena|$virama;
n\u0307→$nga|$virama;
n\u0303→$nya|$virama;
n\u0323→$nna|$virama;
t\u0323h→$ttha|$virama;
t\u0323→$tta|$virama;
r\u0323h→$udha|$virama;
r\u0323→$uddha|$virama;
d\u0323h→$ddha|$virama;
d\u0323→$dda|$virama;
kh→$kha|$virama;
k→$ka|$virama;
gh→$gha|$virama;
g→$ga|$virama;
ch→$cha|$virama;
c→$ca|$virama;
jh→$jha|$virama;
j→$ja|$virama;
ny→$nya|$virama;
tth→$ttha|$virama;
ddh→$ddha|$virama;
th→$tha|$virama;
t→$ta|$virama;
dh→$dha|$virama;
d→$da|$virama;
n→$na|$virama;
ph→$pha|$virama;
p→$pa|$virama;
bh→$bha|$virama;
b→$ba|$virama;
m→$ma|$virama;
y→$ya|$virama;
r\u0331→$rra|$virama;
r→$ra|$virama;
l\u0323→$lla|$virama;
l→$la|$virama;
v→$va|$virama;
w\u0307→$vva|$virama;
w→$va|$virama;
sh→$sha|$virama;
ss→$ssa|$virama;
s\u0323→$ssa|$virama;
s\u0301→$sha|$virama;
s→$sa|$virama;
h→$ha|$virama;
'.'→$danda;
$danda'.'→$doubleDanda;
$depVowelAbove{'~'→$anusvara;
$depVowelBelow{'~'→$chandrabindu;
# convert to dependent forms after consonant with no vowel:
# e.g. kai -→ {ka}{virama}ai -→ {ka}{ai}
#$virama aa→$aa;
$virama a\u0304→$aa;
$virama ai→$ai;
$virama au→$au;
$virama ii→$ii;
$virama i\u0304→$ii;
$virama i→$i;
#$virama uu→$uu;
$virama u\u0304→$uu;
$virama u→$u;
#$virama rrh→$rrh;
$virama r\u0325\u0304→$rrh;
#$virama rh→$rh;
$virama r\u0325a→$rh;
$virama r\u0325→$rh;
$virama l\u0325\u0304→$llh;
$virama lh→$lh;
$virama l\u0325→$lh;
$virama e\u0304→$e;
$virama o\u0304→$o;
$virama a→;
$virama e\u0306→$ce;
$virama o\u0306→$co;
$virama e→$se;
$virama o→$so;
# otherwise convert independent forms when separated by ': k'ai -→ {ka}{virama}{wai}
#$virama''aa→$waa;
$virama''a\u0304→$waa;
$virama''ai→$wai;
$virama''au→$wau;
#$virama''ii→$wii;
$virama''i\u0304→$wii;
$virama''i→$wi;
#$virama''uu→$wuu;
$virama''u\u0304→$wuu;
$virama''u→$wu;
#$virama''rrh→$wrr;
$virama''r\u0325\u0304→$wrr;
#$virama''rh→$wr;
$virama''r\u0325→$wr;
$virama''l\u0325\u0304→$wll;
#$virama''lh→$wl;
$virama''l\u0325→$wl;
$virama''e\u0304→$we;
$virama''o\u0304→$wo;
$virama''a→$wa;
$virama''e\u0306→$wce;
$virama''o\u0306→$wco;
$virama''e→$wse;
$virama''o→$wso;
# no virama
''a\u0304→$waa;
''ai→$wai;
''au→$wau;
''i\u0304→$wii;
''i→$wi;
''u\u0304→$wuu;
''u→$wu;
''r\u0325\u0304→$wrr;
''r\u0325→$wr;
''l\u0325\u0304→$wll;
''l\u0325→$wl;
''e\u0304→$we;
''o\u0304→$wo;
''a→$wa;
''e\u0306→$wce;
''o\u0306→$wco;
''e→$wse;
''o→$wso;
$virama } [$z] → $virama;
$virama } ' ' → $virama ;
$virama}$endThing→;
ʔ→$dgs; # Glottal Stop
0→$zero;
1→$one;
2→$two;
3→$three;
4→$four;
5→$five;
6→$six;
7→$seven;
8→$eight;
9→$nine;
''→;
#:: NFC (NFD) ;
Commit	Line	Data
f3c0d7a5 A	1	# © 2016 and later: Unicode, Inc. and others.
