cldr-transforms 46.0.0-BETA2

package/transforms/Belarusian-Latin-BGN.txt

@@ -0,0 +1,194 @@
+
+# BGN/PCGN 1979 System
+#
+# The BGN/PCGN system for Belarusian (formerly Byelorussian) was
+# designed for use in romanizing names written in the Belarusian
+# Cyrillic alphabet. The Belarusian alphabet contains three
+# letters not present in the Russian alphabet: Іі, Ўў, ’.
+# One obsolete letter Ґґ is included.
+#
+# The Belarusian Alphabet as defined by the BGN (Page 23):
+#
+#   АБВГДЕЁЖЗІЙКЛМНОПРСТУЎФХЦЧШЫЬЭЮЯҐ
+#   абвгдеёжзійклмнопрстуўфхцчшыьэюя’ґ
+#
+# Originally prepared by Michael Everson <everson@evertype.com>
+########################################################################
+::[АБВГДЕЁЖЗІЙКЛМНОПРСТУЎФХЦЧШЩЪЫЬЭЮЯҐабвгдеёжзійклмнопрстуўфхцчшщъыьэюя’ґ] ;
+::NFC;
+$prime  = ʹ ;
+$doublePrime = ʺ ;
+$upperConsonants = [БВГДЖЗЙКЛМНПРСТЎФХЦЧШЬҐ] ;
+$lowerConsonants = [бвгджзйклмнпрстўфхцчшь’ґ] ;
+$consonants = [$upperConsonants $lowerConsonants] ;
+$upperVowels = [АЕЁІОУЫЭЮЯ] ;
+$lowerVowels = [аеёіоуыэюя] ;
+$vowels = [$upperVowels $lowerVowels] ;
+$lower = [$lowerConsonants $lowerVowels] ;
+# Use this $wordBoundary until bug 2034 is fixed in ICU:
+# http://bugs.icu-project.org/cgi-bin/icu-bugs/transliterate?id=2034;expression=boundary;user=guest
+$wordBoundary =  [^[:L:][:M:][:N:]] ;
+# Start of Alphabetic Transformations
+А → A ; # CYRILLIC CAPITAL LETTER A
+а → a ; # CYRILLIC SMALL LETTER A
+Б → B ; # CYRILLIC CAPITAL LETTER BE
+б → b ; # CYRILLIC SMALL LETTER BE
+В → V ; # CYRILLIC CAPITAL LETTER VE
+в → v ; # CYRILLIC SMALL LETTER VE
+Г → H ; # CYRILLIC CAPITAL LETTER GHE
+г → h ; # CYRILLIC SMALL LETTER GHE
+Д → D ; # CYRILLIC CAPITAL LETTER DE
+д → d ; # CYRILLIC SMALL LETTER DE
+Е} $lower → Ye ; # CYRILLIC CAPITAL LETTER IE
+Е → YE ; # CYRILLIC CAPITAL LETTER IE
+е → ye ; # CYRILLIC SMALL LETTER IE
+Ё} $lower → Yo ; # CYRILLIC CAPITAL LETTER IO
+Ё → YO ; # CYRILLIC CAPITAL LETTER IO
+ё → yo ; # CYRILLIC SMALL LETTER IO
+Ж} $lower → Zh ; # CYRILLIC CAPITAL LETTER ZHE
+Ж → ZH ; # CYRILLIC CAPITAL LETTER ZHE
+ж → zh ; # CYRILLIC SMALL LETTER ZHE
+Ш} $lower → Sh;
+Ш → SH;
+ш → sh;
+########################################################################
+#
+# BGN Page 23 Note 1
+#
+# The character sequences зг, цг, сг, тс, and кг may be romanized z·h,
+# k·h, s·h, t·s and ts·h in order to differentiate those romanizations
+# from the digraphs zh, kh, sh, ts, and the letter sequence tsh, which
+# are used to render characters ж, х, ш, ц, and the character sequence тш.
+#
+########################################################################
+ЗГ → Z·H ; # CYRILLIC CAPITAL LETTER ZE
+Зг → Z·h ; # CYRILLIC CAPITAL LETTER ZE
+зг → z·h ; # CYRILLIC SMALL LETTER ZE
+З → Z ; # CYRILLIC CAPITAL LETTER ZE
+з → z ; # CYRILLIC SMALL LETTER ZE
+########################################################################
+#
+# End Rule 1
+#
+########################################################################
+І → I ; # CYRILLIC CAPITAL LETTER BYELORUSSIAN-UKRAINIAN I
+і → i ; # CYRILLIC SMALL LETTER BYELORUSSIAN-UKRAINIAN I
+Й → Y ; # CYRILLIC CAPITAL LETTER I
+й → y ; # CYRILLIC SMALL LETTER I
+########################################################################
+#
+# BGN Page 23 Rule 1
+#
+# кг becomes k·h
+#
+########################################################################
+КГ → K·H ; # CYRILLIC CAPITAL LETTER KA
+Кг → K·h ; # CYRILLIC CAPITAL LETTER KA
+кг → k·h ; # CYRILLIC SMALL LETTER KA
+К → K ; # CYRILLIC CAPITAL LETTER KA
+к → k ; # CYRILLIC SMALL LETTER KA
+########################################################################
+#
