AuroraMiddleware/scuffed-code
1
0
Fork 0
Code Issues Pull Requests Releases Wiki Activity

ICU-1375 Fix Fallback bugs and add Inter-Indic Rountrip tests

Browse Source 
X-SVN-Rev: 6761
This commit is contained in:
Ram Viswanadha 2001-11-11 02:36:10 +00:00
parent a0988947e9
commit d6a383a8f5
36 changed files with 750 additions and 354 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

116
icu4j/src/com/ibm/icu/dev/test/translit/RoundTripTest.java
View File

@ -97,18 +97,114 @@ public class RoundTripTest extends TestFmwk {
.test(null, "[\u0400-\u045F]", null, this, new Legal());
}
public void TestDevanagari() throws IOException, ParseException {
//----------------------------------
// Inter-Indic Tests
//----------------------------------
public void TestDevanagariLatin() throws IOException, ParseException {
new Test("Latin-DEVANAGARI",
TestUtility.LATIN_SCRIPT, TestUtility.DEVANAGARI_SCRIPT)
.test(null, "[:Devanagari:]", null, this, new Legal());
}
public void TestDevanagariTamil() throws IOException, ParseException {
public void TestDevanagariBengali() throws IOException, ParseException {
new Test("BENGALI-DEVANAGARI",
TestUtility.BENGALI_SCRIPT, TestUtility.DEVANAGARI_SCRIPT)
.test("[:BENGALI:]", "[:Devanagari:]",
"[\u0950\u0935\u0912\u0933\u090e\u090D\u0911\u093d\u0929\u0934[\u0958-\u095f]\u09F0\u09F1]", /*roundtrip exclusions*/
this, new Legal());
new Test("DEVANAGARI-BENGALI",
TestUtility.DEVANAGARI_SCRIPT, TestUtility.BENGALI_SCRIPT )
.test( "[:Devanagari:]", "[:BENGALI:]",
"[\u0950\u0935\u0912\u0933\u090e\u090D\u0911\u093d\u0929\u0934[\u0958-\u095f]\u09F0\u09F1]", /*roundtrip exclusions*/
this, new Legal());
}
public void TestDevanagariGurmukhi() throws IOException, ParseException {
new Test("GURMUKHI-DEVANAGARI",
TestUtility.GURMUKHI_SCRIPT, TestUtility.DEVANAGARI_SCRIPT)
.test("[:GURMUKHI:]", "[:Devanagari:]",
"[\u0950\u090D\u090e\u0912\u0911\u090b\u090c\u0934\u0960\u0961\u0937\u0a72\u0a73\u0a74\u093d]", /*roundtrip exclusions*/
this, new Legal());
new Test("DEVANAGARI-GURMUKHI",
TestUtility.DEVANAGARI_SCRIPT, TestUtility.GURMUKHI_SCRIPT )
.test( "[:Devanagari:]", "[:GURMUKHI:]",
"[\u0950\u090D\u090e\u0912\u0911\u090b\u090c\u0934\u0960\u0961\u0937\u0a72\u0a73\u0a74\u093d]", /*roundtrip exclusions*/
this, new Legal());
}
public void TestDevanagariGujarati() throws IOException, ParseException {
new Test("GUJARATI-DEVANAGARI",
TestUtility.GUJARATI_SCRIPT, TestUtility.DEVANAGARI_SCRIPT)
.test("[:GUJARATI:]", "[:Devanagari:]",
"[\u0961\u090c\u090e\u0912]", /*roundtrip exclusions*/
this, new Legal());
new Test("DEVANAGARI-GUJARATI",
TestUtility.DEVANAGARI_SCRIPT, TestUtility.GUJARATI_SCRIPT )
.test( "[:Devanagari:]", "[:GUJARATI:]",
"[\u0961\u090c\u090e\u0912]", /*roundtrip exclusions*/
this, new Legal());
}
public void TestDevanagariOriya() throws IOException, ParseException {
new Test("ORIYA-DEVANAGARI",
TestUtility.ORIYA_SCRIPT, TestUtility.DEVANAGARI_SCRIPT)
.test("[:ORIYA:]", "[:Devanagari:]",
"[\u0950\u090D\u090e\u0912\u0911\u0931\u0935]", /*roundtrip exclusions*/
this, new Legal());
new Test("DEVANAGARI-ORIYA",
TestUtility.DEVANAGARI_SCRIPT, TestUtility.ORIYA_SCRIPT )
.test( "[:Devanagari:]", "[:ORIYA:]",
"[\u0950\u090D\u090e\u0912\u0911\u0931\u0935]", /*roundtrip exclusions*/
this, new Legal());
}
public void TestDevanagariTamil() throws IOException, ParseException {
new Test("Tamil-DEVANAGARI",
TestUtility.TAMIL_SCRIPT, TestUtility.DEVANAGARI_SCRIPT)
.test("[:tamil:]", "[:Devanagari:]", null, this, new Legal());
.test("[:tamil:]", "[:Devanagari:]",
"[\u0950\u090D\u0911\u093d\u0929\u0934[\u0958-\u095f]\u090B\u090C\u0916\u0917\u0918\u091B\u091D\u0920\u0921\u0922\u0925\u0926\u0927\u092B\u092C\u092D\u0936\u0960\u0961]", /*roundtrip exclusions*/
this, new Legal());
new Test("DEVANAGARI-Tamil",
TestUtility.DEVANAGARI_SCRIPT, TestUtility.TAMIL_SCRIPT )
.test( "[:Devanagari:]", "[:tamil:]",
"[\u0950\u090D\u0911\u093d\u0929\u0934[\u0958-\u095f]]", /*roundtrip exclusions*/
this, new Legal());
}
public void TestDevanagariTelugu() throws IOException, ParseException {
new Test("Telugu-DEVANAGARI",
TestUtility.TELUGU_SCRIPT, TestUtility.DEVANAGARI_SCRIPT)
.test("[:telugu:]", "[:Devanagari:]",
"[\u0950\u090D\u0911\u093d\u0929\u0934[\u0958-\u095f]]", /*roundtrip exclusions*/
this, new Legal());
new Test("DEVANAGARI-TELUGU",
TestUtility.DEVANAGARI_SCRIPT, TestUtility.TELUGU_SCRIPT )
.test( "[:Devanagari:]", "[:TELUGU:]",
"[\u0950\u090D\u0911\u093d\u0929\u0934[\u0958-\u095f]]", /*roundtrip exclusions*/
this, new Legal());
}
public void TestDevanagariKannada() throws IOException, ParseException {
new Test("KANNADA-DEVANAGARI",
TestUtility.KANNADA_SCRIPT, TestUtility.DEVANAGARI_SCRIPT)
.test("[:KANNADA:]", "[:Devanagari:]",
"[\u0950\u090D\u0911\u093d\u0929\u0934[\u0958-\u095f]]", /*roundtrip exclusions*/
this, new Legal());
new Test("DEVANAGARI-KANNADA",
TestUtility.DEVANAGARI_SCRIPT, TestUtility.KANNADA_SCRIPT )
.test( "[:Devanagari:]", "[:KANNADA:]",
"[\u0950\u090D\u0911\u093d\u0929\u0934[\u0958-\u095f]]", /*roundtrip exclusions*/
this, new Legal());
}
public void TestDevanagariMalayalam() throws IOException, ParseException {
new Test("MALAYALAM-DEVANAGARI",
TestUtility.MALAYALAM_SCRIPT, TestUtility.DEVANAGARI_SCRIPT)
.test("[:MALAYALAM:]", "[:Devanagari:]",
"[\u0950\u090D\u0911\u093d\u0929\u0934[\u0958-\u095f]]", /*roundtrip exclusions*/
this, new Legal());
new Test("DEVANAGARI-MALAYALAM",
TestUtility.DEVANAGARI_SCRIPT, TestUtility.MALAYALAM_SCRIPT )
.test( "[:Devanagari:]", "[:MALAYALAM:]",
"[\u0950\u090D\u0911\u093d\u0929\u0934[\u0958-\u095f]]", /*roundtrip exclusions*/
this, new Legal());
}
//---------------
// End Indic
//---------------
public static class Legal {
public boolean is(String sourceString) {return true;}
}
@ -262,24 +358,26 @@ public class RoundTripTest extends TestFmwk {
return false;
}
public void test(String sourceRange, String targetRange, String roundtripExclusions, TestLog log, Legal legalSource)
public void test(String sourceRange, String targetRange, String roundtripExclusions,
TestLog log, Legal legalSource)
throws java.io.IOException, java.text.ParseException {
this.legalSource = legalSource;
if (sourceRange != null && sourceRange.length() > 0) {
this.sourceRange = new UnicodeSet(sourceRange);
}else{
this.sourceRange = new UnicodeSet("[a-zA-Z]");
}
if (targetRange != null && targetRange.length() > 0) {
this.targetRange = new UnicodeSet(targetRange);
}
if (roundtripExclusions != null && roundtripExclusions.length() > 0) {
this.roundtripExclusions = new UnicodeSet(roundtripExclusions);
}else{
this.roundtripExclusions = new UnicodeSet(); // empty
}
if (this.sourceRange == null) this.sourceRange = new UnicodeSet("[a-zA-Z]");
if (this.roundtripExclusions == null) this.roundtripExclusions = new UnicodeSet(); // empty
this.log = log;
log.logln(Utility.escape("Source: " + this.sourceRange));

4
icu4j/src/com/ibm/icu/impl/data/Transliterator_Bengali_InterIndic.txt
View File

@ -89,8 +89,8 @@
\u09ed>\ue06d; # DIGIT SEVEN
\u09ee>\ue06e; # DIGIT EIGHT
\u09ef>\ue06f; # DIGIT NINE
# \u09f0>; # UNMAPPED Bengali-InterIndic: LETTER RA WITH MIDDLE DIAGONAL
# \u09f1>; # UNMAPPED Bengali-InterIndic: LETTER RA WITH LOWER DIAGONAL
\u09f0>\ue0f0; # UNMAPPED Bengali-InterIndic: LETTER RA WITH MIDDLE DIAGONAL
\u09f1>\ue0f1; # UNMAPPED Bengali-InterIndic: LETTER RA WITH LOWER DIAGONAL
# \u09f2>; # UNMAPPED Bengali-InterIndic: RUPEE MARK
# \u09f3>; # UNMAPPED Bengali-InterIndic: RUPEE SIGN
# \u09f4>; # UNMAPPED Bengali-InterIndic: CURRENCY NUMERATOR ONE

12
icu4j/src/com/ibm/icu/impl/data/Transliterator_Devanagari_InterIndic.txt
View File

@ -35,11 +35,11 @@
\u090b>\ue00b; # LETTER VOCALIC R
\u090c>\ue00c; # LETTER VOCALIC L
\u090d>\ue00d; # LETTER CANDRA E (For representing English sounds)
#\u090e>\ue00e; # UNMAPPED LETTER SHORT E(For Southern Scripts)
\u090e>\ue00e; # UNMAPPED LETTER SHORT E(For Southern Scripts)
\u090f>\ue00f; # LETTER E
\u0910>\ue010; # LETTER AI
\u0911>\ue011; # LETTER CANDRA O (For representing English sounds)
#\u0912>\ue012; # UNMAPPED LETTER SHORT O (For Southern Scripts)
\u0912>\ue012; # UNMAPPED LETTER SHORT O (For Southern Scripts)
\u0913>\ue013; # LETTER O
\u0914>\ue014; # LETTER AU
\u0915>\ue015; # LETTER KA
@ -70,10 +70,10 @@
\u092e>\ue02e; # LETTER MA
\u092f>\ue02f; # LETTER YA
\u0930>\ue030; # LETTER RA
#\u0931>\ue031; # UNMAPPED LETTER RRA (Eyelash RA for Southern scripts)
\u0931>\ue031; # UNMAPPED LETTER RRA (Eyelash RA for Southern scripts)
\u0932>\ue032; # LETTER LA
\u0933>\ue033; # LETTER LLA
#\u0934>\ue034; # UNMAPPED LETTER LLLA (LLLA for Southern scripts)
\u0934>\ue034; # UNMAPPED LETTER LLLA (LLLA for Southern scripts)
\u0935>\ue035; # LETTER VA
\u0936>\ue036; # LETTER SHA
\u0937>\ue037; # LETTER SSA
@ -89,11 +89,11 @@
\u0943>\ue043; # VOWEL SIGN VOCALIC R
\u0944>\ue044; # VOWEL SIGN VOCALIC RR
\u0945>\ue045; # VOWEL SIGN CANDRA E
#\u0946>\ue046; # UNMAPPED VOWEL SIGN SHORT E
\u0946>\ue046; # UNMAPPED VOWEL SIGN SHORT E
\u0947>\ue047; # VOWEL SIGN E
\u0948>\ue048; # VOWEL SIGN AI
\u0949>\ue049; # VOWEL SIGN CANDRA O
#\u094a>\ue04a; # UNMAPPED VOWEL SIGN SHORT O
\u094a>\ue04a; # UNMAPPED VOWEL SIGN SHORT O
\u094b>\ue04b; # VOWEL SIGN O
\u094c>\ue04c; # VOWEL SIGN AU
\u094d>\ue04d; # SIGN VIRAMA

4
icu4j/src/com/ibm/icu/impl/data/Transliterator_Gujarati_InterIndic.txt
View File

@ -26,10 +26,10 @@
\u0a89>\ue009; # LETTER U
\u0a8a>\ue00a; # LETTER UU
\u0a8b>\ue00b; # LETTER VOCALIC R
# \u0a8d>; # UNMAPPED Gujarati-InterIndic: VOWEL CANDRA E
\u0a8d>\ue00d; # UNMAPPED Gujarati-InterIndic: VOWEL CANDRA E
\u0a8f>\ue00f; # LETTER E
\u0a90>\ue010; # LETTER AI
# \u0a91>; # UNMAPPED Gujarati-InterIndic: VOWEL CANDRA O
\u0a91>\ue011; # UNMAPPED Gujarati-InterIndic: VOWEL CANDRA O
\u0a93>\ue013; # LETTER O
\u0a94>\ue014; # LETTER AU
\u0a95>\ue015; # LETTER KA

10
icu4j/src/com/ibm/icu/impl/data/Transliterator_Gurmukhi_InterIndic.txt
View File

@ -86,10 +86,10 @@
\u0a6d>\ue06d; # DIGIT SEVEN
\u0a6e>\ue06e; # DIGIT EIGHT
\u0a6f>\ue06f; # DIGIT NINE
# \u0a70>; # UNMAPPED Gurmukhi-InterIndic: TIPPI
# \u0a71>; # UNMAPPED Gurmukhi-InterIndic: ADDAK
# \u0a72>; # UNMAPPED Gurmukhi-InterIndic: IRI
# \u0a73>; # UNMAPPED Gurmukhi-InterIndic: URA
# \u0a74>; # UNMAPPED Gurmukhi-InterIndic: EK ONKAR
\u0a70>\ue070; # TIPPI
\u0a71>\ue071; # ADDAK
\u0a72>\ue072; # IRI
\u0a73>\ue073; # URA
\u0a74>\ue074; # EK ONKAR
# :: NFC (NFD) ;
# eof

29
icu4j/src/com/ibm/icu/impl/data/Transliterator_InterIndic_Bengali.txt
View File

@ -27,9 +27,13 @@
\ue00a>\u098a; # LETTER UU
\ue00b>\u098b; # LETTER VOCALIC R
\ue00c>\u098c; # LETTER VOCALIC L
# \ue00f>; # UNMAPPED InterIndic-Bengali: LETTER EE (\u098f = LETTER E)
\ue00d>\u098f; # FALLBACK
\ue00e>\u098f; # FALLBACK
\ue00f>\u098f; # LETTER E
\ue010>\u0990; # LETTER AI
# \ue013>; # UNMAPPED InterIndic-Bengali: LETTER OO (\u0993 = LETTER O)
\ue011>\u0993; # FALLBACK
\ue012>\u0993; # FALLBACK
\ue013>\u0993; # LETTER O
\ue014>\u0994; # LETTER AU
\ue015>\u0995; # LETTER KA
\ue016>\u0996; # LETTER KHA
@ -59,6 +63,7 @@
\ue02e>\u09ae; # LETTER MA
\ue02f>\u09af; # LETTER YA
\ue030>\u09b0; # LETTER RA
\ue031>\u09b0\u09bc; # FALLBACK to RA
\ue032>\u09b2; # LETTER LA
\ue033>\u09b2; # REMAP (indicExceptions.txt): \u09b3>\u09b2 = LETTER LLA>LETTER LA
\ue034>\u09b2; # REMAP (indicExceptions.txt): \u09b4>\u09b2 = LETTER LLLA>LETTER LA
@ -68,7 +73,7 @@
\ue038>\u09b8; # LETTER SA
\ue039>\u09b9; # LETTER HA
\ue03c>\u09bc; # SIGN NUKTA
# \ue03d>; # UNMAPPED InterIndic-Bengali: SIGN AVAGRAHA
\ue03d>; # FALLBACK BLOW AWAY AVAGRAHA
\ue03e>\u09be; # VOWEL SIGN AA
\ue03f>\u09bf; # VOWEL SIGN I
\ue040>\u09c0; # VOWEL SIGN II
@ -77,19 +82,23 @@
\ue043>\u09c3; # VOWEL SIGN VOCALIC R
\ue044>\u09c4; # VOWEL SIGN VOCALIC RR
\ue045>\u09c7; # REMAP (indicExceptions.txt): \u09c5>\u09c7 = VOWEL SIGN CANDRA E>VOWEL SIGN E
# \ue047>; # UNMAPPED InterIndic-Bengali: VOWEL SIGN EE (\u09c7 = VOWEL SIGN E)
\ue046>\u09c7; # FALLBACK
\ue047>\u09c7; # VOWEL SIGN E
\ue048>\u09c8; # VOWEL SIGN AI
\ue049>\u09cb; # REMAP (indicExceptions.txt): \u09c9>\u09cb = VOWEL SIGN CANDRA O>VOWEL SIGN O
# \ue04b>; # UNMAPPED InterIndic-Bengali: VOWEL SIGN OO (\u09cb = VOWEL SIGN O)
\ue04a>\u09cb; # FALLBACK
\ue04b>\u09cb; # VOWEL SIGN O
\ue04c>\u09cc; # VOWEL SIGN AU
\ue04d>\u09cd; # SIGN VIRAMA
# \ue050>; # UNMAPPED InterIndic-Bengali: OM
\ue050>\u0993\u0982; # InterIndic-Bengali: OM
# \ue055>; # UNMAPPED InterIndic-Bengali: LENGTH MARK
\ue056>\u09c8; # REMAP (indicExceptions.txt): \u09d6>\u09c8 = AI LENGTH MARK>VOWEL SIGN AI
\ue057>\u09d7; # AU LENGTH MARK
\ue058>\u0995; # FALLBACK
\ue059>\u0996; # REMAP (indicExceptions.txt): \u09d9>\u0996 = LETTER KHHA>LETTER KHA
\ue05a>\u0997; # REMAP (indicExceptions.txt): \u09da>\u0997 = LETTER GHHA>LETTER GA
\ue05b>\u099c; # REMAP (indicExceptions.txt): \u09db>\u099c = LETTER ZA>LETTER JA
\ue05c>\u09Dc; # FALLBACK
\ue05d>\u09dd; # LETTER RHA
\ue05e>\u09ab; # REMAP (indicExceptions.txt): \u09de>\u09ab = LETTER FA>LETTER PHA
\ue05f>\u09df; # LETTER YYA
@ -108,10 +117,8 @@
\ue06e>\u09ee; # DIGIT EIGHT
\ue06f>\u09ef; # DIGIT NINE
\ue0fa>\u09fa; # ISSHAR
\ue00f>\u098f; # LETTER E
\ue013>\u0993; # LETTER O
\ue031>\u09dc; # LETTER RRA
\ue047>\u09c7; # VOWEL SIGN E
\ue04b>\u09cb; # VOWEL SIGN O
# :: NFC (NFD) ;
# eof