	2	# License & terms of use: http://www.unicode.org/copyright.html#License
	3	#
73c04bcf	4	# File: Latin_InterIndic.txt
f3c0d7a5	5	# Generated from CLDR
73c04bcf	6	#
2ca993e8 A	7
	8	# Latin-InterIndic
	9	#:: NFD;
	10	#\u0E00 reserved
	11	#consonants
73c04bcf A	12	$chandrabindu=\uE001;
	13	$anusvara=\uE002;
	14	$visarga=\uE003;
2ca993e8 A	15	#\u0E004 reserved
2ca993e8 A	16	# w←vowel→ represents the stand-alone form
73c04bcf A	17	$wa=\uE005;
	18	$waa=\uE006;
	19	$wi=\uE007;
	20	$wii=\uE008;
	21	$wu=\uE009;
	22	$wuu=\uE00A;
	23	$wr=\uE00B;
	24	$wl=\uE00C;
	25	$wce=\uE00D; # LETTER CANDRA E
	26	$wse=\uE00E; # LETTER SHORT E
51004dcb	27	$we=\uE00F; # ए LETTER E
73c04bcf A	28	$wai=\uE010;
	29	$wco=\uE011; # LETTER CANDRA O
	30	$wso=\uE012; # LETTER SHORT O
51004dcb	31	$wo=\uE013; # ओ LETTER O
73c04bcf A	32	$wau=\uE014;
	33	$ka=\uE015;
	34	$kha=\uE016;
	35	$ga=\uE017;
	36	$gha=\uE018;
	37	$nga=\uE019;
	38	$ca=\uE01A;
	39	$cha=\uE01B;
	40	$ja=\uE01C;
	41	$jha=\uE01D;
	42	$nya=\uE01E;
	43	$tta=\uE01F;
	44	$ttha=\uE020;
	45	$dda=\uE021;
	46	$ddha=\uE022;
	47	$nna=\uE023;
	48	$ta=\uE024;
	49	$tha=\uE025;
	50	$da=\uE026;
	51	$dha=\uE027;
	52	$na=\uE028;
	53	$ena=\uE029; #compatibility
	54	$pa=\uE02A;
	55	$pha=\uE02B;
	56	$ba=\uE02C;
	57	$bha=\uE02D;
	58	$ma=\uE02E;
	59	$ya=\uE02F;
	60	$ra=\uE030;
	61	$rra=\uE031;
	62	$la=\uE032;
	63	$lla=\uE033;
	64	$ela=\uE034; #compatibility
	65	$va=\uE035;
	66	$vva=\uE081;
	67	$sha=\uE036;
	68	$ssa=\uE037;
	69	$sa=\uE038;
	70	$ha=\uE039;
2ca993e8 A	71	#\u093A Reserved
2ca993e8 A	72	#\u093B Reserved
73c04bcf A	73	$nukta=\uE03C;
73c04bcf A	74	$avagraha=\uE03D; # SIGN AVAGRAHA
2ca993e8	75	# ←vowel→ represents the dependent form
73c04bcf A	76	$aa=\uE03E;
	77	$i=\uE03F;
	78	$ii=\uE040;
	79	$u=\uE041;
	80	$uu=\uE042;
	81	$rh=\uE043;
b331163b	82	$rrh=\uE044;
73c04bcf A	83	$ce=\uE045; #VOWEL SIGN CANDRA E
	84	$se=\uE046; #VOWEL SIGN SHORT E
	85	$e=\uE047;
	86	$ai=\uE048;
	87	$co=\uE049; # VOWEL SIGN CANDRA O
	88	$so=\uE04A; # VOWEL SIGN SHORT O
51004dcb	89	$o=\uE04B; # ो
73c04bcf A	90	$au=\uE04C;
73c04bcf A	91	$virama=\uE04D;
2ca993e8 A	92	# \u094E Reserved
2ca993e8 A	93	# \u094F Reserved
73c04bcf	94	$om = \uE050; # OM
2ca993e8 A	95	# \u0951→; # UNMAPPED STRESS SIGN UDATTA
	96	# \u0952→; # UNMAPPED STRESS SIGN ANUDATTA
	97	# \u0953→; # UNMAPPED GRAVE ACCENT
	98	# \u0954→; # UNMAPPED ACUTE ACCENT
51004dcb A	99	$lm = \uE055;# Telugu Length Mark
	100	$ailm=\uE056;# AI Length Mark
	101	$aulm=\uE057;# AU Length Mark
2ca993e8	102	#urdu compatibity forms
