ICU-62107.0.1.tar.gz

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

diff --git a/icuSources/data/translit/InterIndic_Latin.txt b/icuSources/data/translit/InterIndic_Latin.txt
index a445437113209a8117f1fd08fd499920258cff23..13cd64a7211cd99b3eff533b82208c7ec9f79a83 100644 (file)
--- a/icuSources/data/translit/InterIndic_Latin.txt
+++ b/icuSources/data/translit/InterIndic_Latin.txt
@@ -1,15 +1,18 @@
-# ***************************************************************************
-# *
-# *  Copyright (C) 2004-2014, International Business Machines
-# *  Corporation; Unicode, Inc.; and others.  All Rights Reserved.
-# *
-# ***************************************************************************
+# © 2016 and later: Unicode, Inc. and others.
+# License & terms of use: http://www.unicode.org/copyright.html#License
+#

 # File: InterIndic_Latin.txt
 # File: InterIndic_Latin.txt

-# Generated from CLDR 
+# Generated from CLDR

 #
 #

+
+# InterIndic-Latin
+#\u0E00 reserved
+#consonants

 $chandrabindu=\uE001;
 $anusvara=\uE002;
 $visarga=\uE003;
 $chandrabindu=\uE001;
 $anusvara=\uE002;
 $visarga=\uE003;

+#\u0E004 reserved
+# w←vowel→ represents the stand-alone form

 $wa=\uE005;
 $waa=\uE006;
 $wi=\uE007;
 $wa=\uE005;
 $waa=\uE006;
 $wi=\uE007;


@@ -64,15 +67,18 @@ $sha=\uE036;
 $ssa=\uE037;
 $sa=\uE038;
 $ha=\uE039;
 $ssa=\uE037;
 $sa=\uE038;
 $ha=\uE039;

+#\u093A Reserved
+#\u093B Reserved

 $nukta=\uE03C;
 $avagraha=\uE03D; # SIGN AVAGRAHA
 $nukta=\uE03C;
 $avagraha=\uE03D; # SIGN AVAGRAHA

+# ←vowel→ represents the dependent form

 $aa=\uE03E;
 $i=\uE03F;
 $ii=\uE040;
 $u=\uE041;
 $uu=\uE042;
 $rh=\uE043;
 $aa=\uE03E;
 $i=\uE03F;
 $ii=\uE040;
 $u=\uE041;
 $uu=\uE042;
 $rh=\uE043;

-$lh=\uE044;
+$rrh=\uE044;

 $ce=\uE045; #VOWEL SIGN CANDRA E
 $se=\uE046; #VOWEL SIGN SHORT E
 $e=\uE047;
 $ce=\uE045; #VOWEL SIGN CANDRA E
 $se=\uE046; #VOWEL SIGN SHORT E
 $e=\uE047;


@@ -82,6 +88,8 @@ $so=\uE04A; # VOWEL SIGN SHORT O
 $o=\uE04B;  # ो
 $au=\uE04C;
 $virama=\uE04D;
 $o=\uE04B;  # ो
 $au=\uE04C;
 $virama=\uE04D;

+# \u094E Reserved
+# \u094F Reserved

 $om=\uE050; # OM
 \uE051→;        # UNMAPPED STRESS SIGN UDATTA
 \uE052→;        # UNMAPPED STRESS SIGN ANUDATTA
 $om=\uE050; # OM
 \uE051→;        # UNMAPPED STRESS SIGN UDATTA
 \uE052→;        # UNMAPPED STRESS SIGN ANUDATTA


@@ -90,6 +98,7 @@ $om=\uE050; # OM
 $lm = \uE055;#  Telugu Length Mark
 $ailm=\uE056;#  AI Length Mark
 $aulm=\uE057;#  AU Length Mark
 $lm = \uE055;#  Telugu Length Mark
 $ailm=\uE056;#  AI Length Mark
 $aulm=\uE057;#  AU Length Mark

+#urdu compatibity forms

 $uka=\uE058;
 $ukha=\uE059;
 $ugha=\uE05A;
 $uka=\uE058;
 $ukha=\uE059;
 $ugha=\uE05A;


@@ -100,7 +109,7 @@ $ufa=\uE05E;
 $uya=\uE05F;
 $wrr=\uE060;
 $wll=\uE061;
 $uya=\uE05F;
 $wrr=\uE060;
 $wll=\uE061;

-$rrh=\uE062;
+$lh=\uE062;

 $llh=\uE063;
 $danda=\uE064;
 $doubleDanda=\uE065;
 $llh=\uE063;
 $danda=\uE064;
 $doubleDanda=\uE065;


@@ -114,14 +123,21 @@ $six=\uE06C;      # DIGIT SIX
 $seven=\uE06D;    # DIGIT SEVEN
 $eight=\uE06E;    # DIGIT EIGHT
 $nine=\uE06F;     # DIGIT NINE
 $seven=\uE06D;    # DIGIT SEVEN
 $eight=\uE06E;    # DIGIT EIGHT
 $nine=\uE06F;     # DIGIT NINE

+# Glottal stop

 $dgs=\uE082;
 $dgs=\uE082;