51
icu4j/src/com/ibm/icu/impl/data/Transliterator_InterIndic_Devanagari.txt
View File

@ -28,6 +28,25 @@
\ue022\ue03c > \u095d; #\ue05d LETTER RHA (pronounced RRHA)
\ue02b\ue03c > \u095e; #\ue05e LETTER FA
\ue02f\ue03c > \u095f; #\ue05f LETTER YYA
#Decomposed compatibility transliterations
\ue012\ue057>\u0914; # FALLBACK FOR TAMIL AU
\u0030 > \u0966; # FALLBACK FOR TAMIL
\ue046\ue056 > \u0948; # FALLBACK FOR TELUGU
\ue055>; # FALLBACK BLOW AWAY KANNADA AND TELUGU LENGTH MARK
\ue056>; # FALLBACK BLOW AWAY KANNADA AND TELUGU AI LENGTH MARK
\ue057>; # FALLBACK BLOW AWAY TAMIL AU LENGTH MARK
\ue070>; # FALLBACK BLOW AWAY GURMUKHI
\ue071>; # FALLBACK BLOW AWAY GURMUKHI
\ue072>; # FALLBACK BLOW AWAY GURMUKHI
\ue073>; # FALLBACK BLOW AWAY GURMUKHI
\ue074>; # FALLBACK BLOW AWAY GURMUKHI
#Decomposed compatibility transliterations
\ue032\ue03C > \u0933; # FALLBACK FOR GURMUKHI
\ue038\ue03c > \u0936; # FALLBACK FOR GURMUKHI
\ue001 > \u0901; # SIGN CANDRABINDU
\ue002 > \u0902; # SIGN ANUSVARA
\ue003 > \u0903; # SIGN VISARGA
@ -40,13 +59,13 @@
\ue00b > \u090b; # LETTER VOCALIC R
\ue00c > \u090c; # LETTER VOCALIC L
\ue00d > \u090d; # LETTER CANDRA E (For representing English sounds)
#\ue00e > \u090e; # UNMAPPED LETTER SHORT E(For Southern Scripts)
\ue00e > \u090f;
\ue00e > \u090e; # UNMAPPED LETTER SHORT E(For Southern Scripts)
#\ue00e > \u090f;
\ue00f > \u090f; # LETTER E
\ue010 > \u0910; # LETTER AI
\ue011 > \u0911; # LETTER CANDRA O (For representing English sounds)
#\ue012 > \u0912; # UNMAPPED LETTER SHORT O (For Southern Scripts)
\ue012 > \u0913;
\ue012 > \u0912; # UNMAPPED LETTER SHORT O (For Southern Scripts)
#\ue012 > \u0913;
\ue013 > \u0913; # LETTER O
\ue014 > \u0914; # LETTER AU
\ue015 > \u0915; # LETTER KA
@ -77,12 +96,12 @@
\ue02e > \u092e; # LETTER MA
\ue02f > \u092f; # LETTER YA
\ue030 > \u0930; # LETTER RA
#\ue031 > \u0931; # LETTER RRA (Eyelash RA for Southern scripts)
\ue031 > \u0930;
\ue031 > \u0931; # LETTER RRA (Eyelash RA for Southern scripts)
#\ue031 > \u0930;
\ue032 > \u0932; # LETTER LA
\ue033 > \u0933; # LETTER LLA
#\ue034 > \u0934; # LETTER LLLA (LLLA for Southern scripts)
\ue034 > \u0933;
\ue034 > \u0934; # LETTER LLLA (LLLA for Southern scripts)
#\ue034 > \u0933;
\ue035 > \u0935; # LETTER VA
\ue036 > \u0936; # LETTER SHA
\ue037 > \u0937; # LETTER SSA
@ -98,13 +117,13 @@
\ue043 > \u0943; # VOWEL SIGN VOCALIC R
\ue044 > \u0944; # VOWEL SIGN VOCALIC RR
\ue045 > \u0945; # VOWEL SIGN CANDRA E
#\ue046 > \u0946; # UNMAPPED VOWEL SIGN SHORT E
\ue046 > \u0947;
\ue046 > \u0946; # UNMAPPED VOWEL SIGN SHORT E
#\ue046 > \u0947;
\ue047 > \u0947; # VOWEL SIGN E
\ue048 > \u0948; # VOWEL SIGN AI
\ue049 > \u0949; # VOWEL SIGN CANDRA O
#\ue04a > \u094a; # UNMAPPED VOWEL SIGN SHORT O
\ue04a > \u094b;
\ue04a > \u094a; # UNMAPPED VOWEL SIGN SHORT O
#\ue04a > \u094b;
\ue04b > \u094b; # VOWEL SIGN O
\ue04c > \u094c; # VOWEL SIGN AU
\ue04d > \u094d; # SIGN VIRAMA
@ -125,8 +144,8 @@
\ue061 > \u0961; # LETTER VOCALIC LL
\ue062 > \u0962; # VOWEL SIGN VOCALIC L
\ue063 > \u0963; # VOWEL SIGN VOCALIC LL
# > ; \u0964 # UNMAPPED Devanagari-InterIndic: DANDA
# > ; \u0965 # UNMAPPED Devanagari-InterIndic: DOUBLE DANDA
\ue064 > \u0964; # UNMAPPED Devanagari-InterIndic: DANDA
\ue065 > \u0965; # UNMAPPED Devanagari-InterIndic: DOUBLE DANDA
\ue066 > \u0966; # DIGIT ZERO
\ue067 > \u0967; # DIGIT ONE
\ue068 > \u0968; # DIGIT TWO
@ -137,6 +156,10 @@
\ue06d > \u096d; # DIGIT SEVEN
\ue06e > \u096e; # DIGIT EIGHT
\ue06f > \u096f; # DIGIT NINE
\ue0f0 > \u0930; # FALLBACK RA
\ue0f1 > \u0930; # FALLBACK RA
# \u0970 # UNMAPPED Devanagari-InterIndic: ABBREVIATION SIGN
# :: NFC;
# eof

30
icu4j/src/com/ibm/icu/impl/data/Transliterator_InterIndic_Gujarati.txt
View File

@ -27,9 +27,13 @@
\ue00a>\u0a8a; # LETTER UU
\ue00b>\u0a8b; # LETTER VOCALIC R
\ue00c>\u0ab2\u0ac3; # REMAP (indicExceptions.txt): \u0a8c>\u0ab2\u0ac3 = LETTER VOCALIC L>LETTER LA.VOWEL SIGN VOCALIC R
# \ue00f>; # UNMAPPED InterIndic-Gujarati: LETTER EE (\u0a8f = LETTER E)
\ue00d>\u0a8d; # GUJARATI VOWEL CANDRA E
\ue00e>\u0a8f; # FALLBACK
\ue00f>\u0a8f; # InterIndic-Gujarati: LETTER EE (\u0a8f = LETTER E)
\ue010>\u0a90; # LETTER AI
# \ue013>; # UNMAPPED InterIndic-Gujarati: LETTER OO (\u0a93 = LETTER O)
\ue011>\u0a91; # FALLBACK
\ue012>\u0a93; # FALLBACK
\ue013>\u0a93; # UNMAPPED InterIndic-Gujarati: LETTER OO (\u0a93 = LETTER O)
\ue014>\u0a94; # LETTER AU
\ue015>\u0a95; # LETTER KA
\ue016>\u0a96; # LETTER KHA
@ -51,7 +55,7 @@
\ue026>\u0aa6; # LETTER DA
\ue027>\u0aa7; # LETTER DHA
\ue028>\u0aa8; # LETTER NA
\ue029>\u0aa8; # REMAP (indicExceptions.txt): \u0aa9>\u0aa8 = LETTER NNNA>LETTER NA
\ue029>\u0aa8\u0abc; # FALLBACK to NA+NUKTA
\ue02a>\u0aaa; # LETTER PA
\ue02b>\u0aab; # LETTER PHA
\ue02c>\u0aac; # LETTER BA
@ -59,9 +63,10 @@
\ue02e>\u0aae; # LETTER MA
\ue02f>\u0aaf; # LETTER YA
\ue030>\u0ab0; # LETTER RA
\ue031>\u0ab0\u0abc; # FALLBACK
\ue032>\u0ab2; # LETTER LA
\ue033>\u0ab3; # LETTER LLA
\ue034>\u0ab3; # REMAP (indicExceptions.txt): \u0ab4>\u0ab3 = LETTER LLLA>LETTER LLA
\ue034>\u0ab3\u0abc; # LETTER LLLA>LETTER LLA+NUKTA
\ue035>\u0ab5; # LETTER VA
\ue036>\u0ab6; # LETTER SHA
\ue037>\u0ab7; # LETTER SSA
@ -77,19 +82,23 @@
\ue043>\u0ac3; # VOWEL SIGN VOCALIC R
\ue044>\u0ac4; # VOWEL SIGN VOCALIC RR
\ue045>\u0ac5; # VOWEL SIGN CANDRA E
# \ue047>; # UNMAPPED InterIndic-Gujarati: VOWEL SIGN EE (\u0ac7 = VOWEL SIGN E)
\ue046>\u0ac7; # FALLBACK
\ue047>\u0ac7; # InterIndic-Gujarati: VOWEL SIGN EE (\u0ac7 = VOWEL SIGN E)
\ue048>\u0ac8; # VOWEL SIGN AI
\ue049>\u0ac9; # VOWEL SIGN CANDRA O
# \ue04b>; # UNMAPPED InterIndic-Gujarati: VOWEL SIGN OO (\u0acb = VOWEL SIGN O)
\ue04a>\u0acb; # FALLBACK
\ue04b>\u0acb; # UNMAPPED InterIndic-Gujarati: VOWEL SIGN OO (\u0acb = VOWEL SIGN O)
\ue04c>\u0acc; # VOWEL SIGN AU
\ue04d>\u0acd; # SIGN VIRAMA
\ue050>\u0ad0; # OM
# \ue055>; # UNMAPPED InterIndic-Gujarati: LENGTH MARK
#\ue055>; # UNMAPPED InterIndic-Gujarati: LENGTH MARK
\ue056>\u0ac8; # REMAP (indicExceptions.txt): \u0ad6>\u0ac8 = AI LENGTH MARK>VOWEL SIGN AI
\ue057>\u0acc; # REMAP (indicExceptions.txt): \u0ad7>\u0acc = AU LENGTH MARK>VOWEL SIGN AU
\ue058>\u0a95\u0abc; # FALLBACK
\ue059>\u0a96\u0abc; # REMAP (indicExceptions.txt): \u0ad9>\u0a96\u0abc = LETTER KHHA>LETTER KHA.SIGN NUKTA
\ue05a>\u0a97\u0abc; # REMAP (indicExceptions.txt): \u0ada>\u0a97\u0abc = LETTER GHHA>LETTER GA.SIGN NUKTA
\ue05b>\u0a9c\u0abc; # REMAP (indicExceptions.txt): \u0adb>\u0a9c\u0abc = LETTER ZA>LETTER JA.SIGN NUKTA
\ue05c>\u0aa1\u0abc; # FALLBACK
\ue05d>\u0aa2\u0abc; # REMAP (indicExceptions.txt): \u0add>\u0aa2\u0abc = LETTER RHA>LETTER DDHA.SIGN NUKTA
\ue05e>\u0aab\u0abc; # REMAP (indicExceptions.txt): \u0ade>\u0aab\u0abc = LETTER FA>LETTER PHA.SIGN NUKTA
\ue05f>\u0aaf\u0abc; # REMAP (indicExceptions.txt): \u0adf>\u0aaf\u0abc = LETTER YYA>LETTER YA.SIGN NUKTA
@ -107,11 +116,6 @@
\ue06d>\u0aed; # DIGIT SEVEN
\ue06e>\u0aee; # DIGIT EIGHT
\ue06f>\u0aef; # DIGIT NINE
# \ue080>; # UNMAPPED InterIndic-Gujarati: ISSHAR
\ue00f>\u0a8f; # LETTER E
\ue013>\u0a93; # LETTER O
# \ue083>; # UNMAPPED InterIndic-Gujarati: LETTER RRA (\u0a83 = SIGN VISARGA)
\ue047>\u0ac7; # VOWEL SIGN E
\ue04b>\u0acb; # VOWEL SIGN O
#\ue080>; # UNMAPPED InterIndic-Gujarati: ISSHAR
# :: NFC (NFD) ;
# eof

47
icu4j/src/com/ibm/icu/impl/data/Transliterator_InterIndic_Gurmukhi.txt
View File

@ -17,8 +17,8 @@
# InterIndic-Gurmukhi
#:: NFD (NFC) ;
\ue001>\u0a02; # REMAP (indicExceptions.txt): \u0a01>\u0a02 = SIGN CANDRABINDU>SIGN BINDI
# \ue002>; # UNMAPPED InterIndic-Gurmukhi: SIGN ANUSVARA (\u0a02 = SIGN BINDI)
# \ue003>; # UNMAPPED InterIndic-Gurmukhi: SIGN VISARGA
\ue002>; # FALLBACK BLOW AWAY SIGN ANUSVARA (\u0a02 = SIGN BINDI)
\ue003>; # FALLBACK BLOW AWAY SIGN VISARGA
\ue005>\u0a05; # LETTER A
\ue006>\u0a06; # LETTER AA
\ue007>\u0a07; # LETTER I
@ -26,9 +26,13 @@
\ue009>\u0a09; # LETTER U
\ue00a>\u0a0a; # LETTER UU
\ue00b>\u0a30\u0a3f; # REMAP (indicExceptions.txt): \u0a0b>\u0a30\u0a3f = LETTER VOCALIC R>LETTER RA.VOWEL SIGN I
\ue00c>\u0a07; # REMAP (indicExceptions.txt): \u0a0c>\u0a07 = LETTER VOCALIC L>LETTER I
\ue00c>\u0a33; # FALLBACK
\ue00d>\u0a0f; # FALLBACK
\ue00e>\u0a0f; # FALLBACK
\ue00f>\u0a0f; # LETTER EE
\ue010>\u0a10; # LETTER AI
\ue011>\u0a13; # FALLBACK
\ue012>\u0a13; # FALLBACK
\ue013>\u0a13; # LETTER OO
\ue014>\u0a14; # LETTER AU
\ue015>\u0a15; # LETTER KA
@ -51,7 +55,7 @@
\ue026>\u0a26; # LETTER DA
\ue027>\u0a27; # LETTER DHA
\ue028>\u0a28; # LETTER NA
\ue029>\u0a28; # REMAP (indicExceptions.txt): \u0a29>\u0a28 = LETTER NNNA>LETTER NA
\ue029>\u0a28\u0a3c; # REMAP (indicExceptions.txt): \u0a29>\u0a28 = LETTER NNNA>LETTER NA
\ue02a>\u0a2a; # LETTER PA
\ue02b>\u0a2b; # LETTER PHA
\ue02c>\u0a2c; # LETTER BA
@ -59,6 +63,7 @@
\ue02e>\u0a2e; # LETTER MA
\ue02f>\u0a2f; # LETTER YA
\ue030>\u0a30; # LETTER RA
\ue031>\u0a30\u0a3c; # FALLBACK LETTER RA+NUKTA
\ue032>\u0a32; # LETTER LA
\ue033>\u0a33; # LETTER LLA
\ue034>\u0a33; # REMAP (indicExceptions.txt): \u0a34>\u0a33 = LETTER LLLA>LETTER LLA
@ -68,33 +73,37 @@
\ue038>\u0a38; # LETTER SA
\ue039>\u0a39; # LETTER HA
\ue03c>\u0a3c; # SIGN NUKTA
# \ue03d>; # UNMAPPED InterIndic-Gurmukhi: SIGN AVAGRAHA
\ue03d>; # FALLBACK BLOW AWAY SIGN AVAGRAHA
\ue03e>\u0a3e; # VOWEL SIGN AA
\ue03f>\u0a3f; # VOWEL SIGN I
\ue040>\u0a40; # VOWEL SIGN II
\ue041>\u0a41; # VOWEL SIGN U
\ue042>\u0a42; # VOWEL SIGN UU
# \ue043>; # UNMAPPED InterIndic-Gurmukhi: VOWEL SIGN VOCALIC R
# \ue044>; # UNMAPPED InterIndic-Gurmukhi: VOWEL SIGN VOCALIC RR
\ue043>; # FALLBACK BLOW AWAY VOWEL SIGN VOCALIC R
\ue044>; # FALLBACK BLOW AWAY VOWEL SIGN VOCALIC RR
\ue045>\u0a48; # REMAP (indicExceptions.txt): \u0a45>\u0a48 = VOWEL SIGN CANDRA E>VOWEL SIGN AI
\ue046>\u0a47; # FALLABCK
\ue047>\u0a47; # VOWEL SIGN EE
\ue048>\u0a48; # VOWEL SIGN AI
\ue049>\u0a4c; # REMAP (indicExceptions.txt): \u0a49>\u0a4c = VOWEL SIGN CANDRA O>VOWEL SIGN AU
\ue04a>\u0a4b; # FALLBACK
\ue04b>\u0a4b; # VOWEL SIGN OO
\ue04c>\u0a4c; # VOWEL SIGN AU
\ue04d>\u0a4d; # SIGN VIRAMA
# \ue050>; # UNMAPPED InterIndic-Gurmukhi: OM
# \ue055>; # UNMAPPED InterIndic-Gurmukhi: LENGTH MARK
\ue050>\u0a0f\u0a02; # FALLBACK to OO+BINDI : OM
\ue055>; # FALLBACK BLOW AWAY LENGTH MARK
\ue056>\u0a48; # REMAP (indicExceptions.txt): \u0a56>\u0a48 = AI LENGTH MARK>VOWEL SIGN AI
\ue057>\u0a4c; # REMAP (indicExceptions.txt): \u0a57>\u0a4c = AU LENGTH MARK>VOWEL SIGN AU
\ue058>\u0a15\u0a3c; # FALLBACK RA+ NUKTA
\ue059>\u0a59; # LETTER KHHA
\ue05a>\u0a5a; # LETTER GHHA
\ue05b>\u0a5b; # LETTER ZA
\ue05c>\u0a5c; # LETTER RRA
\ue05d>\u0a22\u0a3c; # REMAP (indicExceptions.txt): \u0a5d>\u0a22\u0a3c = LETTER RHA>LETTER DDHA.SIGN NUKTA
\ue05e>\u0a5e; # LETTER FA
\ue05f>\u0a2f; # REMAP (indicExceptions.txt): \u0a5f>\u0a2f = LETTER YYA>LETTER YA
\ue05f>\u0a2f\u0a3c; # REMAP (indicExceptions.txt): \u0a5f>\u0a2f = LETTER YYA>LETTER YA
\ue060>\u0a30\u0a3f; # REMAP (indicExceptions.txt): \u0a60>\u0a30\u0a3f = LETTER VOCALIC RR>LETTER RA.VOWEL SIGN I
\ue061>\u0a08\u0a3c; # REMAP (indicExceptions.txt): \u0a61>\u0a08\u0a3c = LETTER VOCALIC LL>LETTER II.SIGN NUKTA
\ue061>\u0a33\u0a3c; #
\ue062>\u0a3f\u0a3c; # REMAP (indicExceptions.txt): \u0a62>\u0a3f\u0a3c = VOWEL SIGN VOCALIC L>VOWEL SIGN I.SIGN NUKTA
\ue063>\u0a40\u0a3c; # REMAP (indicExceptions.txt): \u0a63>\u0a40\u0a3c = VOWEL SIGN VOCALIC LL>VOWEL SIGN II.SIGN NUKTA
\ue066>\u0a66; # DIGIT ZERO
@ -107,11 +116,15 @@
\ue06d>\u0a6d; # DIGIT SEVEN
\ue06e>\u0a6e; # DIGIT EIGHT
\ue06f>\u0a6f; # DIGIT NINE
# \ue080>; # UNMAPPED InterIndic-Gurmukhi: ISSHAR
# \ue081>; # UNMAPPED InterIndic-Gurmukhi: LETTER E
# \ue082>; # UNMAPPED InterIndic-Gurmukhi: LETTER O (\u0a02 = SIGN BINDI)
\ue05c>\u0a5c; # LETTER RRA
# \ue084>; # UNMAPPED InterIndic-Gurmukhi: VOWEL SIGN E
# \ue085>; # UNMAPPED InterIndic-Gurmukhi: VOWEL SIGN O (\u0a05 = LETTER A)
\ue070>\u0a70; # TIPPI
\ue071>\u0a71; # ADDAK
\ue072>\u0a72; # IRI
\ue073>\u0a73; # URA
\ue074>\u0a74; # EK ONKAR
\ue080>; # FALLBACK BLOW AWAY ISSHAR
\ue081>; # FALLBACK BLOW AWAY LETTER E
\ue082>; # FALLBACK BLOW AWAY LETTER O (\u0a02 = SIGN BINDI)
\ue084>; # FALLBACK BLOW AWAY VOWEL SIGN E
\ue085>; # FALLBACK BLOW AWAY VOWEL SIGN O (\u0a05 = LETTER A)
# :: NFC (NFD) ;
# eof