73c04bcf A	103	$uka=\uE058;
	104	$ukha=\uE059;
	105	$ugha=\uE05A;
	106	$ujha=\uE05B;
	107	$uddha=\uE05C;
	108	$udha=\uE05D;
	109	$ufa=\uE05E;
	110	$uya=\uE05F;
	111	$wrr=\uE060;
	112	$wll=\uE061;
b331163b	113	$lh=\uE062;
73c04bcf A	114	$llh=\uE063;
	115	$danda=\uE064;
	116	$doubleDanda=\uE065;
51004dcb A	117	$zero=\uE066; # DIGIT ZERO
	118	$one=\uE067; # DIGIT ONE
	119	$two=\uE068; # DIGIT TWO
	120	$three=\uE069; # DIGIT THREE
	121	$four=\uE06A; # DIGIT FOUR
	122	$five=\uE06B; # DIGIT FIVE
	123	$six=\uE06C; # DIGIT SIX
	124	$seven=\uE06D; # DIGIT SEVEN
	125	$eight=\uE06E; # DIGIT EIGHT
	126	$nine=\uE06F; # DIGIT NINE
73c04bcf	127	$dgs=\uE082;
2ca993e8	128	# For all other scripts
73c04bcf A	129	$ecp0=\uE070;
	130	$ecp1=\uE071;
	131	$ecp2=\uE072;
	132	$ecp3=\uE073;
	133	$ecp4=\uE074;
	134	$ecp5=\uE075;
	135	$ecp6=\uE076;
	136	$ecp7=\uE077;
	137	$ecp8=\uE078;
	138	$ecp9=\uE079;
	139	$ecpA=\uE07A;
	140	$ecpB=\uE07B;
	141	$ecpC=\uE07C;
	142	$ecpD=\uE07D;
	143	$ecpE=\uE07E;
	144	$ecpF=\uE07F;
2ca993e8	145	# Khanda-ta
73c04bcf	146	$kta=\uE083;
2ca993e8	147	# ॰→; # nothing in Latin maps to InterIndic ABBREVIATION SIGN
73c04bcf A	148	$depVowelAbove=[\uE03E-\uE040\uE045-\uE04C];
	149	$depVowelBelow=[\uE041-\uE044];
	150	$endThing=[$danda$doubleDanda];
2ca993e8	151	# $x was originally called '§'; $z was '%'
73c04bcf A	152	$x=[$virama$aa$ai$au$ii$i$uu$u$rrh$rh$lh$e$o$se$ce$so$co];
	153	$z=[bcdfghjklmnpqrstvwxyz];
	154	$consonants=[[$ka-$ha]$z[क-ह][ক-হ][ਕ-ਹ][ક-હ][କ-ହ][க-ஹ][క-హ][ಕ-ಹ][ക-ഹ]];
729e4ab9 A	155	\u0315 → $avagraha;
	156	\u0303→$chandrabindu$anusvara;
	157	m\u0310→$chandrabindu;
	158	h\u0323→$visarga;
	159	x→$ka$virama$sa;
2ca993e8 A	160	# convert to independent forms at start of word or syllable:
2ca993e8 A	161	# dependent forms for roundtrip
729e4ab9 A	162	\u0314a\u0304→$aa;
	163	\u0314ai→$ai;
	164	\u0314au→$au;
	165	\u0314ii→$ii;
	166	\u0314i\u0304→$ii;
	167	\u0314i→$i;
	168	\u0314u\u0304→$uu;
	169	\u0314u→$u;
	170	\u0314r\u0325\u0304→$rrh;
	171	\u0314r\u0325→$rh;
	172	\u0314l\u0325\u0304→$llh;
	173	\u0314lh→$lh;
	174	\u0314l\u0325→$lh;
	175	\u0314e\u0304→$e;
	176	\u0314o\u0304→$o;
	177	\u0314a→;
	178	\u0314e\u0306→$ce;
	179	\u0314o\u0306→$co;
	180	\u0314e→$se;
	181	\u0314o→$so;
2ca993e8	182	# preceeded by consonants
729e4ab9 A	183	$consonants{ a\u0304→$aa;
	184	$consonants{ ai→$ai;
	185	$consonants{ au→$au;
	186	$consonants{ ii→$ii;
	187	$consonants{ i\u0304→$ii;
	188	$consonants{ i→$i;
	189	$consonants{ u\u0304→$uu;
	190	$consonants{ u→$u;
	191	$consonants{ r\u0325\u0304→$rrh;