+#Khanda-ta

 $kta=\uE083;
 $depVowelAbove=[\uE03E-\uE040\uE045-\uE04C];
 $depVowelBelow=[\uE041-\uE044];
 $kta=\uE083;
 $depVowelAbove=[\uE03E-\uE040\uE045-\uE04C];
 $depVowelBelow=[\uE041-\uE044];

+# $x was originally called '§'; $z was '%'

 $x=[$aa$ai$au$ii$i$uu$u$rrh$rh$lh$llh$e$o$se$ce$so$co];
 $z=[bcdfghjklmnpqrstvwxyz];
 $vowels=[aeiour\u0304\u0325\u0306];
 $forceIndependentMatra = [^[[:L:][\u0300-\u034C]]];
 $x=[$aa$ai$au$ii$i$uu$u$rrh$rh$lh$llh$e$o$se$ce$so$co];
 $z=[bcdfghjklmnpqrstvwxyz];
 $vowels=[aeiour\u0304\u0325\u0306];
 $forceIndependentMatra = [^[[:L:][\u0300-\u034C]]];

+######################################################################
+# convert from Native letters to Latin letters
+######################################################################
+#transliterations for anusvara

 $anusvara} [$ka$kha$ga$gha$nga]             → n\u0307;
 $anusvara} [$ca$cha$ja$jha$nya]             → n\u0304;
 $anusvara} [$tta$ttha$dda$ddha$nna]         → n\u0323;
 $anusvara} [$ka$kha$ga$gha$nga]             → n\u0307;
 $anusvara} [$ca$cha$ja$jha$nya]             → n\u0304;
 $anusvara} [$tta$ttha$dda$ddha$nna]         → n\u0323;


@@ -129,6 +145,7 @@ $anusvara} [$ta$tha$da$dha$na]              → n;
 $anusvara} [$pa$pha$ba$bha$ma]              → m;
 $anusvara} [$ya$ra$lla$la$va$ssa$sha$sa$ha] → n;
 $anusvara→ m\u0307;
 $anusvara} [$pa$pha$ba$bha$ma]              → m;
 $anusvara} [$ya$ra$lla$la$va$ssa$sha$sa$ha] → n;
 $anusvara→ m\u0307;

+# Urdu compatibility

 $ya$nukta}$x        → y\u0307;
 $ya$nukta$virama    → y\u0307;
 $ya$nukta           → y\u0307a;
 $ya$nukta}$x        → y\u0307;
 $ya$nukta$virama    → y\u0307;
 $ya$nukta           → y\u0307a;


@@ -186,6 +203,7 @@ $ela                → l\u0331a;
 $uya}$x             → y\u0307;
 $uya$virama         → y\u0307;
 $uya                → y\u0307a;
 $uya}$x             → y\u0307;
 $uya$virama         → y\u0307;
 $uya                → y\u0307a;

+# normal consonants

 $ka$virama}$ha→k'';
 $ka}$x→k;
 $ka$virama→k;
 $ka$virama}$ha→k'';
 $ka}$x→k;
 $ka$virama→k;


@@ -312,6 +330,7 @@ $sa$virama}$ssa→s'';
 $sa$virama}$sa→s'';
 $sa}$x→s;
 $sa$virama→s;
 $sa$virama}$sa→s'';
 $sa}$x→s;
 $sa$virama→s;

+#for gurmukhi

 $sa$nukta}$x→s\u0301;
 $sa$nukta$virama→s\u0301;
 $sa$nukta→s\u0301a;
 $sa$nukta}$x→s\u0301;
 $sa$nukta$virama→s\u0301;
 $sa$nukta→s\u0301a;


@@ -325,6 +344,7 @@ $ssa→s\u0323a;
 $ha}$x→h;
 $ha$virama→h;
 $ha→ha;
 $ha}$x→h;
 $ha$virama→h;
 $ha→ha;

+# dependent vowels (should never occur except following consonants)

 $forceIndependentMatra{$aa  → \u0314a\u0304;
 $forceIndependentMatra{$ai  → \u0314ai;
 $forceIndependentMatra{$au  → \u0314au;
 $forceIndependentMatra{$aa  → \u0314a\u0304;
 $forceIndependentMatra{$ai  → \u0314ai;
 $forceIndependentMatra{$au  → \u0314au;


@@ -338,6 +358,7 @@ $forceIndependentMatra{$llh → \u0314l\u0325\u0304;
 $forceIndependentMatra{$lh  → \u0314l\u0325;
 $forceIndependentMatra{$e   → \u0314e\u0304;
 $forceIndependentMatra{$o   → \u0314o\u0304;
 $forceIndependentMatra{$lh  → \u0314l\u0325;
 $forceIndependentMatra{$e   → \u0314e\u0304;
 $forceIndependentMatra{$o   → \u0314o\u0304;

+#extra vowels

 $forceIndependentMatra{$ce  → \u0314e\u0306;
 $forceIndependentMatra{$co  → \u0314o\u0306;
 $forceIndependentMatra{$se  → \u0314e;
 $forceIndependentMatra{$ce  → \u0314e\u0306;
 $forceIndependentMatra{$co  → \u0314o\u0306;
 $forceIndependentMatra{$se  → \u0314e;