20
icu4j/src/com/ibm/icu/impl/data/Transliterator_InterIndic_Kannada.txt
View File

@ -27,8 +27,12 @@
\ue00a>\u0c8a; # LETTER UU
\ue00b>\u0c8b; # LETTER VOCALIC R
\ue00c>\u0c8c; # LETTER VOCALIC L
\ue00d>\u0c8e; # LETTER E
\ue00e>\u0c8e; # FALLBACK
\ue00f>\u0c8f; # LETTER EE
\ue010>\u0c90; # LETTER AI
\ue011>\u0c92; # FALLBACK
\ue012>\u0c92; # LETTER O
\ue013>\u0c93; # LETTER OO
\ue014>\u0c94; # LETTER AU
\ue015>\u0c95; # LETTER KA
@ -59,6 +63,7 @@
\ue02e>\u0cae; # LETTER MA
\ue02f>\u0caf; # LETTER YA
\ue030>\u0cb0; # LETTER RA
\ue031>\u0cb1; # LETTER RRA
\ue032>\u0cb2; # LETTER LA
\ue033>\u0cb3; # LETTER LLA
\ue034>\u0cb3; # REMAP (indicExceptions.txt): \u0cb4>\u0cb3 = LETTER LLLA>LETTER LLA
@ -67,8 +72,10 @@
\ue037>\u0cb7; # LETTER SSA
\ue038>\u0cb8; # LETTER SA
\ue039>\u0cb9; # LETTER HA
# \ue03c>; # UNMAPPED InterIndic-Kannada: SIGN NUKTA
# \ue03d>; # UNMAPPED InterIndic-Kannada: SIGN AVAGRAHA
\ue03c>; # FALLBACK BLOW AWAY NUKTA
\ue03d>; # FALLBACK BLOW AWAY AVAGRAHA
\ue03e>\u0cbe; # VOWEL SIGN AA
\ue03f>\u0cbf; # VOWEL SIGN I
\ue040>\u0cc0; # VOWEL SIGN II
@ -77,9 +84,11 @@
\ue043>\u0cc3; # VOWEL SIGN VOCALIC R
\ue044>\u0cc4; # VOWEL SIGN VOCALIC RR
\ue045>\u0cc6; # REMAP (indicExceptions.txt): \u0cc5>\u0cc6 = VOWEL SIGN CANDRA E>VOWEL SIGN E
\ue046>\u0cc6; # VOWEL SIGN E
\ue047>\u0cc7; # VOWEL SIGN EE
\ue048>\u0cc8; # VOWEL SIGN AI
\ue049>\u0cca; # REMAP (indicExceptions.txt): \u0cc9>\u0cca = VOWEL SIGN CANDRA O>VOWEL SIGN O
\ue04a>\u0cca; # VOWEL SIGN O
\ue04b>\u0ccb; # VOWEL SIGN OO
\ue04c>\u0ccc; # VOWEL SIGN AU
\ue04d>\u0ccd; # SIGN VIRAMA
@ -87,9 +96,11 @@
\ue055>\u0cd5; # LENGTH MARK
\ue056>\u0cd6; # AI LENGTH MARK
\ue057>\u0ccc; # REMAP (indicExceptions.txt): \u0cd7>\u0ccc = AU LENGTH MARK>VOWEL SIGN AU
\ue058>\u0c95; # FALLBACK
\ue059>\u0c96; # REMAP (indicExceptions.txt): \u0cd9>\u0c96 = LETTER KHHA>LETTER KHA
\ue05a>\u0c97; # REMAP (indicExceptions.txt): \u0cda>\u0c97 = LETTER GHHA>LETTER GA
\ue05b>\u0c9c; # REMAP (indicExceptions.txt): \u0cdb>\u0c9c = LETTER ZA>LETTER JA
\ue05c>\u0ca2; # FALLBACK
\ue05d>\u0ca2; # REMAP (indicExceptions.txt): \u0cdd>\u0ca2 = LETTER RHA>LETTER DDHA
\ue05e>\u0cde; # LETTER FA
\ue05f>\u0caf; # REMAP (indicExceptions.txt): \u0cdf>\u0caf = LETTER YYA>LETTER YA
@ -108,10 +119,5 @@
\ue06e>\u0cee; # DIGIT EIGHT
\ue06f>\u0cef; # DIGIT NINE
# \ue080>; # UNMAPPED InterIndic-Kannada: ISSHAR
\ue00e>\u0c8e; # LETTER E
\ue012>\u0c92; # LETTER O
\ue031>\u0cb1; # LETTER RRA
\ue046>\u0cc6; # VOWEL SIGN E
\ue04a>\u0cca; # VOWEL SIGN O
# :: NFC (NFD) ;
# eof

66
icu4j/src/com/ibm/icu/impl/data/Transliterator_InterIndic_Latin.txt
View File

@ -87,7 +87,7 @@
$virama=\ue04d;
# \u094e Reserved
# \u094f Reserved
#\u0950>\ue050; # OM
$om=\ue050; # OM
# \u0951>; # UNMAPPED STRESS SIGN UDATTA
# \u0952>; # UNMAPPED STRESS SIGN ANUDATTA
# \u0953>; # UNMAPPED GRAVE ACCENT
@ -238,6 +238,9 @@
$ra$virama}$ha>r'';
$ra}$x>r;
$ra>ra;
$rra$virama}$ha>r\u0331'';
$rra}$x>r\u0331;
$rra>r\u0331a;
$la$virama}$ha>l'';
$la}$x>l;
$la>la;
@ -267,8 +270,8 @@
$ena > n\u0331a ;
$uka}$x > q ;
$uka > qa ;
$ukha}$x > k\u0323 ;
$ukha > k\u0323a ;
$ukha}$x > k\u0331h\u0331 ;
$ukha > k\u0331h\u0331a ;
$ugha}$x > g\u0307 ;
$ugha > g\u0307a ;
$ujha}$x > z ;
@ -277,8 +280,8 @@
$udha > r\u0323ha;
$uddha}$x> r\u0323 ;
$uddha > r\u0323a ;
$ufa}$x > f\u0323 ;
$ufa > f\u0323a ;
$ufa}$x > f ;
$ufa > fa ;
# dependent vowels (should never occur except following consonants)
$aa > a\u0304 ;
$ai > ai ;
@ -300,33 +303,33 @@
$se > e ;
$so > o ;
# independent vowels (when following consonants)
a}$waa > ''a\u0304 ;
$z}$waa > ''a\u0304 ;
a}$wai > ''ai ;
$z}$wai > ''ai ;
a}$wau > ''au ;
$z}$wau > ''au ;
a}$wii > ''i\u0304 ;
$z}$wii > ''i\u0304 ;
a}$wi > ''i ;
$z}$wi > ''i ;
a}$wuu > ''u\u0304 ;
$z}$wuu > ''u\u0304 ;
a}$wu > ''u ;
$z}$wu > ''u ;
$z}$wrr > ''r\u0325\u0304 ;
$z}$wr > ''r\u0325 ;
$z}$wll > ''l\u0325\u0304 ;
$z}$wl > ''l\u0325 ;
$z}$we > ''e\u0304 ;
$z}$wo > ''o\u0304 ;
a}$wa > ''a ;
$z}$wa > ''a ;
a{$waa > ''a\u0304 ;
$z{$waa > ''a\u0304 ;
a{$wai > ''ai ;
$z{$wai > ''ai ;
a{$wau > ''au ;
$z{$wau > ''au ;
a{$wii > ''i\u0304 ;
$z{$wii > ''i\u0304 ;
a{$wi > ''i ;
$z{$wi > ''i ;
a{$wuu > ''u\u0304 ;
$z{$wuu > ''u\u0304 ;
a{$wu > ''u ;
$z{$wu > ''u ;
$z{$wrr > ''r\u0325\u0304 ;
$z{$wr > ''r\u0325 ;
$z{$wll > ''l\u0325\u0304 ;
$z{$wl > ''l\u0325 ;
$z{$we > ''e\u0304 ;
$z{$wo > ''o\u0304 ;
a{$wa > ''a ;
$z{$wa > ''a ;
#extra vowels
$z}$wce > ''e\u0306 ;
$z}$wco > ''o\u0306 ;
$z}$wse > ''e ;
$z}$wso > ''o ;
$z{$wce > ''e\u0306 ;
$z{$wco > ''o\u0306 ;
$z{$wse > ''e ;
$z{$wso > ''o ;
# independent vowels (otherwise)
$waa > a\u0304 ;
$wai > ai ;
@ -347,6 +350,7 @@
$wco > o\u0306 ;
$wse > e ;
$wso > o ;
$om > ''om ;
#stress marks
$avagraha > \u0315;
$chandrabindu$anusvara>'-'\u0303;

35
icu4j/src/com/ibm/icu/impl/data/Transliterator_InterIndic_Malayalam.txt
View File

@ -27,8 +27,12 @@
\ue00a>\u0d0a; # LETTER UU
\ue00b>\u0d0b; # LETTER VOCALIC R
\ue00c>\u0d0c; # LETTER VOCALIC L
\ue00d>\u0d0e; # FALLLBACK LETTER E
\ue00e>\u0d0e; # LETTER E
\ue00f>\u0d0f; # LETTER EE
\ue010>\u0d10; # LETTER AI
\ue011>\u0d12; # FALLBACK TO O
\ue012>\u0d12; # LETTER O
\ue013>\u0d13; # LETTER OO
\ue014>\u0d14; # LETTER AU
\ue015>\u0d15; # LETTER KA
@ -59,6 +63,7 @@
\ue02e>\u0d2e; # LETTER MA
\ue02f>\u0d2f; # LETTER YA
\ue030>\u0d30; # LETTER RA
\ue031>\u0d31; # LETTER RRA
\ue032>\u0d32; # LETTER LA
\ue033>\u0d33; # LETTER LLA
\ue034>\u0d34; # LETTER LLLA
@ -67,36 +72,42 @@
\ue037>\u0d37; # LETTER SSA
\ue038>\u0d38; # LETTER SA
\ue039>\u0d39; # LETTER HA
# \ue03c>; # UNMAPPED InterIndic-Malayalam: SIGN NUKTA
# \ue03d>; # UNMAPPED InterIndic-Malayalam: SIGN AVAGRAHA
\ue03c>; # FALLBACK BLOW AWAY NUKTA
\ue03d>; # FALLBACK BLOW AWAY AVAGRAHA
\ue03e>\u0d3e; # VOWEL SIGN AA
\ue03f>\u0d3f; # VOWEL SIGN I
\ue040>\u0d40; # VOWEL SIGN II
\ue041>\u0d41; # VOWEL SIGN U
\ue042>\u0d42; # VOWEL SIGN UU
\ue043>\u0d43; # VOWEL SIGN VOCALIC R
# \ue044>; # UNMAPPED InterIndic-Malayalam: VOWEL SIGN VOCALIC RR
\ue044>; # FALLBACK BLOW AWAY VOWEL SIGN VOCALIC RR
\ue045>\u0d3e; # REMAP (indicExceptions.txt): \u0d45>\u0d3e = VOWEL SIGN CANDRA E>VOWEL SIGN AA
\ue046>\u0d46; # VOWEL SIGN E
\ue047>\u0d47; # VOWEL SIGN EE
\ue048>\u0d48; # VOWEL SIGN AI
\ue049>\u0d4b; # REMAP (indicExceptions.txt): \u0d49>\u0d4b = VOWEL SIGN CANDRA O>VOWEL SIGN OO
\ue04a>\u0d4a; # VOWEL SIGN O
\ue04b>\u0d4b; # VOWEL SIGN OO
\ue04c>\u0d4c; # VOWEL SIGN AU
\ue04d>\u0d4d; # SIGN VIRAMA
# \ue050>; # UNMAPPED InterIndic-Malayalam: OM
# \ue055>; # UNMAPPED InterIndic-Malayalam: LENGTH MARK
\ue050>\u0d13\u0d02; # UNMAPPED InterIndic-Malayalam: OM
\ue055>; # FALLBACK BLOW AWAY LENGTH MARK
\ue056>\u0d48; # REMAP (indicExceptions.txt): \u0d56>\u0d48 = AI LENGTH MARK>VOWEL SIGN AI
\ue057>\u0d57; # AU LENGTH MARK
\ue058>\u0d15; # FALLBACK
\ue059>\u0d16; # REMAP (indicExceptions.txt): \u0d59>\u0d16 = LETTER KHHA>LETTER KHA
\ue05a>\u0d17; # REMAP (indicExceptions.txt): \u0d5a>\u0d17 = LETTER GHHA>LETTER GA
\ue05b>\u0d1c; # REMAP (indicExceptions.txt): \u0d5b>\u0d1c = LETTER ZA>LETTER JA
\ue05d>\u0d22; # REMAP (indicExceptions.txt): \u0d5d>\u0d22 = LETTER RHA>LETTER DDHA
\ue05c>\u0d21; # FALLBACK
\ue05e>\u0d2b; # REMAP (indicExceptions.txt): \u0d5e>\u0d2b = LETTER FA>LETTER PHA
\ue05f>\u0d2f; # REMAP (indicExceptions.txt): \u0d5f>\u0d2f = LETTER YYA>LETTER YA
\ue060>\u0d60; # LETTER VOCALIC RR
\ue061>\u0d61; # LETTER VOCALIC LL
# \ue062>; # UNMAPPED InterIndic-Malayalam: VOWEL SIGN VOCALIC L
# \ue063>; # UNMAPPED InterIndic-Malayalam: VOWEL SIGN VOCALIC LL
\ue062>; # FALLBACK BLOW AWAY VOWEL SIGN VOCALIC L
\ue063>; # FALLBACK BLOW AWAY VOWEL SIGN VOCALIC LL
\ue066>\u0d66; # DIGIT ZERO
\ue067>\u0d67; # DIGIT ONE
\ue068>\u0d68; # DIGIT TWO
@ -108,10 +119,10 @@
\ue06e>\u0d6e; # DIGIT EIGHT
\ue06f>\u0d6f; # DIGIT NINE
# \ue080>; # UNMAPPED InterIndic-Malayalam: ISSHAR
\ue00e>\u0d0e; # LETTER E
\ue012>\u0d12; # LETTER O
\ue031>\u0d31; # LETTER RRA
\ue046>\u0d46; # VOWEL SIGN E
\ue04a>\u0d4a; # VOWEL SIGN O
# :: NFC (NFD) ;
# eof