	192	$consonants{ r\u0325a→$rh;
	193	$consonants{ r\u0325→$rh;
	194	$consonants{ l\u0325\u0304→$llh;
	195	$consonants{ lh→$lh;
	196	$consonants{ l\u0325→$lh;
	197	$consonants{ e\u0304→$e;
	198	$consonants{ o\u0304→$o;
	199	$consonants{ e\u0306→$ce;
	200	$consonants{ o\u0306→$co;
	201	$consonants{ e→$se;
	202	$consonants{ o→$so;
2ca993e8	203	# e.g. keai -→ {ka}{e}{wai}; k'ai -→ {ka}{wai}; (ai) -→ ({wai})
729e4ab9 A	204	a\u0304→$waa;
	205	ai→$wai;
	206	au→$wau;
	207	i\u0304→$wii;
	208	i→$wi;
	209	u\u0304→$wuu;
	210	u→$wu;
	211	r\u0325\u0304→$wrr;
	212	r\u0325→$wr;
	213	l\u0325\u0304→$wll;
	214	lh→$wl;
	215	l\u0325→$wl;
	216	e\u0304→$we;
	217	o\u0304→$wo;
	218	a→$wa;
	219	e\u0306→$wce;
	220	o\u0306→$wco;
	221	e→$wse;
	222	''om→$om;
	223	o→$wso;
2ca993e8	224	# rules for anusvara
51004dcb A	225	n}r\u0325 → $na\|$virama;
	226	n}l\u0325 → $na\|$virama;
	227	n}na → $na\|$virama;
	228	n\u0307}[kg] → $anusvara;
	229	n\u0307}n\u0307 → $anusvara;
	230	n\u0304}[cj] → $anusvara;
	231	n\u0304}n\u0303 → $anusvara;
729e4ab9	232	n\u0323}[tdn]\u0323 → $anusvara;
51004dcb A	233	n}[tdn] → $anusvara;
	234	m}[pbm] → $anusvara;
	235	n}[ylvshr] → $anusvara;
	236	m\u0307 → $anusvara;
2ca993e8	237	#urdu compatibility
729e4ab9 A	238	q→$uka\|$virama;
	239	k\u0331h\u0331→$ukha \|$virama;
	240	g\u0307→ $ugha \| $virama;
	241	z → $ujha \|$virama;
	242	f → $ufa\|$virama;
	243	t\u0331→$kta;
2ca993e8	244	# dev
729e4ab9 A	245	y\u0307→$uya\|$virama;
	246	l\u0331→$ela\|$virama;
	247	n\u0331→$ena\|$virama;
	248	n\u0307→$nga\|$virama;
	249	n\u0303→$nya\|$virama;
	250	n\u0323→$nna\|$virama;
	251	t\u0323h→$ttha\|$virama;
	252	t\u0323→$tta\|$virama;
	253	r\u0323h→$udha\|$virama;
	254	r\u0323→$uddha\|$virama;
	255	d\u0323h→$ddha\|$virama;
	256	d\u0323→$dda\|$virama;
	257	kh→$kha\|$virama;
	258	k→$ka\|$virama;
	259	gh→$gha\|$virama;
	260	g→$ga\|$virama;
	261	ch→$cha\|$virama;
	262	c→$ca\|$virama;
	263	jh→$jha\|$virama;
	264	j→$ja\|$virama;
	265	ny→$nya\|$virama;
	266	tth→$ttha\|$virama;
	267	ddh→$ddha\|$virama;
	268	th→$tha\|$virama;
	269	t→$ta\|$virama;
	270	dh→$dha\|$virama;
	271	d→$da\|$virama;
	272	n→$na\|$virama;
	273	ph→$pha\|$virama;
	274	p→$pa\|$virama;
	275	bh→$bha\|$virama;
	276	b→$ba\|$virama;
	277	m→$ma\|$virama;
	278	y→$ya\|$virama;
	279	r\u0331→$rra\|$virama;
	280	r→$ra\|$virama;
	281	l\u0323→$lla\|$virama;
	282	l→$la\|$virama;
	283	v→$va\|$virama;
	284	w\u0307→$vva\|$virama;
	285	w→$va\|$virama;
	286	sh→$sha\|$virama;
	287	ss→$ssa\|$virama;
	288	s\u0323→$ssa\|$virama;