+# End Rule 1
+#
+########################################################################
+Л → L ; # CYRILLIC CAPITAL LETTER EL
+л → l ; # CYRILLIC SMALL LETTER EL
+М → M ; # CYRILLIC CAPITAL LETTER EM
+м → m ; # CYRILLIC SMALL LETTER EM
+Н → N ; # CYRILLIC CAPITAL LETTER EN
+н → n ; # CYRILLIC SMALL LETTER EN
+О → O ; # CYRILLIC CAPITAL LETTER O
+о → o ; # CYRILLIC SMALL LETTER O
+П → P ; # CYRILLIC CAPITAL LETTER PE
+п → p ; # CYRILLIC SMALL LETTER PE
+Р → R ; # CYRILLIC CAPITAL LETTER ER
+р → r ; # CYRILLIC SMALL LETTER ER
+########################################################################
+#
+# BGN Page 23 Rule 1
+#
+# сг becomes s·h
+#
+########################################################################
+СГ → S·H ; # CYRILLIC CAPITAL LETTER ES
+Сг → S·h ; # CYRILLIC CAPITAL LETTER ES
+сг → s·h ; # CYRILLIC SMALL LETTER ES
+С → S ; # CYRILLIC CAPITAL LETTER ES
+с → s ; # CYRILLIC SMALL LETTER ES
+########################################################################
+#
+# End Rule 1
+#
+########################################################################
+#
+########################################################################
+#
+# BGN Page 23 Rule 1
+#
+# тс becomes t·s
+#
+########################################################################
+ТС → T·S ; # CYRILLIC CAPITAL LETTER TE
+Тс → T·s ; # CYRILLIC CAPITAL LETTER TE
+тс → t·s ; # CYRILLIC SMALL LETTER TE
+Т → T ; # CYRILLIC CAPITAL LETTER TE
+т → t ; # CYRILLIC SMALL LETTER TE
+########################################################################
+#
+# End Rule 1
+#
+########################################################################
+У → U ; # CYRILLIC CAPITAL LETTER U
+у → u ; # CYRILLIC SMALL LETTER U
+Ў → W ; # CYRILLIC CAPITAL LETTER SHORT U
+ў → w ; # CYRILLIC SMALL LETTER SHORT U
+Ф → F ; # CYRILLIC CAPITAL LETTER EF
+ф → f ; # CYRILLIC SMALL LETTER EF
+Х} $lower → Kh ; # CYRILLIC CAPITAL LETTER HA
+Х → KH ; # CYRILLIC CAPITAL LETTER HA
+х → kh ; # CYRILLIC SMALL LETTER HA
+########################################################################
+#
+# BGN Page 23 Rule 1
+#
+# цг becomes ts·h
+#
+########################################################################
+ЦГ → TS·H ; # CYRILLIC CAPITAL LETTER TSE
+Цг → Ts·h ; # CYRILLIC CAPITAL LETTER TSE
+цг → ts·h ; # CYRILLIC SMALL LETTER TSE
+Ц} $lower → Ts ; # CYRILLIC CAPITAL LETTER TSE
+Ц → TS ; # CYRILLIC CAPITAL LETTER TSE
+ц → ts ; # CYRILLIC SMALL LETTER TSE
+########################################################################
+#
+# End Rule 1
+#
+########################################################################
+Ч} $lower → Ch ; # CYRILLIC CAPITAL LETTER CHE
+Ч → CH ; # CYRILLIC CAPITAL LETTER CHE
+ч → ch ; # CYRILLIC SMALL LETTER CHE
+Ы → Y ; # CYRILLIC CAPITAL LETTER YERU
+ы → y ; # CYRILLIC SMALL LETTER YERU
+Ь → $prime ; # CYRILLIC CAPITAL LETTER SOFT SIGN
+ь → $prime ; # CYRILLIC SMALL LETTER SOFT SIGN
+Э → E ; # CYRILLIC CAPITAL LETTER E
+э → e ; # CYRILLIC SMALL LETTER E
+Ю} $lower → Yu ; # CYRILLIC CAPITAL LETTER YU
+Ю → YU ; # CYRILLIC CAPITAL LETTER YU
+ю → yu ; # CYRILLIC SMALL LETTER YU
+Я} $lower → Ya ; # CYRILLIC CAPITAL LETTER YA
+Я → YA ; # CYRILLIC CAPITAL LETTER YA
+я → ya ; # CYRILLIC SMALL LETTER YA
+’ → $doublePrime ; # LEFT SINGLE QUOTATION MARK
+########################################################################
+#
+# BGN Page 23 Note 2
+#
+# The obsolete character ґ should be romanized g.
+#
+########################################################################
+Ґ → G ; # CYRILLIC CAPITAL LETTER GHE WITH UPTURN
+ґ → g ; # CYRILLIC SMALL LETTER GHE WITH UPTURN
+