32
icu4j/src/com/ibm/icu/impl/data/Transliterator_InterIndic_Oriya.txt
View File

@ -27,9 +27,13 @@
\ue00a>\u0b0a; # LETTER UU
\ue00b>\u0b0b; # LETTER VOCALIC R
\ue00c>\u0b0c; # LETTER VOCALIC L
# \ue00f>; # UNMAPPED InterIndic-Oriya: LETTER EE (\u0b0f = LETTER E)
\ue00d>\u0b0f; # FALLBACK
\ue00e>\u0b0f; # FALLBACK
\ue00f>\u0b0f; # LETTER E
\ue010>\u0b10; # LETTER AI
# \ue013>; # UNMAPPED InterIndic-Oriya: LETTER OO (\u0b13 = LETTER O)
\ue011>\u0b13; # FALLBACK
\ue012>\u0b13; # FALLBACK
\ue013>\u0b13; # UNMAPPED InterIndic-Oriya: LETTER OO (\u0b13 = LETTER O)
\ue014>\u0b14; # LETTER AU
\ue015>\u0b15; # LETTER KA
\ue016>\u0b16; # LETTER KHA
@ -51,7 +55,7 @@
\ue026>\u0b26; # LETTER DA
\ue027>\u0b27; # LETTER DHA
\ue028>\u0b28; # LETTER NA
\ue029>\u0b28; # REMAP (indicExceptions.txt): \u0b29>\u0b28 = LETTER NNNA>LETTER NA
\ue029>\u0b28\u0b3c; # REMAP (indicExceptions.txt): \u0b29>\u0b28 = LETTER NNNA>LETTER NA
\ue02a>\u0b2a; # LETTER PA
\ue02b>\u0b2b; # LETTER PHA
\ue02c>\u0b2c; # LETTER BA
@ -59,9 +63,10 @@
\ue02e>\u0b2e; # LETTER MA
\ue02f>\u0b2f; # LETTER YA
\ue030>\u0b30; # LETTER RA
\ue031>\u0b5c; # LETTER RRA
\ue032>\u0b32; # LETTER LA
\ue033>\u0b33; # LETTER LLA
\ue034>\u0b33; # REMAP (indicExceptions.txt): \u0b34>\u0b33 = LETTER LLLA>LETTER LLA
\ue034>\u0b33\u0b3c; # REMAP (indicExceptions.txt): \u0b34>\u0b33 = LETTER LLLA>LETTER LLA
\ue035>\u0b2c; # REMAP (indicExceptions.txt): \u0b35>\u0b2c = LETTER VA>LETTER BA
\ue036>\u0b36; # LETTER SHA
\ue037>\u0b37; # LETTER SSA
@ -76,11 +81,13 @@
\ue042>\u0b42; # VOWEL SIGN UU
\ue043>\u0b43; # VOWEL SIGN VOCALIC R
\ue044>\u0b43\u0b3c; # REMAP (indicExceptions.txt): \u0b44>\u0b43\u0b3c = VOWEL SIGN VOCALIC RR>VOWEL SIGN VOCALIC R.SIGN NUKTA
\ue045>\u0b47; # REMAP (indicExceptions.txt): \u0b45>\u0b47 = VOWEL SIGN CANDRA E>VOWEL SIGN E
# \ue047>; # UNMAPPED InterIndic-Oriya: VOWEL SIGN EE (\u0b47 = VOWEL SIGN E)
\ue045>\u0b47; # FALLBACK
\ue046>\u0b47; # FALLBACK
\ue047>\u0b47; # VOWEL SIGN E
\ue048>\u0b48; # VOWEL SIGN AI
\ue049>\u0b4b; # REMAP (indicExceptions.txt): \u0b49>\u0b4b = VOWEL SIGN CANDRA O>VOWEL SIGN O
# \ue04b>; # UNMAPPED InterIndic-Oriya: VOWEL SIGN OO (\u0b4b = VOWEL SIGN O)
\ue049>\u0b4b; # FALLBACK
\ue04a>\u0b4b; # FALLBACK
\ue04b>\u0b4b; # VOWEL SIGN E
\ue04c>\u0b4c; # VOWEL SIGN AU
\ue04d>\u0b4d; # SIGN VIRAMA
\ue050>\u0b13\u0b01; # REMAP (indicExceptions.txt): \u0b50>\u0b13\u0b01 = OM>LETTER O.SIGN CANDRABINDU
@ -88,8 +95,10 @@
\ue056>\u0b56; # AI LENGTH MARK
\ue057>\u0b57; # AU LENGTH MARK
\ue059>\u0b16\u0b3c; # REMAP (indicExceptions.txt): \u0b59>\u0b16\u0b3c = LETTER KHHA>LETTER KHA.SIGN NUKTA
\ue058>\u0b15\u0b3c; # FALLBACK
\ue05a>\u0b17\u0b3c; # REMAP (indicExceptions.txt): \u0b5a>\u0b17\u0b3c = LETTER GHHA>LETTER GA.SIGN NUKTA
\ue05b>\u0b1c\u0b3c; # REMAP (indicExceptions.txt): \u0b5b>\u0b1c\u0b3c = LETTER ZA>LETTER JA.SIGN NUKTA
\ue05c>\u0b21\u0b3c; # FALLBACK
\ue05d>\u0b5d; # LETTER RHA
\ue05e>\u0b2b\u0b3c; # REMAP (indicExceptions.txt): \u0b5e>\u0b2b\u0b3c = LETTER FA>LETTER PHA.SIGN NUKTA
\ue05f>\u0b5f; # LETTER YYA
@ -108,10 +117,7 @@
\ue06e>\u0b6e; # DIGIT EIGHT
\ue06f>\u0b6f; # DIGIT NINE
\ue070>\u0b70; # ISSHAR
\ue00e>\u0b0f; # LETTER E
\ue013>\u0b13; # LETTER O
\ue031>\u0b5c; # LETTER RRA
\ue047>\u0b47; # VOWEL SIGN E
\ue04b>\u0b4b; # VOWEL SIGN O
# :: NFC (NFD) ;
# eof

29
icu4j/src/com/ibm/icu/impl/data/Transliterator_InterIndic_Tamil.txt
View File

@ -16,7 +16,7 @@
#--------------------------------------------------------------------
# InterIndic-Tamil
#:: NFD (NFC) ;
# \ue001>; # UNMAPPED InterIndic-Tamil: SIGN CANDRABINDU
\ue001>\u0b82; # FALLBACK SIGN CANDRABINDU
\ue002>\u0b82; # SIGN ANUSVARA
\ue003>\u0b83; # SIGN VISARGA
\ue005>\u0b85; # LETTER A
@ -27,8 +27,12 @@
\ue00a>\u0b8a; # LETTER UU
\ue00b>\u0bb0\u0bbf; # REMAP (indicExceptions.txt): \u0b8b>\u0bb0\u0bbf = LETTER VOCALIC R>LETTER RA.VOWEL SIGN I
\ue00c>\u0b87; # REMAP (indicExceptions.txt): \u0b8c>\u0b87 = LETTER VOCALIC L>LETTER I
\ue00d>\u0b8f; # FALLBACK
\ue00e>\u0b8e; # LETTER E
\ue00f>\u0b8f; # LETTER EE
\ue010>\u0b90; # LETTER AI
\ue011>\u0b92; # FALLBACK
\ue012>\u0b92; # LETTER O
\ue013>\u0b93; # LETTER OO
\ue014>\u0b94; # LETTER AU
\ue015>\u0b95; # LETTER KA
@ -59,6 +63,7 @@
\ue02e>\u0bae; # LETTER MA
\ue02f>\u0baf; # LETTER YA
\ue030>\u0bb0; # LETTER RA
\ue031>\u0bb1; # LETTER RRA
\ue032>\u0bb2; # LETTER LA
\ue033>\u0bb3; # LETTER LLA
\ue034>\u0bb4; # LETTER LLLA
@ -67,8 +72,10 @@
\ue037>\u0bb7; # LETTER SSA
\ue038>\u0bb8; # LETTER SA
\ue039>\u0bb9; # LETTER HA
# \ue03c>; # UNMAPPED InterIndic-Tamil: SIGN NUKTA
# \ue03d>; # UNMAPPED InterIndic-Tamil: SIGN AVAGRAHA
\ue03c>; # FALLBACK BLOW AWAY NUKTA
\ue03d>; # FALLBACK BLOW AWAY AVAGRAHA
\ue03e>\u0bbe; # VOWEL SIGN AA
\ue03f>\u0bbf; # VOWEL SIGN I
\ue040>\u0bc0; # VOWEL SIGN II
@ -77,27 +84,32 @@
\ue043>\u0bcd\u0bb0\u0bbf; # REMAP (indicExceptions.txt): \u0bc3>\u0bcd\u0bb0\u0bbf = VOWEL SIGN VOCALIC R>SIGN VIRAMA.LETTER RA.VOWEL SIGN I
\ue044>\u0bcd\u0bb0\u0bbf; # REMAP (indicExceptions.txt): \u0bc4>\u0bcd\u0bb0\u0bbf = VOWEL SIGN VOCALIC RR>SIGN VIRAMA.LETTER RA.VOWEL SIGN I
\ue045>\u0bbe; # REMAP (indicExceptions.txt): \u0bc5>\u0bbe = VOWEL SIGN CANDRA E>VOWEL SIGN AA
\ue046>\u0bc6; # VOWEL SIGN E
\ue047>\u0bc7; # VOWEL SIGN EE
\ue048>\u0bc8; # VOWEL SIGN AI
\ue049>\u0bbe; # REMAP (indicExceptions.txt): \u0bc9>\u0bbe = VOWEL SIGN CANDRA O>VOWEL SIGN AA
\ue04a>\u0bca; # VOWEL SIGN O
\ue04b>\u0bcb; # VOWEL SIGN OO
\ue04c>\u0bcc; # VOWEL SIGN AU
\ue04d>\u0bcd; # SIGN VIRAMA
\ue050>\u0b93\u0bae\u0bcd; # REMAP (indicExceptions.txt): \u0bd0>\u0b93\u0bae\u0bcd = OM>LETTER OO.LETTER MA.SIGN VIRAMA
# \ue055>; # UNMAPPED InterIndic-Tamil: LENGTH MARK
\ue056>\u0bc8; # REMAP (indicExceptions.txt): \u0bd6>\u0bc8 = AI LENGTH MARK>VOWEL SIGN AI
\ue057>\u0bd7; # AU LENGTH MARK
\ue058>\u0b95; # FALLBACK
\ue059>\u0b95; # REMAP (indicExceptions.txt): \u0bd9>\u0b95 = LETTER KHHA>LETTER KA
\ue05a>\u0b95; # REMAP (indicExceptions.txt): \u0bda>\u0b95 = LETTER GHHA>LETTER KA
\ue05b>\u0b9c; # REMAP (indicExceptions.txt): \u0bdb>\u0b9c = LETTER ZA>LETTER JA
\ue05c>\u0ba4; # FALLBACK
\ue05d>\u0b9f; # REMAP (indicExceptions.txt): \u0bdd>\u0b9f = LETTER RHA>LETTER TTA
\ue05e>\u0baa; # REMAP (indicExceptions.txt): \u0bde>\u0baa = LETTER FA>LETTER PA
\ue05f>\u0baf; # REMAP (indicExceptions.txt): \u0bdf>\u0baf = LETTER YYA>LETTER YA
\ue060>\u0bb0\u0bbf; # REMAP (indicExceptions.txt): \u0be0>\u0bb0\u0bbf = LETTER VOCALIC RR>LETTER RA.VOWEL SIGN I
\ue061>\u0b88; # REMAP (indicExceptions.txt): \u0be1>\u0b88 = LETTER VOCALIC LL>LETTER II
# \ue062>; # UNMAPPED InterIndic-Tamil: VOWEL SIGN VOCALIC L
# \ue063>; # UNMAPPED InterIndic-Tamil: VOWEL SIGN VOCALIC LL
# \ue066>; # UNMAPPED InterIndic-Tamil: DIGIT ZERO
\ue062>\u0bbf;# FALLBACK VOWEL SIGN VOCALIC L
\ue063>\u0bc0;# FALLBACK VOWEL SIGN VOCALIC LL
\ue066>\u0030; # FALLBACK DIGIT ZERO
\ue067>\u0be7; # DIGIT ONE
\ue068>\u0be8; # DIGIT TWO
\ue069>\u0be9; # DIGIT THREE
@ -108,10 +120,5 @@
\ue06e>\u0bee; # DIGIT EIGHT
\ue06f>\u0bef; # DIGIT NINE
# \ue080>; # UNMAPPED InterIndic-Tamil: ISSHAR
\ue00e>\u0b8e; # LETTER E
\ue012>\u0b92; # LETTER O
\ue031>\u0bb1; # LETTER RRA
\ue046>\u0bc6; # VOWEL SIGN E
\ue04a>\u0bca; # VOWEL SIGN O
# :: NFC (NFD) ;
# eof

15
icu4j/src/com/ibm/icu/impl/data/Transliterator_InterIndic_Telugu.txt
View File

@ -27,8 +27,12 @@
\ue00a>\u0c0a; # LETTER UU
\ue00b>\u0c0b; # LETTER VOCALIC R
\ue00c>\u0c0c; # LETTER VOCALIC L
\ue00d>\u0c0E; # FALLBACK MAPPING
\ue00e>\u0c0E; # LETTER E
\ue00f>\u0c0f; # LETTER EE
\ue010>\u0c10; # LETTER AI
\ue011>\u0c12; # FALBACK MAPPING
\ue012>\u0c12; # LETTER O
\ue013>\u0c13; # LETTER OO
\ue014>\u0c14; # LETTER AU
\ue015>\u0c15; # LETTER KA
@ -67,8 +71,10 @@
\ue037>\u0c37; # LETTER SSA
\ue038>\u0c38; # LETTER SA
\ue039>\u0c39; # LETTER HA
# \ue03c>; # UNMAPPED InterIndic-Telugu: SIGN NUKTA
# \ue03d>; # UNMAPPED InterIndic-Telugu: SIGN AVAGRAHA
\ue03c>; # FALLBACK BLOW AWAY NUKTA
\ue03d>; # FALLBACK BLOW AWAY AVAGRAHA
\ue03e>\u0c3e; # VOWEL SIGN AA
\ue03f>\u0c3f; # VOWEL SIGN I
\ue040>\u0c40; # VOWEL SIGN II
@ -87,9 +93,11 @@
\ue055>\u0c55; # LENGTH MARK
\ue056>\u0c56; # AI LENGTH MARK
\ue057>\u0c4c; # REMAP (indicExceptions.txt): \u0c57>\u0c4c = AU LENGTH MARK>VOWEL SIGN AU
\ue058>\u0c15; # REMAP
\ue059>\u0c16; # REMAP (indicExceptions.txt): \u0c59>\u0c16 = LETTER KHHA>LETTER KHA
\ue05a>\u0c17; # REMAP (indicExceptions.txt): \u0c5a>\u0c17 = LETTER GHHA>LETTER GA
\ue05b>\u0c1c; # REMAP (indicExceptions.txt): \u0c5b>\u0c1c = LETTER ZA>LETTER JA
\ue05c>\u0c22; # REMAP
\ue05d>\u0c22; # REMAP (indicExceptions.txt): \u0c5d>\u0c22 = LETTER RHA>LETTER DDHA
\ue05e>\u0c2b; # REMAP (indicExceptions.txt): \u0c5e>\u0c2b = LETTER FA>LETTER PHA
\ue05f>\u0c2f; # REMAP (indicExceptions.txt): \u0c5f>\u0c2f = LETTER YYA>LETTER YA
@ -108,8 +116,7 @@
\ue06e>\u0c6e; # DIGIT EIGHT
\ue06f>\u0c6f; # DIGIT NINE
# \ue080>; # UNMAPPED InterIndic-Telugu: ISSHAR
\ue00e>\u0c0e; # LETTER E
\ue012>\u0c12; # LETTER O
\ue031>\u0c31; # LETTER RRA
\ue046>\u0c46; # VOWEL SIGN E
\ue04a>\u0c4a; # VOWEL SIGN O

49
icu4j/src/com/ibm/icu/impl/data/Transliterator_Latin_InterIndic.txt
View File

@ -87,7 +87,7 @@
$virama=\ue04d;
# \u094e Reserved
# \u094f Reserved
#\u0950>\ue050; # OM
$om = \ue050; # OM
# \u0951>; # UNMAPPED STRESS SIGN UDATTA
# \u0952>; # UNMAPPED STRESS SIGN ANUDATTA
# \u0953>; # UNMAPPED GRAVE ACCENT
@ -151,19 +151,19 @@
'-'h\u0323>$visarga;
mm>$anusvara;
x>$visarga;
aa>$waa;
# convert to independent forms at start of word or syllable:
# e.g. keai -> {ka}{e}{wai}; k'ai -> {ka}{wai}; (ai) -> ({wai})
a\u0304>$waa;
ai>$wai;
au>$wau;
ii>$wii;
i\u0304>$wii;
i>$wi;
uu>$wuu;
#uu>$wuu;
u\u0304>$wuu;
u>$wu;
rrh>$wrr;
#rrh>$wrr;
r\u0325\u0304>$wrr;
rh>$wr;
#rh>$wr;
r\u0325>$wr;
l\u0325\u0304>$wll;
lh>$wl;l\u0325>$wl;
@ -173,6 +173,7 @@
e\u0306>$wce;
o\u0306>$wco;
e>$wse;
''om>$om;
o>$wso;
n}na > $na|$virama;
n\u0307}[kg] > $anusvara;
@ -184,6 +185,15 @@
m}[pbm] > $anusvara;
n} [yrlvsh] > $anusvara;
'-'m\u0307 > $anusvara;
#urdu compatibility
q>$uka|$virama;
k\u0331h\u0331>$ukha |$virama;
g\u0307> $ugha | $virama;
z > $ujha |$virama;
f > $ufa|$virama;
# dev
y\u0307>$uya|$virama;
l\u0331>$ela|$virama;
n\u0331>$ena|$virama;
@ -198,7 +208,6 @@
d\u0323>$dda|$virama;
kh>$kha|$virama;
k>$ka|$virama;
q>$ka|$virama;
gh>$gha|$virama;
g>$ga|$virama;
ch>$cha|$virama;
@ -219,36 +228,37 @@
b>$ba|$virama;
m>$ma|$virama;
y>$ya|$virama;
r\u0331>$rra|$virama;
r>$ra|$virama;
l\u0323a>$lla;
l>$la|$virama;
v>$va|$virama;
f>$va|$virama;
w>$va|$virama;
sh>$sha|$virama;
ss>$ssa|$virama;
s\u0323>$ssa|$virama;
s\u0301>$sha|$virama;
s>$sa|$virama;
z>$sa|$virama;
h>$ha|$virama;
'.'>$danda;
$danda'.'>$doubleDanda;
$depVowelAbove{'~'>$anusvara;
$depVowelBelow{'~'>$chandrabindu;
$virama aa>$aa;
# 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;
#$virama uu>$uu;
$virama u\u0304>$uu;
$virama u>$u;
$virama rrh>$rrh;
#$virama rrh>$rrh;
$virama r\u0325\u0304>$rrh;
$virama rh>$rh;
#$virama rh>$rh;
$virama r\u0325a>$rh;
$virama r\u0325>$rh;
$virama l\u0325\u0304>$llh;
@ -261,22 +271,23 @@
$virama o\u0306>$co;
$virama e>$se;
$virama o>$so;
$virama''aa>$waa;
# 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;
#$virama''ii>$wii;
$virama''i\u0304>$wii;
$virama''i>$wi;
$virama''uu>$wuu;
#$virama''uu>$wuu;
$virama''u\u0304>$wuu;
$virama''u>$wu;
$virama''rrh>$wrr;
#$virama''rrh>$wrr;
$virama''r\u0325\u0304>$wrr;
$virama''rh>$wr;
#$virama''rh>$wr;
$virama''r\u0325>$wr;
$virama''l\u0325\u0304>$wll;
$virama''lh>$wl;
#$virama''lh>$wl;
$virama''l\u0325>$wl;
$virama''e\u0304>$we;
$virama''o\u0304>$wo;

1
icu4j/src/com/ibm/icu/impl/data/Transliterator_Tamil_InterIndic.txt
View File

@ -16,6 +16,7 @@
#--------------------------------------------------------------------
# Tamil-InterIndic
#:: NFD (NFC) ;
\u0bc6\u0bd7>\ue04c;
\u0b82>\ue002; # SIGN ANUSVARA
\u0b83>\ue003; # SIGN VISARGA
\u0b85>\ue005; # LETTER A

2
icu4j/src/com/ibm/icu/impl/data/Transliterator_index.txt
View File

@ -150,7 +150,7 @@ Gurmukhi-Tamil:alias:NFD;Gurmukhi-InterIndic;InterIndic-Tamil;NFC
Gurmukhi-Telugu:alias:NFD;Gurmukhi-InterIndic;InterIndic-Telugu;NFC
Gurmukhi-Kannada:alias:NFD;Gurmukhi-InterIndic;InterIndic-Kannada;NFC
Gurmukhi-Malayalam:alias:NFD;Gurmukhi-InterIndic;InterIndic-Malayalam;NFC
Gujarati-Devanagari:alias:Gujarati-InterIndic;InterIndic-Devanagari;NFC
Gujarati-Devanagari:alias:NFD;Gujarati-InterIndic;InterIndic-Devanagari;NFC
Gujarati-Bengali:alias:NFD;Gujarati-InterIndic;InterIndic-Bengali;NFC
Gujarati-Gurmukhi:alias:NFD;Gujarati-InterIndic;InterIndic-Gurmukhi;NFC
Gujarati-Oriya:alias:NFD;Gujarati-InterIndic;InterIndic-Oriya;NFC