@@ -357,10 +378,12 @@ $llh → l\u0325\u0304;
 $lh  → l\u0325;
 $e   → e\u0304;
 $o   → o\u0304;
 $lh  → l\u0325;
 $e   → e\u0304;
 $o   → o\u0304;

+#extra vowels

 $ce  → e\u0306;
 $co  → o\u0306;
 $se  → e;
 $so  → o;
 $ce  → e\u0306;
 $co  → o\u0306;
 $se  → e;
 $so  → o;

+#dependent vowels when following independent vowels. Generally Illegal only for roundtripping

 $waa} $x → a\u0304\u0314;
 $wai} $x → ai\u0314;
 $wau} $x → au\u0314;
 $waa} $x → a\u0304\u0314;
 $wai} $x → ai\u0314;
 $wau} $x → au\u0314;


@@ -375,11 +398,13 @@ $wl } $x → l\u0325\u0314;
 $we } $x → e\u0304\u0314;
 $wo } $x → o\u0304\u0314;
 $wa } $x → a\u0314;
 $we } $x → e\u0304\u0314;
 $wo } $x → o\u0304\u0314;
 $wa } $x → a\u0314;

+#extra vowels

 $wce} $x → e\u0306\u0314;
 $wco} $x → o\u0306\u0314;
 $wse} $x → e\u0314;
 $wso} $x → o\u0314;
 $om} $x → ''om\u0314;
 $wce} $x → e\u0306\u0314;
 $wco} $x → o\u0306\u0314;
 $wse} $x → e\u0314;
 $wso} $x → o\u0314;
 $om} $x → ''om\u0314;

+# independent vowels when preceeded by vowels

 $vowels{$waa  → ''a\u0304;
 $vowels{$wai  → ''ai;
 $vowels{$wau  → ''au;
 $vowels{$waa  → ''a\u0304;
 $vowels{$wai  → ''ai;
 $vowels{$wau  → ''au;


@@ -394,10 +419,12 @@ $vowels{$wl   → ''l\u0325;
 $vowels{$we   → ''e\u0304;
 $vowels{$wo   → ''o\u0304;
 $vowels{$wa   → ''a;
 $vowels{$we   → ''e\u0304;
 $vowels{$wo   → ''o\u0304;
 $vowels{$wa   → ''a;

+#extra vowels

 $vowels{$wce  → ''e\u0306;
 $vowels{$wco  → ''o\u0306;
 $vowels{$wse  → ''e;
 $vowels{$wso  → ''o;
 $vowels{$wce  → ''e\u0306;
 $vowels{$wco  → ''o\u0306;
 $vowels{$wse  → ''e;
 $vowels{$wso  → ''o;

+# independent vowels (otherwise)

 $waa → a\u0304;
 $wai → ai;
 $wau → au;
 $waa → a\u0304;
 $wai → ai;
 $wau → au;


@@ -412,15 +439,18 @@ $wl  → l\u0325;
 $we  → e\u0304;
 $wo  → o\u0304;
 $wa  → a;
 $we  → e\u0304;
 $wo  → o\u0304;
 $wa  → a;

+#extra vowels

 $wce → e\u0306;
 $wco → o\u0306;
 $wse → e;
 $wso → o;
 $om → ''om;
 $wce → e\u0306;
 $wco → o\u0306;
 $wse → e;
 $wso → o;
 $om → ''om;

+#stress marks

 $avagraha → \u0315;
 $chandrabindu$anusvara→\u0303;
 $chandrabindu → m\u0310;
 $visarga→h\u0323;
 $avagraha → \u0315;
 $chandrabindu$anusvara→\u0303;
 $chandrabindu → m\u0310;
 $visarga→h\u0323;

+#numbers

 $zero  → 0;
 $one   → 1;
 $two   → 2;
 $zero  → 0;
 $one   → 1;
 $two   → 2;


@@ -439,9 +469,11 @@ $kta→t\u0331;
 $danda→'.';
 $doubleDanda→'.';
 \uE070→;       # ABBREVIATION SIGN
 $danda→'.';
 $doubleDanda→'.';
 \uE070→;       # ABBREVIATION SIGN

+# LETTER RA WITH MIDDLE DIAGONAL

 \uE071}$x→ra;
 \uE071$virama→r;
 \uE071→ra;
 \uE071}$x→ra;
 \uE071$virama→r;
 \uE071→ra;

+# LETTER RA WITH LOWER DIAGONAL

 \uE072}$x→ra;
 \uE072$virama→r;
 \uE072→ra;
 \uE072}$x→ra;
 \uE072$virama→r;
 \uE072→ra;


@@ -460,3 +492,4 @@ $doubleDanda→'.';
 \uE07F→;       # URA
 \uE080→;       # EK ONKAR
 \uE004→;       # DEVANAGARI VOWEL SIGN SHORT A
 \uE07F→;       # URA
 \uE080→;       # EK ONKAR
 \uE004→;       # DEVANAGARI VOWEL SIGN SHORT A

+