	289	s\u0301→$sha\|$virama;
	290	s→$sa\|$virama;
	291	h→$ha\|$virama;
	292	'.'→$danda;
	293	$danda'.'→$doubleDanda;
	294	$depVowelAbove{'~'→$anusvara;
	295	$depVowelBelow{'~'→$chandrabindu;
2ca993e8 A	296	# convert to dependent forms after consonant with no vowel:
	297	# e.g. kai -→ {ka}{virama}ai -→ {ka}{ai}
	298	#$virama aa→$aa;
729e4ab9 A	299	$virama a\u0304→$aa;
	300	$virama ai→$ai;
	301	$virama au→$au;
	302	$virama ii→$ii;
	303	$virama i\u0304→$ii;
	304	$virama i→$i;
2ca993e8	305	#$virama uu→$uu;
729e4ab9 A	306	$virama u\u0304→$uu;
729e4ab9 A	307	$virama u→$u;
2ca993e8	308	#$virama rrh→$rrh;
729e4ab9	309	$virama r\u0325\u0304→$rrh;
2ca993e8	310	#$virama rh→$rh;
729e4ab9 A	311	$virama r\u0325a→$rh;
	312	$virama r\u0325→$rh;
	313	$virama l\u0325\u0304→$llh;
	314	$virama lh→$lh;
	315	$virama l\u0325→$lh;
	316	$virama e\u0304→$e;
	317	$virama o\u0304→$o;
	318	$virama a→;
	319	$virama e\u0306→$ce;
	320	$virama o\u0306→$co;
	321	$virama e→$se;
	322	$virama o→$so;
2ca993e8 A	323	# otherwise convert independent forms when separated by ': k'ai -→ {ka}{virama}{wai}
2ca993e8 A	324	#$virama''aa→$waa;
729e4ab9 A	325	$virama''a\u0304→$waa;
	326	$virama''ai→$wai;
	327	$virama''au→$wau;
2ca993e8	328	#$virama''ii→$wii;
729e4ab9 A	329	$virama''i\u0304→$wii;
729e4ab9 A	330	$virama''i→$wi;
2ca993e8	331	#$virama''uu→$wuu;
729e4ab9 A	332	$virama''u\u0304→$wuu;
729e4ab9 A	333	$virama''u→$wu;
2ca993e8	334	#$virama''rrh→$wrr;
729e4ab9	335	$virama''r\u0325\u0304→$wrr;
2ca993e8	336	#$virama''rh→$wr;
729e4ab9 A	337	$virama''r\u0325→$wr;
729e4ab9 A	338	$virama''l\u0325\u0304→$wll;
2ca993e8	339	#$virama''lh→$wl;
729e4ab9 A	340	$virama''l\u0325→$wl;
	341	$virama''e\u0304→$we;
	342	$virama''o\u0304→$wo;
	343	$virama''a→$wa;
	344	$virama''e\u0306→$wce;
	345	$virama''o\u0306→$wco;
	346	$virama''e→$wse;
	347	$virama''o→$wso;
2ca993e8	348	# no virama
729e4ab9 A	349	''a\u0304→$waa;
	350	''ai→$wai;
	351	''au→$wau;
	352	''i\u0304→$wii;
	353	''i→$wi;
	354	''u\u0304→$wuu;
	355	''u→$wu;
	356	''r\u0325\u0304→$wrr;
	357	''r\u0325→$wr;
	358	''l\u0325\u0304→$wll;
	359	''l\u0325→$wl;
	360	''e\u0304→$we;
	361	''o\u0304→$wo;
	362	''a→$wa;
	363	''e\u0306→$wce;
	364	''o\u0306→$wco;
	365	''e→$wse;
	366	''o→$wso;
	367	$virama } [$z] → $virama;
	368	$virama } ' ' → $virama ;
	369	$virama}$endThing→;
	370	ʔ→$dgs; # Glottal Stop
	371	0→$zero;
	372	1→$one;
	373	2→$two;
	374	3→$three;
	375	4→$four;
	376	5→$five;
	377	6→$six;
	378	7→$seven;
	379	8→$eight;
	380	9→$nine;
	381	''→;
2ca993e8 A	382	#:: NFC (NFD) ;
2ca993e8 A	383