116
icu4j/src/com/ibm/test/translit/RoundTripTest.java
View File

@ -97,18 +97,114 @@ public class RoundTripTest extends TestFmwk {
.test(null, "[\u0400-\u045F]", null, this, new Legal());
}
public void TestDevanagari() throws IOException, ParseException {
//----------------------------------
// Inter-Indic Tests
//----------------------------------
public void TestDevanagariLatin() throws IOException, ParseException {
new Test("Latin-DEVANAGARI",
TestUtility.LATIN_SCRIPT, TestUtility.DEVANAGARI_SCRIPT)
.test(null, "[:Devanagari:]", null, this, new Legal());
}
public void TestDevanagariTamil() throws IOException, ParseException {
public void TestDevanagariBengali() throws IOException, ParseException {
new Test("BENGALI-DEVANAGARI",
TestUtility.BENGALI_SCRIPT, TestUtility.DEVANAGARI_SCRIPT)
.test("[:BENGALI:]", "[:Devanagari:]",
"[\u0950\u0935\u0912\u0933\u090e\u090D\u0911\u093d\u0929\u0934[\u0958-\u095f]\u09F0\u09F1]", /*roundtrip exclusions*/
this, new Legal());
new Test("DEVANAGARI-BENGALI",
TestUtility.DEVANAGARI_SCRIPT, TestUtility.BENGALI_SCRIPT )
.test( "[:Devanagari:]", "[:BENGALI:]",
"[\u0950\u0935\u0912\u0933\u090e\u090D\u0911\u093d\u0929\u0934[\u0958-\u095f]\u09F0\u09F1]", /*roundtrip exclusions*/
this, new Legal());
}
public void TestDevanagariGurmukhi() throws IOException, ParseException {
new Test("GURMUKHI-DEVANAGARI",
TestUtility.GURMUKHI_SCRIPT, TestUtility.DEVANAGARI_SCRIPT)
.test("[:GURMUKHI:]", "[:Devanagari:]",
"[\u0950\u090D\u090e\u0912\u0911\u090b\u090c\u0934\u0960\u0961\u0937\u0a72\u0a73\u0a74\u093d]", /*roundtrip exclusions*/
this, new Legal());
new Test("DEVANAGARI-GURMUKHI",
TestUtility.DEVANAGARI_SCRIPT, TestUtility.GURMUKHI_SCRIPT )
.test( "[:Devanagari:]", "[:GURMUKHI:]",
"[\u0950\u090D\u090e\u0912\u0911\u090b\u090c\u0934\u0960\u0961\u0937\u0a72\u0a73\u0a74\u093d]", /*roundtrip exclusions*/
this, new Legal());
}
public void TestDevanagariGujarati() throws IOException, ParseException {
new Test("GUJARATI-DEVANAGARI",
TestUtility.GUJARATI_SCRIPT, TestUtility.DEVANAGARI_SCRIPT)
.test("[:GUJARATI:]", "[:Devanagari:]",
"[\u0961\u090c\u090e\u0912]", /*roundtrip exclusions*/
this, new Legal());
new Test("DEVANAGARI-GUJARATI",
TestUtility.DEVANAGARI_SCRIPT, TestUtility.GUJARATI_SCRIPT )
.test( "[:Devanagari:]", "[:GUJARATI:]",
"[\u0961\u090c\u090e\u0912]", /*roundtrip exclusions*/
this, new Legal());
}
public void TestDevanagariOriya() throws IOException, ParseException {
new Test("ORIYA-DEVANAGARI",
TestUtility.ORIYA_SCRIPT, TestUtility.DEVANAGARI_SCRIPT)
.test("[:ORIYA:]", "[:Devanagari:]",
"[\u0950\u090D\u090e\u0912\u0911\u0931\u0935]", /*roundtrip exclusions*/
this, new Legal());
new Test("DEVANAGARI-ORIYA",
TestUtility.DEVANAGARI_SCRIPT, TestUtility.ORIYA_SCRIPT )
.test( "[:Devanagari:]", "[:ORIYA:]",
"[\u0950\u090D\u090e\u0912\u0911\u0931\u0935]", /*roundtrip exclusions*/
this, new Legal());
}
public void TestDevanagariTamil() throws IOException, ParseException {
new Test("Tamil-DEVANAGARI",
TestUtility.TAMIL_SCRIPT, TestUtility.DEVANAGARI_SCRIPT)
.test("[:tamil:]", "[:Devanagari:]", null, this, new Legal());
.test("[:tamil:]", "[:Devanagari:]",
"[\u0950\u090D\u0911\u093d\u0929\u0934[\u0958-\u095f]\u090B\u090C\u0916\u0917\u0918\u091B\u091D\u0920\u0921\u0922\u0925\u0926\u0927\u092B\u092C\u092D\u0936\u0960\u0961]", /*roundtrip exclusions*/
this, new Legal());
new Test("DEVANAGARI-Tamil",
TestUtility.DEVANAGARI_SCRIPT, TestUtility.TAMIL_SCRIPT )
.test( "[:Devanagari:]", "[:tamil:]",
"[\u0950\u090D\u0911\u093d\u0929\u0934[\u0958-\u095f]]", /*roundtrip exclusions*/
this, new Legal());
}
public void TestDevanagariTelugu() throws IOException, ParseException {
new Test("Telugu-DEVANAGARI",
TestUtility.TELUGU_SCRIPT, TestUtility.DEVANAGARI_SCRIPT)
.test("[:telugu:]", "[:Devanagari:]",
"[\u0950\u090D\u0911\u093d\u0929\u0934[\u0958-\u095f]]", /*roundtrip exclusions*/
this, new Legal());
new Test("DEVANAGARI-TELUGU",
TestUtility.DEVANAGARI_SCRIPT, TestUtility.TELUGU_SCRIPT )
.test( "[:Devanagari:]", "[:TELUGU:]",
"[\u0950\u090D\u0911\u093d\u0929\u0934[\u0958-\u095f]]", /*roundtrip exclusions*/
this, new Legal());
}
public void TestDevanagariKannada() throws IOException, ParseException {
new Test("KANNADA-DEVANAGARI",
TestUtility.KANNADA_SCRIPT, TestUtility.DEVANAGARI_SCRIPT)
.test("[:KANNADA:]", "[:Devanagari:]",
"[\u0950\u090D\u0911\u093d\u0929\u0934[\u0958-\u095f]]", /*roundtrip exclusions*/
this, new Legal());
new Test("DEVANAGARI-KANNADA",
TestUtility.DEVANAGARI_SCRIPT, TestUtility.KANNADA_SCRIPT )
.test( "[:Devanagari:]", "[:KANNADA:]",
"[\u0950\u090D\u0911\u093d\u0929\u0934[\u0958-\u095f]]", /*roundtrip exclusions*/
this, new Legal());
}
public void TestDevanagariMalayalam() throws IOException, ParseException {
new Test("MALAYALAM-DEVANAGARI",
TestUtility.MALAYALAM_SCRIPT, TestUtility.DEVANAGARI_SCRIPT)
.test("[:MALAYALAM:]", "[:Devanagari:]",
"[\u0950\u090D\u0911\u093d\u0929\u0934[\u0958-\u095f]]", /*roundtrip exclusions*/
this, new Legal());
new Test("DEVANAGARI-MALAYALAM",
TestUtility.DEVANAGARI_SCRIPT, TestUtility.MALAYALAM_SCRIPT )
.test( "[:Devanagari:]", "[:MALAYALAM:]",
"[\u0950\u090D\u0911\u093d\u0929\u0934[\u0958-\u095f]]", /*roundtrip exclusions*/
this, new Legal());
}
//---------------
// End Indic
//---------------
public static class Legal {
public boolean is(String sourceString) {return true;}
}
@ -262,24 +358,26 @@ public class RoundTripTest extends TestFmwk {
return false;
}
public void test(String sourceRange, String targetRange, String roundtripExclusions, TestLog log, Legal legalSource)
public void test(String sourceRange, String targetRange, String roundtripExclusions,
TestLog log, Legal legalSource)
throws java.io.IOException, java.text.ParseException {
this.legalSource = legalSource;
if (sourceRange != null && sourceRange.length() > 0) {
this.sourceRange = new UnicodeSet(sourceRange);
}else{
this.sourceRange = new UnicodeSet("[a-zA-Z]");
}
if (targetRange != null && targetRange.length() > 0) {
this.targetRange = new UnicodeSet(targetRange);
}
if (roundtripExclusions != null && roundtripExclusions.length() > 0) {
this.roundtripExclusions = new UnicodeSet(roundtripExclusions);
}else{
this.roundtripExclusions = new UnicodeSet(); // empty
}
if (this.sourceRange == null) this.sourceRange = new UnicodeSet("[a-zA-Z]");
if (this.roundtripExclusions == null) this.roundtripExclusions = new UnicodeSet(); // empty
this.log = log;
log.logln(Utility.escape("Source: " + this.sourceRange));

4
icu4j/src/com/ibm/text/resources/Transliterator_Bengali_InterIndic.txt
View File

@ -89,8 +89,8 @@
\u09ed>\ue06d; # DIGIT SEVEN
\u09ee>\ue06e; # DIGIT EIGHT
\u09ef>\ue06f; # DIGIT NINE
# \u09f0>; # UNMAPPED Bengali-InterIndic: LETTER RA WITH MIDDLE DIAGONAL
# \u09f1>; # UNMAPPED Bengali-InterIndic: LETTER RA WITH LOWER DIAGONAL
\u09f0>\ue0f0; # UNMAPPED Bengali-InterIndic: LETTER RA WITH MIDDLE DIAGONAL
\u09f1>\ue0f1; # UNMAPPED Bengali-InterIndic: LETTER RA WITH LOWER DIAGONAL
# \u09f2>; # UNMAPPED Bengali-InterIndic: RUPEE MARK
# \u09f3>; # UNMAPPED Bengali-InterIndic: RUPEE SIGN
# \u09f4>; # UNMAPPED Bengali-InterIndic: CURRENCY NUMERATOR ONE

12
icu4j/src/com/ibm/text/resources/Transliterator_Devanagari_InterIndic.txt
View File

@ -35,11 +35,11 @@
\u090b>\ue00b; # LETTER VOCALIC R
\u090c>\ue00c; # LETTER VOCALIC L
\u090d>\ue00d; # LETTER CANDRA E (For representing English sounds)
#\u090e>\ue00e; # UNMAPPED LETTER SHORT E(For Southern Scripts)
\u090e>\ue00e; # UNMAPPED LETTER SHORT E(For Southern Scripts)
\u090f>\ue00f; # LETTER E
\u0910>\ue010; # LETTER AI
\u0911>\ue011; # LETTER CANDRA O (For representing English sounds)
#\u0912>\ue012; # UNMAPPED LETTER SHORT O (For Southern Scripts)
\u0912>\ue012; # UNMAPPED LETTER SHORT O (For Southern Scripts)
\u0913>\ue013; # LETTER O
\u0914>\ue014; # LETTER AU
\u0915>\ue015; # LETTER KA
@ -70,10 +70,10 @@
\u092e>\ue02e; # LETTER MA
\u092f>\ue02f; # LETTER YA
\u0930>\ue030; # LETTER RA
#\u0931>\ue031; # UNMAPPED LETTER RRA (Eyelash RA for Southern scripts)
\u0931>\ue031; # UNMAPPED LETTER RRA (Eyelash RA for Southern scripts)
\u0932>\ue032; # LETTER LA
\u0933>\ue033; # LETTER LLA
#\u0934>\ue034; # UNMAPPED LETTER LLLA (LLLA for Southern scripts)
\u0934>\ue034; # UNMAPPED LETTER LLLA (LLLA for Southern scripts)
\u0935>\ue035; # LETTER VA
\u0936>\ue036; # LETTER SHA
\u0937>\ue037; # LETTER SSA
@ -89,11 +89,11 @@
\u0943>\ue043; # VOWEL SIGN VOCALIC R
\u0944>\ue044; # VOWEL SIGN VOCALIC RR
\u0945>\ue045; # VOWEL SIGN CANDRA E
#\u0946>\ue046; # UNMAPPED VOWEL SIGN SHORT E
\u0946>\ue046; # UNMAPPED VOWEL SIGN SHORT E
\u0947>\ue047; # VOWEL SIGN E
\u0948>\ue048; # VOWEL SIGN AI
\u0949>\ue049; # VOWEL SIGN CANDRA O
#\u094a>\ue04a; # UNMAPPED VOWEL SIGN SHORT O
\u094a>\ue04a; # UNMAPPED VOWEL SIGN SHORT O
\u094b>\ue04b; # VOWEL SIGN O
\u094c>\ue04c; # VOWEL SIGN AU
\u094d>\ue04d; # SIGN VIRAMA

4
icu4j/src/com/ibm/text/resources/Transliterator_Gujarati_InterIndic.txt
View File

@ -26,10 +26,10 @@
\u0a89>\ue009; # LETTER U
\u0a8a>\ue00a; # LETTER UU
\u0a8b>\ue00b; # LETTER VOCALIC R
# \u0a8d>; # UNMAPPED Gujarati-InterIndic: VOWEL CANDRA E
\u0a8d>\ue00d; # UNMAPPED Gujarati-InterIndic: VOWEL CANDRA E
\u0a8f>\ue00f; # LETTER E
\u0a90>\ue010; # LETTER AI
# \u0a91>; # UNMAPPED Gujarati-InterIndic: VOWEL CANDRA O
\u0a91>\ue011; # UNMAPPED Gujarati-InterIndic: VOWEL CANDRA O
\u0a93>\ue013; # LETTER O
\u0a94>\ue014; # LETTER AU
\u0a95>\ue015; # LETTER KA

10
icu4j/src/com/ibm/text/resources/Transliterator_Gurmukhi_InterIndic.txt
View File

@ -86,10 +86,10 @@
\u0a6d>\ue06d; # DIGIT SEVEN
\u0a6e>\ue06e; # DIGIT EIGHT
\u0a6f>\ue06f; # DIGIT NINE
# \u0a70>; # UNMAPPED Gurmukhi-InterIndic: TIPPI
# \u0a71>; # UNMAPPED Gurmukhi-InterIndic: ADDAK
# \u0a72>; # UNMAPPED Gurmukhi-InterIndic: IRI
# \u0a73>; # UNMAPPED Gurmukhi-InterIndic: URA
# \u0a74>; # UNMAPPED Gurmukhi-InterIndic: EK ONKAR
\u0a70>\ue070; # TIPPI
\u0a71>\ue071; # ADDAK
\u0a72>\ue072; # IRI
\u0a73>\ue073; # URA
\u0a74>\ue074; # EK ONKAR
# :: NFC (NFD) ;
# eof

29
icu4j/src/com/ibm/text/resources/Transliterator_InterIndic_Bengali.txt
View File

@ -27,9 +27,13 @@
\ue00a>\u098a; # LETTER UU
\ue00b>\u098b; # LETTER VOCALIC R
\ue00c>\u098c; # LETTER VOCALIC L
# \ue00f>; # UNMAPPED InterIndic-Bengali: LETTER EE (\u098f = LETTER E)
\ue00d>\u098f; # FALLBACK
\ue00e>\u098f; # FALLBACK
\ue00f>\u098f; # LETTER E
\ue010>\u0990; # LETTER AI
# \ue013>; # UNMAPPED InterIndic-Bengali: LETTER OO (\u0993 = LETTER O)
\ue011>\u0993; # FALLBACK
\ue012>\u0993; # FALLBACK
\ue013>\u0993; # LETTER O
\ue014>\u0994; # LETTER AU
\ue015>\u0995; # LETTER KA
\ue016>\u0996; # LETTER KHA
@ -59,6 +63,7 @@
\ue02e>\u09ae; # LETTER MA
\ue02f>\u09af; # LETTER YA
\ue030>\u09b0; # LETTER RA
\ue031>\u09b0\u09bc; # FALLBACK to RA
\ue032>\u09b2; # LETTER LA
\ue033>\u09b2; # REMAP (indicExceptions.txt): \u09b3>\u09b2 = LETTER LLA>LETTER LA
\ue034>\u09b2; # REMAP (indicExceptions.txt): \u09b4>\u09b2 = LETTER LLLA>LETTER LA
@ -68,7 +73,7 @@
\ue038>\u09b8; # LETTER SA
\ue039>\u09b9; # LETTER HA
\ue03c>\u09bc; # SIGN NUKTA
# \ue03d>; # UNMAPPED InterIndic-Bengali: SIGN AVAGRAHA
\ue03d>; # FALLBACK BLOW AWAY AVAGRAHA
\ue03e>\u09be; # VOWEL SIGN AA
\ue03f>\u09bf; # VOWEL SIGN I
\ue040>\u09c0; # VOWEL SIGN II
@ -77,19 +82,23 @@
\ue043>\u09c3; # VOWEL SIGN VOCALIC R
\ue044>\u09c4; # VOWEL SIGN VOCALIC RR
\ue045>\u09c7; # REMAP (indicExceptions.txt): \u09c5>\u09c7 = VOWEL SIGN CANDRA E>VOWEL SIGN E
# \ue047>; # UNMAPPED InterIndic-Bengali: VOWEL SIGN EE (\u09c7 = VOWEL SIGN E)
\ue046>\u09c7; # FALLBACK
\ue047>\u09c7; # VOWEL SIGN E
\ue048>\u09c8; # VOWEL SIGN AI
\ue049>\u09cb; # REMAP (indicExceptions.txt): \u09c9>\u09cb = VOWEL SIGN CANDRA O>VOWEL SIGN O
# \ue04b>; # UNMAPPED InterIndic-Bengali: VOWEL SIGN OO (\u09cb = VOWEL SIGN O)
\ue04a>\u09cb; # FALLBACK
\ue04b>\u09cb; # VOWEL SIGN O
\ue04c>\u09cc; # VOWEL SIGN AU
\ue04d>\u09cd; # SIGN VIRAMA
# \ue050>; # UNMAPPED InterIndic-Bengali: OM
\ue050>\u0993\u0982; # InterIndic-Bengali: OM
# \ue055>; # UNMAPPED InterIndic-Bengali: LENGTH MARK
\ue056>\u09c8; # REMAP (indicExceptions.txt): \u09d6>\u09c8 = AI LENGTH MARK>VOWEL SIGN AI
\ue057>\u09d7; # AU LENGTH MARK
\ue058>\u0995; # FALLBACK
\ue059>\u0996; # REMAP (indicExceptions.txt): \u09d9>\u0996 = LETTER KHHA>LETTER KHA
\ue05a>\u0997; # REMAP (indicExceptions.txt): \u09da>\u0997 = LETTER GHHA>LETTER GA
\ue05b>\u099c; # REMAP (indicExceptions.txt): \u09db>\u099c = LETTER ZA>LETTER JA
\ue05c>\u09Dc; # FALLBACK
\ue05d>\u09dd; # LETTER RHA
\ue05e>\u09ab; # REMAP (indicExceptions.txt): \u09de>\u09ab = LETTER FA>LETTER PHA
\ue05f>\u09df; # LETTER YYA
@ -108,10 +117,8 @@
\ue06e>\u09ee; # DIGIT EIGHT
\ue06f>\u09ef; # DIGIT NINE
\ue0fa>\u09fa; # ISSHAR
\ue00f>\u098f; # LETTER E
\ue013>\u0993; # LETTER O
\ue031>\u09dc; # LETTER RRA
\ue047>\u09c7; # VOWEL SIGN E
\ue04b>\u09cb; # VOWEL SIGN O
# :: NFC (NFD) ;
# eof

51
icu4j/src/com/ibm/text/resources/Transliterator_InterIndic_Devanagari.txt
View File

@ -28,6 +28,25 @@
\ue022\ue03c > \u095d; #\ue05d LETTER RHA (pronounced RRHA)
\ue02b\ue03c > \u095e; #\ue05e LETTER FA
\ue02f\ue03c > \u095f; #\ue05f LETTER YYA
#Decomposed compatibility transliterations
\ue012\ue057>\u0914; # FALLBACK FOR TAMIL AU
\u0030 > \u0966; # FALLBACK FOR TAMIL
\ue046\ue056 > \u0948; # FALLBACK FOR TELUGU
\ue055>; # FALLBACK BLOW AWAY KANNADA AND TELUGU LENGTH MARK
\ue056>; # FALLBACK BLOW AWAY KANNADA AND TELUGU AI LENGTH MARK
\ue057>; # FALLBACK BLOW AWAY TAMIL AU LENGTH MARK
\ue070>; # FALLBACK BLOW AWAY GURMUKHI
\ue071>; # FALLBACK BLOW AWAY GURMUKHI
\ue072>; # FALLBACK BLOW AWAY GURMUKHI
\ue073>; # FALLBACK BLOW AWAY GURMUKHI
\ue074>; # FALLBACK BLOW AWAY GURMUKHI
#Decomposed compatibility transliterations
\ue032\ue03C > \u0933; # FALLBACK FOR GURMUKHI
\ue038\ue03c > \u0936; # FALLBACK FOR GURMUKHI
\ue001 > \u0901; # SIGN CANDRABINDU
\ue002 > \u0902; # SIGN ANUSVARA
\ue003 > \u0903; # SIGN VISARGA
@ -40,13 +59,13 @@
\ue00b > \u090b; # LETTER VOCALIC R
\ue00c > \u090c; # LETTER VOCALIC L
\ue00d > \u090d; # LETTER CANDRA E (For representing English sounds)
#\ue00e > \u090e; # UNMAPPED LETTER SHORT E(For Southern Scripts)
\ue00e > \u090f;
\ue00e > \u090e; # UNMAPPED LETTER SHORT E(For Southern Scripts)
#\ue00e > \u090f;
\ue00f > \u090f; # LETTER E
\ue010 > \u0910; # LETTER AI
\ue011 > \u0911; # LETTER CANDRA O (For representing English sounds)
#\ue012 > \u0912; # UNMAPPED LETTER SHORT O (For Southern Scripts)
\ue012 > \u0913;
\ue012 > \u0912; # UNMAPPED LETTER SHORT O (For Southern Scripts)
#\ue012 > \u0913;
\ue013 > \u0913; # LETTER O
\ue014 > \u0914; # LETTER AU
\ue015 > \u0915; # LETTER KA
@ -77,12 +96,12 @@
\ue02e > \u092e; # LETTER MA
\ue02f > \u092f; # LETTER YA
\ue030 > \u0930; # LETTER RA
#\ue031 > \u0931; # LETTER RRA (Eyelash RA for Southern scripts)
\ue031 > \u0930;
\ue031 > \u0931; # LETTER RRA (Eyelash RA for Southern scripts)
#\ue031 > \u0930;
\ue032 > \u0932; # LETTER LA
\ue033 > \u0933; # LETTER LLA
#\ue034 > \u0934; # LETTER LLLA (LLLA for Southern scripts)
\ue034 > \u0933;
\ue034 > \u0934; # LETTER LLLA (LLLA for Southern scripts)
#\ue034 > \u0933;
\ue035 > \u0935; # LETTER VA
\ue036 > \u0936; # LETTER SHA
\ue037 > \u0937; # LETTER SSA
@ -98,13 +117,13 @@
\ue043 > \u0943; # VOWEL SIGN VOCALIC R
\ue044 > \u0944; # VOWEL SIGN VOCALIC RR
\ue045 > \u0945; # VOWEL SIGN CANDRA E
#\ue046 > \u0946; # UNMAPPED VOWEL SIGN SHORT E
\ue046 > \u0947;
\ue046 > \u0946; # UNMAPPED VOWEL SIGN SHORT E
#\ue046 > \u0947;
\ue047 > \u0947; # VOWEL SIGN E
\ue048 > \u0948; # VOWEL SIGN AI
\ue049 > \u0949; # VOWEL SIGN CANDRA O
#\ue04a > \u094a; # UNMAPPED VOWEL SIGN SHORT O
\ue04a > \u094b;
\ue04a > \u094a; # UNMAPPED VOWEL SIGN SHORT O
#\ue04a > \u094b;
\ue04b > \u094b; # VOWEL SIGN O
\ue04c > \u094c; # VOWEL SIGN AU
\ue04d > \u094d; # SIGN VIRAMA
@ -125,8 +144,8 @@
\ue061 > \u0961; # LETTER VOCALIC LL
\ue062 > \u0962; # VOWEL SIGN VOCALIC L
\ue063 > \u0963; # VOWEL SIGN VOCALIC LL
# > ; \u0964 # UNMAPPED Devanagari-InterIndic: DANDA
# > ; \u0965 # UNMAPPED Devanagari-InterIndic: DOUBLE DANDA
\ue064 > \u0964; # UNMAPPED Devanagari-InterIndic: DANDA
\ue065 > \u0965; # UNMAPPED Devanagari-InterIndic: DOUBLE DANDA
\ue066 > \u0966; # DIGIT ZERO
\ue067 > \u0967; # DIGIT ONE
\ue068 > \u0968; # DIGIT TWO
@ -137,6 +156,10 @@
\ue06d > \u096d; # DIGIT SEVEN
\ue06e > \u096e; # DIGIT EIGHT
\ue06f > \u096f; # DIGIT NINE
\ue0f0 > \u0930; # FALLBACK RA
\ue0f1 > \u0930; # FALLBACK RA
# \u0970 # UNMAPPED Devanagari-InterIndic: ABBREVIATION SIGN
# :: NFC;
# eof

30
icu4j/src/com/ibm/text/resources/Transliterator_InterIndic_Gujarati.txt
View File

@ -27,9 +27,13 @@
\ue00a>\u0a8a; # LETTER UU
\ue00b>\u0a8b; # LETTER VOCALIC R
\ue00c>\u0ab2\u0ac3; # REMAP (indicExceptions.txt): \u0a8c>\u0ab2\u0ac3 = LETTER VOCALIC L>LETTER LA.VOWEL SIGN VOCALIC R
# \ue00f>; # UNMAPPED InterIndic-Gujarati: LETTER EE (\u0a8f = LETTER E)
\ue00d>\u0a8d; # GUJARATI VOWEL CANDRA E
\ue00e>\u0a8f; # FALLBACK
\ue00f>\u0a8f; # InterIndic-Gujarati: LETTER EE (\u0a8f = LETTER E)
\ue010>\u0a90; # LETTER AI
# \ue013>; # UNMAPPED InterIndic-Gujarati: LETTER OO (\u0a93 = LETTER O)
\ue011>\u0a91; # FALLBACK
\ue012>\u0a93; # FALLBACK
\ue013>\u0a93; # UNMAPPED InterIndic-Gujarati: LETTER OO (\u0a93 = LETTER O)
\ue014>\u0a94; # LETTER AU
\ue015>\u0a95; # LETTER KA
\ue016>\u0a96; # LETTER KHA
@ -51,7 +55,7 @@
\ue026>\u0aa6; # LETTER DA
\ue027>\u0aa7; # LETTER DHA
\ue028>\u0aa8; # LETTER NA
\ue029>\u0aa8; # REMAP (indicExceptions.txt): \u0aa9>\u0aa8 = LETTER NNNA>LETTER NA
\ue029>\u0aa8\u0abc; # FALLBACK to NA+NUKTA
\ue02a>\u0aaa; # LETTER PA
\ue02b>\u0aab; # LETTER PHA
\ue02c>\u0aac; # LETTER BA
@ -59,9 +63,10 @@
\ue02e>\u0aae; # LETTER MA
\ue02f>\u0aaf; # LETTER YA
\ue030>\u0ab0; # LETTER RA
\ue031>\u0ab0\u0abc; # FALLBACK
\ue032>\u0ab2; # LETTER LA
\ue033>\u0ab3; # LETTER LLA
\ue034>\u0ab3; # REMAP (indicExceptions.txt): \u0ab4>\u0ab3 = LETTER LLLA>LETTER LLA
\ue034>\u0ab3\u0abc; # LETTER LLLA>LETTER LLA+NUKTA
\ue035>\u0ab5; # LETTER VA
\ue036>\u0ab6; # LETTER SHA
\ue037>\u0ab7; # LETTER SSA
@ -77,19 +82,23 @@
\ue043>\u0ac3; # VOWEL SIGN VOCALIC R
\ue044>\u0ac4; # VOWEL SIGN VOCALIC RR
\ue045>\u0ac5; # VOWEL SIGN CANDRA E
# \ue047>; # UNMAPPED InterIndic-Gujarati: VOWEL SIGN EE (\u0ac7 = VOWEL SIGN E)
\ue046>\u0ac7; # FALLBACK
\ue047>\u0ac7; # InterIndic-Gujarati: VOWEL SIGN EE (\u0ac7 = VOWEL SIGN E)
\ue048>\u0ac8; # VOWEL SIGN AI
\ue049>\u0ac9; # VOWEL SIGN CANDRA O
# \ue04b>; # UNMAPPED InterIndic-Gujarati: VOWEL SIGN OO (\u0acb = VOWEL SIGN O)
\ue04a>\u0acb; # FALLBACK
\ue04b>\u0acb; # UNMAPPED InterIndic-Gujarati: VOWEL SIGN OO (\u0acb = VOWEL SIGN O)
\ue04c>\u0acc; # VOWEL SIGN AU
\ue04d>\u0acd; # SIGN VIRAMA
\ue050>\u0ad0; # OM
# \ue055>; # UNMAPPED InterIndic-Gujarati: LENGTH MARK
#\ue055>; # UNMAPPED InterIndic-Gujarati: LENGTH MARK
\ue056>\u0ac8; # REMAP (indicExceptions.txt): \u0ad6>\u0ac8 = AI LENGTH MARK>VOWEL SIGN AI
\ue057>\u0acc; # REMAP (indicExceptions.txt): \u0ad7>\u0acc = AU LENGTH MARK>VOWEL SIGN AU
\ue058>\u0a95\u0abc; # FALLBACK
\ue059>\u0a96\u0abc; # REMAP (indicExceptions.txt): \u0ad9>\u0a96\u0abc = LETTER KHHA>LETTER KHA.SIGN NUKTA
\ue05a>\u0a97\u0abc; # REMAP (indicExceptions.txt): \u0ada>\u0a97\u0abc = LETTER GHHA>LETTER GA.SIGN NUKTA
\ue05b>\u0a9c\u0abc; # REMAP (indicExceptions.txt): \u0adb>\u0a9c\u0abc = LETTER ZA>LETTER JA.SIGN NUKTA
\ue05c>\u0aa1\u0abc; # FALLBACK
\ue05d>\u0aa2\u0abc; # REMAP (indicExceptions.txt): \u0add>\u0aa2\u0abc = LETTER RHA>LETTER DDHA.SIGN NUKTA
\ue05e>\u0aab\u0abc; # REMAP (indicExceptions.txt): \u0ade>\u0aab\u0abc = LETTER FA>LETTER PHA.SIGN NUKTA
\ue05f>\u0aaf\u0abc; # REMAP (indicExceptions.txt): \u0adf>\u0aaf\u0abc = LETTER YYA>LETTER YA.SIGN NUKTA
@ -107,11 +116,6 @@
\ue06d>\u0aed; # DIGIT SEVEN
\ue06e>\u0aee; # DIGIT EIGHT
\ue06f>\u0aef; # DIGIT NINE
# \ue080>; # UNMAPPED InterIndic-Gujarati: ISSHAR
\ue00f>\u0a8f; # LETTER E
\ue013>\u0a93; # LETTER O
# \ue083>; # UNMAPPED InterIndic-Gujarati: LETTER RRA (\u0a83 = SIGN VISARGA)
\ue047>\u0ac7; # VOWEL SIGN E
\ue04b>\u0acb; # VOWEL SIGN O
#\ue080>; # UNMAPPED InterIndic-Gujarati: ISSHAR
# :: NFC (NFD) ;
# eof

47
icu4j/src/com/ibm/text/resources/Transliterator_InterIndic_Gurmukhi.txt
View File

@ -17,8 +17,8 @@
# InterIndic-Gurmukhi
#:: NFD (NFC) ;
\ue001>\u0a02; # REMAP (indicExceptions.txt): \u0a01>\u0a02 = SIGN CANDRABINDU>SIGN BINDI
# \ue002>; # UNMAPPED InterIndic-Gurmukhi: SIGN ANUSVARA (\u0a02 = SIGN BINDI)
# \ue003>; # UNMAPPED InterIndic-Gurmukhi: SIGN VISARGA
\ue002>; # FALLBACK BLOW AWAY SIGN ANUSVARA (\u0a02 = SIGN BINDI)
\ue003>; # FALLBACK BLOW AWAY SIGN VISARGA
\ue005>\u0a05; # LETTER A
\ue006>\u0a06; # LETTER AA
\ue007>\u0a07; # LETTER I
@ -26,9 +26,13 @@
\ue009>\u0a09; # LETTER U
\ue00a>\u0a0a; # LETTER UU
\ue00b>\u0a30\u0a3f; # REMAP (indicExceptions.txt): \u0a0b>\u0a30\u0a3f = LETTER VOCALIC R>LETTER RA.VOWEL SIGN I
\ue00c>\u0a07; # REMAP (indicExceptions.txt): \u0a0c>\u0a07 = LETTER VOCALIC L>LETTER I
\ue00c>\u0a33; # FALLBACK
\ue00d>\u0a0f; # FALLBACK
\ue00e>\u0a0f; # FALLBACK
\ue00f>\u0a0f; # LETTER EE
\ue010>\u0a10; # LETTER AI
\ue011>\u0a13; # FALLBACK
\ue012>\u0a13; # FALLBACK
\ue013>\u0a13; # LETTER OO
\ue014>\u0a14; # LETTER AU
\ue015>\u0a15; # LETTER KA
@ -51,7 +55,7 @@
\ue026>\u0a26; # LETTER DA
\ue027>\u0a27; # LETTER DHA
\ue028>\u0a28; # LETTER NA
\ue029>\u0a28; # REMAP (indicExceptions.txt): \u0a29>\u0a28 = LETTER NNNA>LETTER NA
\ue029>\u0a28\u0a3c; # REMAP (indicExceptions.txt): \u0a29>\u0a28 = LETTER NNNA>LETTER NA
\ue02a>\u0a2a; # LETTER PA
\ue02b>\u0a2b; # LETTER PHA
\ue02c>\u0a2c; # LETTER BA
@ -59,6 +63,7 @@
\ue02e>\u0a2e; # LETTER MA
\ue02f>\u0a2f; # LETTER YA
\ue030>\u0a30; # LETTER RA
\ue031>\u0a30\u0a3c; # FALLBACK LETTER RA+NUKTA
\ue032>\u0a32; # LETTER LA
\ue033>\u0a33; # LETTER LLA
\ue034>\u0a33; # REMAP (indicExceptions.txt): \u0a34>\u0a33 = LETTER LLLA>LETTER LLA
@ -68,33 +73,37 @@
\ue038>\u0a38; # LETTER SA
\ue039>\u0a39; # LETTER HA
\ue03c>\u0a3c; # SIGN NUKTA
# \ue03d>; # UNMAPPED InterIndic-Gurmukhi: SIGN AVAGRAHA
\ue03d>; # FALLBACK BLOW AWAY SIGN AVAGRAHA
\ue03e>\u0a3e; # VOWEL SIGN AA
\ue03f>\u0a3f; # VOWEL SIGN I
\ue040>\u0a40; # VOWEL SIGN II
\ue041>\u0a41; # VOWEL SIGN U
\ue042>\u0a42; # VOWEL SIGN UU
# \ue043>; # UNMAPPED InterIndic-Gurmukhi: VOWEL SIGN VOCALIC R
# \ue044>; # UNMAPPED InterIndic-Gurmukhi: VOWEL SIGN VOCALIC RR
\ue043>; # FALLBACK BLOW AWAY VOWEL SIGN VOCALIC R
\ue044>; # FALLBACK BLOW AWAY VOWEL SIGN VOCALIC RR
\ue045>\u0a48; # REMAP (indicExceptions.txt): \u0a45>\u0a48 = VOWEL SIGN CANDRA E>VOWEL SIGN AI
\ue046>\u0a47; # FALLABCK
\ue047>\u0a47; # VOWEL SIGN EE
\ue048>\u0a48; # VOWEL SIGN AI
\ue049>\u0a4c; # REMAP (indicExceptions.txt): \u0a49>\u0a4c = VOWEL SIGN CANDRA O>VOWEL SIGN AU
\ue04a>\u0a4b; # FALLBACK
\ue04b>\u0a4b; # VOWEL SIGN OO
\ue04c>\u0a4c; # VOWEL SIGN AU
\ue04d>\u0a4d; # SIGN VIRAMA
# \ue050>; # UNMAPPED InterIndic-Gurmukhi: OM
# \ue055>; # UNMAPPED InterIndic-Gurmukhi: LENGTH MARK
\ue050>\u0a0f\u0a02; # FALLBACK to OO+BINDI : OM
\ue055>; # FALLBACK BLOW AWAY LENGTH MARK
\ue056>\u0a48; # REMAP (indicExceptions.txt): \u0a56>\u0a48 = AI LENGTH MARK>VOWEL SIGN AI
\ue057>\u0a4c; # REMAP (indicExceptions.txt): \u0a57>\u0a4c = AU LENGTH MARK>VOWEL SIGN AU
\ue058>\u0a15\u0a3c; # FALLBACK RA+ NUKTA
\ue059>\u0a59; # LETTER KHHA
\ue05a>\u0a5a; # LETTER GHHA
\ue05b>\u0a5b; # LETTER ZA
\ue05c>\u0a5c; # LETTER RRA
\ue05d>\u0a22\u0a3c; # REMAP (indicExceptions.txt): \u0a5d>\u0a22\u0a3c = LETTER RHA>LETTER DDHA.SIGN NUKTA
\ue05e>\u0a5e; # LETTER FA
\ue05f>\u0a2f; # REMAP (indicExceptions.txt): \u0a5f>\u0a2f = LETTER YYA>LETTER YA
\ue05f>\u0a2f\u0a3c; # REMAP (indicExceptions.txt): \u0a5f>\u0a2f = LETTER YYA>LETTER YA
\ue060>\u0a30\u0a3f; # REMAP (indicExceptions.txt): \u0a60>\u0a30\u0a3f = LETTER VOCALIC RR>LETTER RA.VOWEL SIGN I
\ue061>\u0a08\u0a3c; # REMAP (indicExceptions.txt): \u0a61>\u0a08\u0a3c = LETTER VOCALIC LL>LETTER II.SIGN NUKTA
\ue061>\u0a33\u0a3c; #
\ue062>\u0a3f\u0a3c; # REMAP (indicExceptions.txt): \u0a62>\u0a3f\u0a3c = VOWEL SIGN VOCALIC L>VOWEL SIGN I.SIGN NUKTA
\ue063>\u0a40\u0a3c; # REMAP (indicExceptions.txt): \u0a63>\u0a40\u0a3c = VOWEL SIGN VOCALIC LL>VOWEL SIGN II.SIGN NUKTA
\ue066>\u0a66; # DIGIT ZERO
@ -107,11 +116,15 @@
\ue06d>\u0a6d; # DIGIT SEVEN
\ue06e>\u0a6e; # DIGIT EIGHT
\ue06f>\u0a6f; # DIGIT NINE
# \ue080>; # UNMAPPED InterIndic-Gurmukhi: ISSHAR
# \ue081>; # UNMAPPED InterIndic-Gurmukhi: LETTER E
# \ue082>; # UNMAPPED InterIndic-Gurmukhi: LETTER O (\u0a02 = SIGN BINDI)
\ue05c>\u0a5c; # LETTER RRA
# \ue084>; # UNMAPPED InterIndic-Gurmukhi: VOWEL SIGN E
# \ue085>; # UNMAPPED InterIndic-Gurmukhi: VOWEL SIGN O (\u0a05 = LETTER A)
\ue070>\u0a70; # TIPPI
\ue071>\u0a71; # ADDAK
\ue072>\u0a72; # IRI
\ue073>\u0a73; # URA
\ue074>\u0a74; # EK ONKAR
\ue080>; # FALLBACK BLOW AWAY ISSHAR
\ue081>; # FALLBACK BLOW AWAY LETTER E
\ue082>; # FALLBACK BLOW AWAY LETTER O (\u0a02 = SIGN BINDI)
\ue084>; # FALLBACK BLOW AWAY VOWEL SIGN E
\ue085>; # FALLBACK BLOW AWAY VOWEL SIGN O (\u0a05 = LETTER A)
# :: NFC (NFD) ;
# eof

20
icu4j/src/com/ibm/text/resources/Transliterator_InterIndic_Kannada.txt
View File

@ -27,8 +27,12 @@
\ue00a>\u0c8a; # LETTER UU
\ue00b>\u0c8b; # LETTER VOCALIC R
\ue00c>\u0c8c; # LETTER VOCALIC L
\ue00d>\u0c8e; # LETTER E
\ue00e>\u0c8e; # FALLBACK
\ue00f>\u0c8f; # LETTER EE
\ue010>\u0c90; # LETTER AI
\ue011>\u0c92; # FALLBACK
\ue012>\u0c92; # LETTER O
\ue013>\u0c93; # LETTER OO
\ue014>\u0c94; # LETTER AU
\ue015>\u0c95; # LETTER KA
@ -59,6 +63,7 @@
\ue02e>\u0cae; # LETTER MA
\ue02f>\u0caf; # LETTER YA
\ue030>\u0cb0; # LETTER RA
\ue031>\u0cb1; # LETTER RRA
\ue032>\u0cb2; # LETTER LA
\ue033>\u0cb3; # LETTER LLA
\ue034>\u0cb3; # REMAP (indicExceptions.txt): \u0cb4>\u0cb3 = LETTER LLLA>LETTER LLA
@ -67,8 +72,10 @@
\ue037>\u0cb7; # LETTER SSA
\ue038>\u0cb8; # LETTER SA
\ue039>\u0cb9; # LETTER HA
# \ue03c>; # UNMAPPED InterIndic-Kannada: SIGN NUKTA
# \ue03d>; # UNMAPPED InterIndic-Kannada: SIGN AVAGRAHA
\ue03c>; # FALLBACK BLOW AWAY NUKTA
\ue03d>; # FALLBACK BLOW AWAY AVAGRAHA
\ue03e>\u0cbe; # VOWEL SIGN AA
\ue03f>\u0cbf; # VOWEL SIGN I
\ue040>\u0cc0; # VOWEL SIGN II
@ -77,9 +84,11 @@
\ue043>\u0cc3; # VOWEL SIGN VOCALIC R
\ue044>\u0cc4; # VOWEL SIGN VOCALIC RR
\ue045>\u0cc6; # REMAP (indicExceptions.txt): \u0cc5>\u0cc6 = VOWEL SIGN CANDRA E>VOWEL SIGN E
\ue046>\u0cc6; # VOWEL SIGN E
\ue047>\u0cc7; # VOWEL SIGN EE
\ue048>\u0cc8; # VOWEL SIGN AI
\ue049>\u0cca; # REMAP (indicExceptions.txt): \u0cc9>\u0cca = VOWEL SIGN CANDRA O>VOWEL SIGN O
\ue04a>\u0cca; # VOWEL SIGN O
\ue04b>\u0ccb; # VOWEL SIGN OO
\ue04c>\u0ccc; # VOWEL SIGN AU
\ue04d>\u0ccd; # SIGN VIRAMA
@ -87,9 +96,11 @@
\ue055>\u0cd5; # LENGTH MARK
\ue056>\u0cd6; # AI LENGTH MARK
\ue057>\u0ccc; # REMAP (indicExceptions.txt): \u0cd7>\u0ccc = AU LENGTH MARK>VOWEL SIGN AU
\ue058>\u0c95; # FALLBACK
\ue059>\u0c96; # REMAP (indicExceptions.txt): \u0cd9>\u0c96 = LETTER KHHA>LETTER KHA
\ue05a>\u0c97; # REMAP (indicExceptions.txt): \u0cda>\u0c97 = LETTER GHHA>LETTER GA
\ue05b>\u0c9c; # REMAP (indicExceptions.txt): \u0cdb>\u0c9c = LETTER ZA>LETTER JA
\ue05c>\u0ca2; # FALLBACK
\ue05d>\u0ca2; # REMAP (indicExceptions.txt): \u0cdd>\u0ca2 = LETTER RHA>LETTER DDHA
\ue05e>\u0cde; # LETTER FA
\ue05f>\u0caf; # REMAP (indicExceptions.txt): \u0cdf>\u0caf = LETTER YYA>LETTER YA
@ -108,10 +119,5 @@
\ue06e>\u0cee; # DIGIT EIGHT
\ue06f>\u0cef; # DIGIT NINE
# \ue080>; # UNMAPPED InterIndic-Kannada: ISSHAR
\ue00e>\u0c8e; # LETTER E
\ue012>\u0c92; # LETTER O
\ue031>\u0cb1; # LETTER RRA
\ue046>\u0cc6; # VOWEL SIGN E
\ue04a>\u0cca; # VOWEL SIGN O
# :: NFC (NFD) ;
# eof

66
icu4j/src/com/ibm/text/resources/Transliterator_InterIndic_Latin.txt
View File

@ -87,7 +87,7 @@
$virama=\ue04d;
# \u094e Reserved
# \u094f Reserved
#\u0950>\ue050; # OM
$om=\ue050; # OM
# \u0951>; # UNMAPPED STRESS SIGN UDATTA
# \u0952>; # UNMAPPED STRESS SIGN ANUDATTA
# \u0953>; # UNMAPPED GRAVE ACCENT
@ -238,6 +238,9 @@
$ra$virama}$ha>r'';
$ra}$x>r;
$ra>ra;
$rra$virama}$ha>r\u0331'';
$rra}$x>r\u0331;
$rra>r\u0331a;
$la$virama}$ha>l'';
$la}$x>l;
$la>la;
@ -267,8 +270,8 @@
$ena > n\u0331a ;
$uka}$x > q ;
$uka > qa ;
$ukha}$x > k\u0323 ;
$ukha > k\u0323a ;
$ukha}$x > k\u0331h\u0331 ;
$ukha > k\u0331h\u0331a ;
$ugha}$x > g\u0307 ;
$ugha > g\u0307a ;
$ujha}$x > z ;
@ -277,8 +280,8 @@
$udha > r\u0323ha;
$uddha}$x> r\u0323 ;
$uddha > r\u0323a ;
$ufa}$x > f\u0323 ;
$ufa > f\u0323a ;
$ufa}$x > f ;
$ufa > fa ;
# dependent vowels (should never occur except following consonants)
$aa > a\u0304 ;
$ai > ai ;
@ -300,33 +303,33 @@
$se > e ;
$so > o ;
# independent vowels (when following consonants)
a}$waa > ''a\u0304 ;
$z}$waa > ''a\u0304 ;
a}$wai > ''ai ;
$z}$wai > ''ai ;
a}$wau > ''au ;
$z}$wau > ''au ;
a}$wii > ''i\u0304 ;
$z}$wii > ''i\u0304 ;
a}$wi > ''i ;
$z}$wi > ''i ;
a}$wuu > ''u\u0304 ;
$z}$wuu > ''u\u0304 ;
a}$wu > ''u ;
$z}$wu > ''u ;
$z}$wrr > ''r\u0325\u0304 ;
$z}$wr > ''r\u0325 ;
$z}$wll > ''l\u0325\u0304 ;
$z}$wl > ''l\u0325 ;
$z}$we > ''e\u0304 ;
$z}$wo > ''o\u0304 ;
a}$wa > ''a ;
$z}$wa > ''a ;
a{$waa > ''a\u0304 ;
$z{$waa > ''a\u0304 ;
a{$wai > ''ai ;
$z{$wai > ''ai ;
a{$wau > ''au ;
$z{$wau > ''au ;
a{$wii > ''i\u0304 ;
$z{$wii > ''i\u0304 ;
a{$wi > ''i ;
$z{$wi > ''i ;
a{$wuu > ''u\u0304 ;
$z{$wuu > ''u\u0304 ;
a{$wu > ''u ;
$z{$wu > ''u ;
$z{$wrr > ''r\u0325\u0304 ;
$z{$wr > ''r\u0325 ;
$z{$wll > ''l\u0325\u0304 ;
$z{$wl > ''l\u0325 ;
$z{$we > ''e\u0304 ;
$z{$wo > ''o\u0304 ;
a{$wa > ''a ;
$z{$wa > ''a ;
#extra vowels
$z}$wce > ''e\u0306 ;
$z}$wco > ''o\u0306 ;
$z}$wse > ''e ;
$z}$wso > ''o ;
$z{$wce > ''e\u0306 ;
$z{$wco > ''o\u0306 ;
$z{$wse > ''e ;
$z{$wso > ''o ;
# independent vowels (otherwise)
$waa > a\u0304 ;
$wai > ai ;
@ -347,6 +350,7 @@
$wco > o\u0306 ;
$wse > e ;
$wso > o ;
$om > ''om ;
#stress marks
$avagraha > \u0315;
$chandrabindu$anusvara>'-'\u0303;

35
icu4j/src/com/ibm/text/resources/Transliterator_InterIndic_Malayalam.txt
View File

@ -27,8 +27,12 @@
\ue00a>\u0d0a; # LETTER UU
\ue00b>\u0d0b; # LETTER VOCALIC R
\ue00c>\u0d0c; # LETTER VOCALIC L
\ue00d>\u0d0e; # FALLLBACK LETTER E
\ue00e>\u0d0e; # LETTER E
\ue00f>\u0d0f; # LETTER EE
\ue010>\u0d10; # LETTER AI
\ue011>\u0d12; # FALLBACK TO O
\ue012>\u0d12; # LETTER O
\ue013>\u0d13; # LETTER OO
\ue014>\u0d14; # LETTER AU
\ue015>\u0d15; # LETTER KA
@ -59,6 +63,7 @@
\ue02e>\u0d2e; # LETTER MA
\ue02f>\u0d2f; # LETTER YA
\ue030>\u0d30; # LETTER RA
\ue031>\u0d31; # LETTER RRA
\ue032>\u0d32; # LETTER LA
\ue033>\u0d33; # LETTER LLA
\ue034>\u0d34; # LETTER LLLA
@ -67,36 +72,42 @@
\ue037>\u0d37; # LETTER SSA
\ue038>\u0d38; # LETTER SA
\ue039>\u0d39; # LETTER HA
# \ue03c>; # UNMAPPED InterIndic-Malayalam: SIGN NUKTA
# \ue03d>; # UNMAPPED InterIndic-Malayalam: SIGN AVAGRAHA
\ue03c>; # FALLBACK BLOW AWAY NUKTA
\ue03d>; # FALLBACK BLOW AWAY AVAGRAHA
\ue03e>\u0d3e; # VOWEL SIGN AA
\ue03f>\u0d3f; # VOWEL SIGN I
\ue040>\u0d40; # VOWEL SIGN II
\ue041>\u0d41; # VOWEL SIGN U
\ue042>\u0d42; # VOWEL SIGN UU
\ue043>\u0d43; # VOWEL SIGN VOCALIC R
# \ue044>; # UNMAPPED InterIndic-Malayalam: VOWEL SIGN VOCALIC RR
\ue044>; # FALLBACK BLOW AWAY VOWEL SIGN VOCALIC RR
\ue045>\u0d3e; # REMAP (indicExceptions.txt): \u0d45>\u0d3e = VOWEL SIGN CANDRA E>VOWEL SIGN AA
\ue046>\u0d46; # VOWEL SIGN E
\ue047>\u0d47; # VOWEL SIGN EE
\ue048>\u0d48; # VOWEL SIGN AI
\ue049>\u0d4b; # REMAP (indicExceptions.txt): \u0d49>\u0d4b = VOWEL SIGN CANDRA O>VOWEL SIGN OO
\ue04a>\u0d4a; # VOWEL SIGN O
\ue04b>\u0d4b; # VOWEL SIGN OO
\ue04c>\u0d4c; # VOWEL SIGN AU
\ue04d>\u0d4d; # SIGN VIRAMA
# \ue050>; # UNMAPPED InterIndic-Malayalam: OM
# \ue055>; # UNMAPPED InterIndic-Malayalam: LENGTH MARK
\ue050>\u0d13\u0d02; # UNMAPPED InterIndic-Malayalam: OM
\ue055>; # FALLBACK BLOW AWAY LENGTH MARK
\ue056>\u0d48; # REMAP (indicExceptions.txt): \u0d56>\u0d48 = AI LENGTH MARK>VOWEL SIGN AI
\ue057>\u0d57; # AU LENGTH MARK
\ue058>\u0d15; # FALLBACK
\ue059>\u0d16; # REMAP (indicExceptions.txt): \u0d59>\u0d16 = LETTER KHHA>LETTER KHA
\ue05a>\u0d17; # REMAP (indicExceptions.txt): \u0d5a>\u0d17 = LETTER GHHA>LETTER GA
\ue05b>\u0d1c; # REMAP (indicExceptions.txt): \u0d5b>\u0d1c = LETTER ZA>LETTER JA
\ue05d>\u0d22; # REMAP (indicExceptions.txt): \u0d5d>\u0d22 = LETTER RHA>LETTER DDHA
\ue05c>\u0d21; # FALLBACK
\ue05e>\u0d2b; # REMAP (indicExceptions.txt): \u0d5e>\u0d2b = LETTER FA>LETTER PHA
\ue05f>\u0d2f; # REMAP (indicExceptions.txt): \u0d5f>\u0d2f = LETTER YYA>LETTER YA
\ue060>\u0d60; # LETTER VOCALIC RR
\ue061>\u0d61; # LETTER VOCALIC LL
# \ue062>; # UNMAPPED InterIndic-Malayalam: VOWEL SIGN VOCALIC L
# \ue063>; # UNMAPPED InterIndic-Malayalam: VOWEL SIGN VOCALIC LL
\ue062>; # FALLBACK BLOW AWAY VOWEL SIGN VOCALIC L
\ue063>; # FALLBACK BLOW AWAY VOWEL SIGN VOCALIC LL
\ue066>\u0d66; # DIGIT ZERO
\ue067>\u0d67; # DIGIT ONE
\ue068>\u0d68; # DIGIT TWO
@ -108,10 +119,10 @@
\ue06e>\u0d6e; # DIGIT EIGHT
\ue06f>\u0d6f; # DIGIT NINE
# \ue080>; # UNMAPPED InterIndic-Malayalam: ISSHAR
\ue00e>\u0d0e; # LETTER E
\ue012>\u0d12; # LETTER O
\ue031>\u0d31; # LETTER RRA
\ue046>\u0d46; # VOWEL SIGN E
\ue04a>\u0d4a; # VOWEL SIGN O
# :: NFC (NFD) ;
# eof

32
icu4j/src/com/ibm/text/resources/Transliterator_InterIndic_Oriya.txt
View File

@ -27,9 +27,13 @@
\ue00a>\u0b0a; # LETTER UU
\ue00b>\u0b0b; # LETTER VOCALIC R
\ue00c>\u0b0c; # LETTER VOCALIC L
# \ue00f>; # UNMAPPED InterIndic-Oriya: LETTER EE (\u0b0f = LETTER E)
\ue00d>\u0b0f; # FALLBACK
\ue00e>\u0b0f; # FALLBACK
\ue00f>\u0b0f; # LETTER E
\ue010>\u0b10; # LETTER AI
# \ue013>; # UNMAPPED InterIndic-Oriya: LETTER OO (\u0b13 = LETTER O)
\ue011>\u0b13; # FALLBACK
\ue012>\u0b13; # FALLBACK
\ue013>\u0b13; # UNMAPPED InterIndic-Oriya: LETTER OO (\u0b13 = LETTER O)
\ue014>\u0b14; # LETTER AU
\ue015>\u0b15; # LETTER KA
\ue016>\u0b16; # LETTER KHA
@ -51,7 +55,7 @@
\ue026>\u0b26; # LETTER DA
\ue027>\u0b27; # LETTER DHA
\ue028>\u0b28; # LETTER NA
\ue029>\u0b28; # REMAP (indicExceptions.txt): \u0b29>\u0b28 = LETTER NNNA>LETTER NA
\ue029>\u0b28\u0b3c; # REMAP (indicExceptions.txt): \u0b29>\u0b28 = LETTER NNNA>LETTER NA
\ue02a>\u0b2a; # LETTER PA
\ue02b>\u0b2b; # LETTER PHA
\ue02c>\u0b2c; # LETTER BA
@ -59,9 +63,10 @@
\ue02e>\u0b2e; # LETTER MA
\ue02f>\u0b2f; # LETTER YA
\ue030>\u0b30; # LETTER RA
\ue031>\u0b5c; # LETTER RRA
\ue032>\u0b32; # LETTER LA
\ue033>\u0b33; # LETTER LLA
\ue034>\u0b33; # REMAP (indicExceptions.txt): \u0b34>\u0b33 = LETTER LLLA>LETTER LLA
\ue034>\u0b33\u0b3c; # REMAP (indicExceptions.txt): \u0b34>\u0b33 = LETTER LLLA>LETTER LLA
\ue035>\u0b2c; # REMAP (indicExceptions.txt): \u0b35>\u0b2c = LETTER VA>LETTER BA
\ue036>\u0b36; # LETTER SHA
\ue037>\u0b37; # LETTER SSA
@ -76,11 +81,13 @@
\ue042>\u0b42; # VOWEL SIGN UU
\ue043>\u0b43; # VOWEL SIGN VOCALIC R
\ue044>\u0b43\u0b3c; # REMAP (indicExceptions.txt): \u0b44>\u0b43\u0b3c = VOWEL SIGN VOCALIC RR>VOWEL SIGN VOCALIC R.SIGN NUKTA
\ue045>\u0b47; # REMAP (indicExceptions.txt): \u0b45>\u0b47 = VOWEL SIGN CANDRA E>VOWEL SIGN E
# \ue047>; # UNMAPPED InterIndic-Oriya: VOWEL SIGN EE (\u0b47 = VOWEL SIGN E)
\ue045>\u0b47; # FALLBACK
\ue046>\u0b47; # FALLBACK
\ue047>\u0b47; # VOWEL SIGN E
\ue048>\u0b48; # VOWEL SIGN AI
\ue049>\u0b4b; # REMAP (indicExceptions.txt): \u0b49>\u0b4b = VOWEL SIGN CANDRA O>VOWEL SIGN O
# \ue04b>; # UNMAPPED InterIndic-Oriya: VOWEL SIGN OO (\u0b4b = VOWEL SIGN O)
\ue049>\u0b4b; # FALLBACK
\ue04a>\u0b4b; # FALLBACK
\ue04b>\u0b4b; # VOWEL SIGN E
\ue04c>\u0b4c; # VOWEL SIGN AU
\ue04d>\u0b4d; # SIGN VIRAMA
\ue050>\u0b13\u0b01; # REMAP (indicExceptions.txt): \u0b50>\u0b13\u0b01 = OM>LETTER O.SIGN CANDRABINDU
@ -88,8 +95,10 @@
\ue056>\u0b56; # AI LENGTH MARK
\ue057>\u0b57; # AU LENGTH MARK
\ue059>\u0b16\u0b3c; # REMAP (indicExceptions.txt): \u0b59>\u0b16\u0b3c = LETTER KHHA>LETTER KHA.SIGN NUKTA
\ue058>\u0b15\u0b3c; # FALLBACK
\ue05a>\u0b17\u0b3c; # REMAP (indicExceptions.txt): \u0b5a>\u0b17\u0b3c = LETTER GHHA>LETTER GA.SIGN NUKTA
\ue05b>\u0b1c\u0b3c; # REMAP (indicExceptions.txt): \u0b5b>\u0b1c\u0b3c = LETTER ZA>LETTER JA.SIGN NUKTA
\ue05c>\u0b21\u0b3c; # FALLBACK
\ue05d>\u0b5d; # LETTER RHA
\ue05e>\u0b2b\u0b3c; # REMAP (indicExceptions.txt): \u0b5e>\u0b2b\u0b3c = LETTER FA>LETTER PHA.SIGN NUKTA
\ue05f>\u0b5f; # LETTER YYA
@ -108,10 +117,7 @@
\ue06e>\u0b6e; # DIGIT EIGHT
\ue06f>\u0b6f; # DIGIT NINE
\ue070>\u0b70; # ISSHAR
\ue00e>\u0b0f; # LETTER E
\ue013>\u0b13; # LETTER O
\ue031>\u0b5c; # LETTER RRA
\ue047>\u0b47; # VOWEL SIGN E
\ue04b>\u0b4b; # VOWEL SIGN O
# :: NFC (NFD) ;
# eof

29
icu4j/src/com/ibm/text/resources/Transliterator_InterIndic_Tamil.txt
View File

@ -16,7 +16,7 @@
#--------------------------------------------------------------------
# InterIndic-Tamil
#:: NFD (NFC) ;
# \ue001>; # UNMAPPED InterIndic-Tamil: SIGN CANDRABINDU
\ue001>\u0b82; # FALLBACK SIGN CANDRABINDU
\ue002>\u0b82; # SIGN ANUSVARA
\ue003>\u0b83; # SIGN VISARGA
\ue005>\u0b85; # LETTER A
@ -27,8 +27,12 @@
\ue00a>\u0b8a; # LETTER UU
\ue00b>\u0bb0\u0bbf; # REMAP (indicExceptions.txt): \u0b8b>\u0bb0\u0bbf = LETTER VOCALIC R>LETTER RA.VOWEL SIGN I
\ue00c>\u0b87; # REMAP (indicExceptions.txt): \u0b8c>\u0b87 = LETTER VOCALIC L>LETTER I
\ue00d>\u0b8f; # FALLBACK
\ue00e>\u0b8e; # LETTER E
\ue00f>\u0b8f; # LETTER EE
\ue010>\u0b90; # LETTER AI
\ue011>\u0b92; # FALLBACK
\ue012>\u0b92; # LETTER O
\ue013>\u0b93; # LETTER OO
\ue014>\u0b94; # LETTER AU
\ue015>\u0b95; # LETTER KA
@ -59,6 +63,7 @@
\ue02e>\u0bae; # LETTER MA
\ue02f>\u0baf; # LETTER YA
\ue030>\u0bb0; # LETTER RA
\ue031>\u0bb1; # LETTER RRA
\ue032>\u0bb2; # LETTER LA
\ue033>\u0bb3; # LETTER LLA
\ue034>\u0bb4; # LETTER LLLA
@ -67,8 +72,10 @@
\ue037>\u0bb7; # LETTER SSA
\ue038>\u0bb8; # LETTER SA
\ue039>\u0bb9; # LETTER HA
# \ue03c>; # UNMAPPED InterIndic-Tamil: SIGN NUKTA
# \ue03d>; # UNMAPPED InterIndic-Tamil: SIGN AVAGRAHA
\ue03c>; # FALLBACK BLOW AWAY NUKTA
\ue03d>; # FALLBACK BLOW AWAY AVAGRAHA
\ue03e>\u0bbe; # VOWEL SIGN AA
\ue03f>\u0bbf; # VOWEL SIGN I
\ue040>\u0bc0; # VOWEL SIGN II
@ -77,27 +84,32 @@
\ue043>\u0bcd\u0bb0\u0bbf; # REMAP (indicExceptions.txt): \u0bc3>\u0bcd\u0bb0\u0bbf = VOWEL SIGN VOCALIC R>SIGN VIRAMA.LETTER RA.VOWEL SIGN I
\ue044>\u0bcd\u0bb0\u0bbf; # REMAP (indicExceptions.txt): \u0bc4>\u0bcd\u0bb0\u0bbf = VOWEL SIGN VOCALIC RR>SIGN VIRAMA.LETTER RA.VOWEL SIGN I
\ue045>\u0bbe; # REMAP (indicExceptions.txt): \u0bc5>\u0bbe = VOWEL SIGN CANDRA E>VOWEL SIGN AA
\ue046>\u0bc6; # VOWEL SIGN E
\ue047>\u0bc7; # VOWEL SIGN EE
\ue048>\u0bc8; # VOWEL SIGN AI
\ue049>\u0bbe; # REMAP (indicExceptions.txt): \u0bc9>\u0bbe = VOWEL SIGN CANDRA O>VOWEL SIGN AA
\ue04a>\u0bca; # VOWEL SIGN O
\ue04b>\u0bcb; # VOWEL SIGN OO
\ue04c>\u0bcc; # VOWEL SIGN AU
\ue04d>\u0bcd; # SIGN VIRAMA
\ue050>\u0b93\u0bae\u0bcd; # REMAP (indicExceptions.txt): \u0bd0>\u0b93\u0bae\u0bcd = OM>LETTER OO.LETTER MA.SIGN VIRAMA
# \ue055>; # UNMAPPED InterIndic-Tamil: LENGTH MARK
\ue056>\u0bc8; # REMAP (indicExceptions.txt): \u0bd6>\u0bc8 = AI LENGTH MARK>VOWEL SIGN AI
\ue057>\u0bd7; # AU LENGTH MARK
\ue058>\u0b95; # FALLBACK
\ue059>\u0b95; # REMAP (indicExceptions.txt): \u0bd9>\u0b95 = LETTER KHHA>LETTER KA
\ue05a>\u0b95; # REMAP (indicExceptions.txt): \u0bda>\u0b95 = LETTER GHHA>LETTER KA
\ue05b>\u0b9c; # REMAP (indicExceptions.txt): \u0bdb>\u0b9c = LETTER ZA>LETTER JA
\ue05c>\u0ba4; # FALLBACK
\ue05d>\u0b9f; # REMAP (indicExceptions.txt): \u0bdd>\u0b9f = LETTER RHA>LETTER TTA
\ue05e>\u0baa; # REMAP (indicExceptions.txt): \u0bde>\u0baa = LETTER FA>LETTER PA
\ue05f>\u0baf; # REMAP (indicExceptions.txt): \u0bdf>\u0baf = LETTER YYA>LETTER YA
\ue060>\u0bb0\u0bbf; # REMAP (indicExceptions.txt): \u0be0>\u0bb0\u0bbf = LETTER VOCALIC RR>LETTER RA.VOWEL SIGN I
\ue061>\u0b88; # REMAP (indicExceptions.txt): \u0be1>\u0b88 = LETTER VOCALIC LL>LETTER II
# \ue062>; # UNMAPPED InterIndic-Tamil: VOWEL SIGN VOCALIC L
# \ue063>; # UNMAPPED InterIndic-Tamil: VOWEL SIGN VOCALIC LL
# \ue066>; # UNMAPPED InterIndic-Tamil: DIGIT ZERO
\ue062>\u0bbf;# FALLBACK VOWEL SIGN VOCALIC L
\ue063>\u0bc0;# FALLBACK VOWEL SIGN VOCALIC LL
\ue066>\u0030; # FALLBACK DIGIT ZERO
\ue067>\u0be7; # DIGIT ONE
\ue068>\u0be8; # DIGIT TWO
\ue069>\u0be9; # DIGIT THREE
@ -108,10 +120,5 @@
\ue06e>\u0bee; # DIGIT EIGHT
\ue06f>\u0bef; # DIGIT NINE
# \ue080>; # UNMAPPED InterIndic-Tamil: ISSHAR
\ue00e>\u0b8e; # LETTER E
\ue012>\u0b92; # LETTER O
\ue031>\u0bb1; # LETTER RRA
\ue046>\u0bc6; # VOWEL SIGN E
\ue04a>\u0bca; # VOWEL SIGN O
# :: NFC (NFD) ;
# eof

15
icu4j/src/com/ibm/text/resources/Transliterator_InterIndic_Telugu.txt
View File

@ -27,8 +27,12 @@
\ue00a>\u0c0a; # LETTER UU
\ue00b>\u0c0b; # LETTER VOCALIC R
\ue00c>\u0c0c; # LETTER VOCALIC L
\ue00d>\u0c0E; # FALLBACK MAPPING
\ue00e>\u0c0E; # LETTER E
\ue00f>\u0c0f; # LETTER EE
\ue010>\u0c10; # LETTER AI
\ue011>\u0c12; # FALBACK MAPPING
\ue012>\u0c12; # LETTER O
\ue013>\u0c13; # LETTER OO
\ue014>\u0c14; # LETTER AU
\ue015>\u0c15; # LETTER KA
@ -67,8 +71,10 @@
\ue037>\u0c37; # LETTER SSA
\ue038>\u0c38; # LETTER SA
\ue039>\u0c39; # LETTER HA
# \ue03c>; # UNMAPPED InterIndic-Telugu: SIGN NUKTA
# \ue03d>; # UNMAPPED InterIndic-Telugu: SIGN AVAGRAHA
\ue03c>; # FALLBACK BLOW AWAY NUKTA
\ue03d>; # FALLBACK BLOW AWAY AVAGRAHA
\ue03e>\u0c3e; # VOWEL SIGN AA
\ue03f>\u0c3f; # VOWEL SIGN I
\ue040>\u0c40; # VOWEL SIGN II
@ -87,9 +93,11 @@
\ue055>\u0c55; # LENGTH MARK
\ue056>\u0c56; # AI LENGTH MARK
\ue057>\u0c4c; # REMAP (indicExceptions.txt): \u0c57>\u0c4c = AU LENGTH MARK>VOWEL SIGN AU
\ue058>\u0c15; # REMAP
\ue059>\u0c16; # REMAP (indicExceptions.txt): \u0c59>\u0c16 = LETTER KHHA>LETTER KHA
\ue05a>\u0c17; # REMAP (indicExceptions.txt): \u0c5a>\u0c17 = LETTER GHHA>LETTER GA
\ue05b>\u0c1c; # REMAP (indicExceptions.txt): \u0c5b>\u0c1c = LETTER ZA>LETTER JA
\ue05c>\u0c22; # REMAP
\ue05d>\u0c22; # REMAP (indicExceptions.txt): \u0c5d>\u0c22 = LETTER RHA>LETTER DDHA
\ue05e>\u0c2b; # REMAP (indicExceptions.txt): \u0c5e>\u0c2b = LETTER FA>LETTER PHA
\ue05f>\u0c2f; # REMAP (indicExceptions.txt): \u0c5f>\u0c2f = LETTER YYA>LETTER YA
@ -108,8 +116,7 @@
\ue06e>\u0c6e; # DIGIT EIGHT
\ue06f>\u0c6f; # DIGIT NINE
# \ue080>; # UNMAPPED InterIndic-Telugu: ISSHAR
\ue00e>\u0c0e; # LETTER E
\ue012>\u0c12; # LETTER O
\ue031>\u0c31; # LETTER RRA
\ue046>\u0c46; # VOWEL SIGN E
\ue04a>\u0c4a; # VOWEL SIGN O

49
icu4j/src/com/ibm/text/resources/Transliterator_Latin_InterIndic.txt
View File

@ -87,7 +87,7 @@
$virama=\ue04d;
# \u094e Reserved
# \u094f Reserved
#\u0950>\ue050; # OM
$om = \ue050; # OM
# \u0951>; # UNMAPPED STRESS SIGN UDATTA
# \u0952>; # UNMAPPED STRESS SIGN ANUDATTA
# \u0953>; # UNMAPPED GRAVE ACCENT
@ -151,19 +151,19 @@
'-'h\u0323>$visarga;
mm>$anusvara;
x>$visarga;
aa>$waa;
# convert to independent forms at start of word or syllable:
# e.g. keai -> {ka}{e}{wai}; k'ai -> {ka}{wai}; (ai) -> ({wai})
a\u0304>$waa;
ai>$wai;
au>$wau;
ii>$wii;
i\u0304>$wii;
i>$wi;
uu>$wuu;
#uu>$wuu;
u\u0304>$wuu;
u>$wu;
rrh>$wrr;
#rrh>$wrr;
r\u0325\u0304>$wrr;
rh>$wr;
#rh>$wr;
r\u0325>$wr;
l\u0325\u0304>$wll;
lh>$wl;l\u0325>$wl;
@ -173,6 +173,7 @@
e\u0306>$wce;
o\u0306>$wco;
e>$wse;
''om>$om;
o>$wso;
n}na > $na|$virama;
n\u0307}[kg] > $anusvara;
@ -184,6 +185,15 @@
m}[pbm] > $anusvara;
n} [yrlvsh] > $anusvara;
'-'m\u0307 > $anusvara;
#urdu compatibility
q>$uka|$virama;
k\u0331h\u0331>$ukha |$virama;
g\u0307> $ugha | $virama;
z > $ujha |$virama;
f > $ufa|$virama;
# dev
y\u0307>$uya|$virama;
l\u0331>$ela|$virama;
n\u0331>$ena|$virama;
@ -198,7 +208,6 @@
d\u0323>$dda|$virama;
kh>$kha|$virama;
k>$ka|$virama;
q>$ka|$virama;
gh>$gha|$virama;
g>$ga|$virama;
ch>$cha|$virama;
@ -219,36 +228,37 @@
b>$ba|$virama;
m>$ma|$virama;
y>$ya|$virama;
r\u0331>$rra|$virama;
r>$ra|$virama;
l\u0323a>$lla;
l>$la|$virama;
v>$va|$virama;
f>$va|$virama;
w>$va|$virama;
sh>$sha|$virama;
ss>$ssa|$virama;
s\u0323>$ssa|$virama;
s\u0301>$sha|$virama;
s>$sa|$virama;
z>$sa|$virama;
h>$ha|$virama;
'.'>$danda;
$danda'.'>$doubleDanda;
$depVowelAbove{'~'>$anusvara;
$depVowelBelow{'~'>$chandrabindu;
$virama aa>$aa;
# 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;
#$virama uu>$uu;
$virama u\u0304>$uu;
$virama u>$u;
$virama rrh>$rrh;
#$virama rrh>$rrh;
$virama r\u0325\u0304>$rrh;
$virama rh>$rh;
#$virama rh>$rh;
$virama r\u0325a>$rh;
$virama r\u0325>$rh;
$virama l\u0325\u0304>$llh;
@ -261,22 +271,23 @@
$virama o\u0306>$co;
$virama e>$se;
$virama o>$so;
$virama''aa>$waa;
# 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;
#$virama''ii>$wii;
$virama''i\u0304>$wii;
$virama''i>$wi;
$virama''uu>$wuu;
#$virama''uu>$wuu;
$virama''u\u0304>$wuu;
$virama''u>$wu;
$virama''rrh>$wrr;
#$virama''rrh>$wrr;
$virama''r\u0325\u0304>$wrr;
$virama''rh>$wr;
#$virama''rh>$wr;
$virama''r\u0325>$wr;
$virama''l\u0325\u0304>$wll;
$virama''lh>$wl;
#$virama''lh>$wl;
$virama''l\u0325>$wl;
$virama''e\u0304>$we;
$virama''o\u0304>$wo;

1
icu4j/src/com/ibm/text/resources/Transliterator_Tamil_InterIndic.txt
View File

@ -16,6 +16,7 @@
#--------------------------------------------------------------------
# Tamil-InterIndic
#:: NFD (NFC) ;
\u0bc6\u0bd7>\ue04c;
\u0b82>\ue002; # SIGN ANUSVARA
\u0b83>\ue003; # SIGN VISARGA
\u0b85>\ue005; # LETTER A

2
icu4j/src/com/ibm/text/resources/Transliterator_index.txt
View File

@ -150,7 +150,7 @@ Gurmukhi-Tamil:alias:NFD;Gurmukhi-InterIndic;InterIndic-Tamil;NFC
Gurmukhi-Telugu:alias:NFD;Gurmukhi-InterIndic;InterIndic-Telugu;NFC
Gurmukhi-Kannada:alias:NFD;Gurmukhi-InterIndic;InterIndic-Kannada;NFC
Gurmukhi-Malayalam:alias:NFD;Gurmukhi-InterIndic;InterIndic-Malayalam;NFC
Gujarati-Devanagari:alias:Gujarati-InterIndic;InterIndic-Devanagari;NFC
Gujarati-Devanagari:alias:NFD;Gujarati-InterIndic;InterIndic-Devanagari;NFC
Gujarati-Bengali:alias:NFD;Gujarati-InterIndic;InterIndic-Bengali;NFC
Gujarati-Gurmukhi:alias:NFD;Gujarati-InterIndic;InterIndic-Gurmukhi;NFC
Gujarati-Oriya:alias:NFD;Gujarati-InterIndic;InterIndic-Oriya;NFC