package/transforms/Bengali-Arabic.json

@@ -0,0 +1,9 @@
+{
+  "_visibility": "external",
+  "_alias": "Bengali-Arabic",
+  "_aliasBcp47": "und-Arab-t-und-beng",
+  "_source": "Beng",
+  "_direction": "forward",
+  "_target": "Arab",
+  "_rulesFile": "Bengali-Arabic.txt"
+}

package/transforms/Bengali-Arabic.txt

@@ -0,0 +1,7 @@
+
+::[।-॥ঁ-ঃঅ-ঌএ-ঐও-নপ-রলশ-হ়-ৄে-ৈো-্ৗড়-ঢ়য়-ৣ০-৺ৎ];
+::NFD;
+::Bengali-InterIndic;
+::InterIndic-Arabic;
+::NFC;
+

package/transforms/Bengali-Devanagari.json

@@ -0,0 +1,9 @@
+{
+  "_visibility": "external",
+  "_alias": "Bengali-Devanagari",
+  "_aliasBcp47": "und-Deva-t-und-beng",
+  "_source": "Beng",
+  "_direction": "forward",
+  "_target": "Deva",
+  "_rulesFile": "Bengali-Devanagari.txt"
+}

package/transforms/Bengali-Devanagari.txt

@@ -0,0 +1,7 @@
+
+::[।-॥ঁ-ঃঅ-ঌএ-ঐও-নপ-রলশ-হ়-ৄে-ৈো-্ৗড়-ঢ়য়-ৣ০-৺ৎ];
+::NFD;
+::Bengali-InterIndic;
+::InterIndic-Devanagari;
+::NFC;
+

package/transforms/Bengali-Gujarati.json

@@ -0,0 +1,9 @@
+{
+  "_visibility": "external",
+  "_alias": "Bengali-Gujarati",
+  "_aliasBcp47": "und-Gujr-t-und-beng",
+  "_source": "Beng",
+  "_direction": "forward",
+  "_target": "Gujr",
+  "_rulesFile": "Bengali-Gujarati.txt"
+}

package/transforms/Bengali-Gujarati.txt

@@ -0,0 +1,7 @@
+
+::[।-॥ঁ-ঃঅ-ঌএ-ঐও-নপ-রলশ-হ়-ৄে-ৈো-্ৗড়-ঢ়য়-ৣ০-৺ৎ];
+::NFD;
+::Bengali-InterIndic;
+::InterIndic-Gujarati;
+::NFC;
+

package/transforms/Bengali-Gurmukhi.json

@@ -0,0 +1,9 @@
+{
+  "_visibility": "external",
+  "_alias": "Bengali-Gurmukhi",
+  "_aliasBcp47": "und-Guru-t-und-beng",
+  "_source": "Beng",
+  "_direction": "forward",
+  "_target": "Guru",
+  "_rulesFile": "Bengali-Gurmukhi.txt"
+}

package/transforms/Bengali-Gurmukhi.txt

@@ -0,0 +1,7 @@
+
+::[।-॥ঁ-ঃঅ-ঌএ-ঐও-নপ-রলশ-হ়-ৄে-ৈো-্ৗড়-ঢ়য়-ৣ০-৺ৎ];
+::NFD;
+::Bengali-InterIndic;
+::InterIndic-Gurmukhi;
+::NFC;
+

package/transforms/Bengali-InterIndic.json

@@ -0,0 +1,7 @@
+{
+  "_visibility": "internal",
+  "_source": "Bengali",
+  "_direction": "forward",
+  "_target": "InterIndic",
+  "_rulesFile": "Bengali-InterIndic.txt"
+}

package/transforms/Bengali-InterIndic.txt

@@ -0,0 +1,99 @@
+
+# Bengali-InterIndic
+ো→\uE04B; # VOWEL SIGN O
+ৌ→\uE04C; # VOWEL SIGN AU
+ঁ→\uE001; # SIGN CANDRABINDU
+ং→\uE002; # SIGN ANUSVARA
+ঃ→\uE003; # SIGN VISARGA
+অ→\uE005; # LETTER A
+আ→\uE006; # LETTER AA
+ই→\uE007; # LETTER I
+ঈ→\uE008; # LETTER II
+উ→\uE009; # LETTER U
+ঊ→\uE00A; # LETTER UU
+ঋ→\uE00B; # LETTER VOCALIC R
+ঌ→\uE00C; # LETTER VOCALIC L
+এ→\uE00F; # LETTER E
+ঐ→\uE010; # LETTER AI
+ও→\uE013; # LETTER O
+ঔ→\uE014; # LETTER AU
+ক→\uE015; # LETTER KA
+খ→\uE016; # LETTER KHA
+গ→\uE017; # LETTER GA
+ঘ→\uE018; # LETTER GHA
+ঙ→\uE019; # LETTER NGA
+চ→\uE01A; # LETTER CA
+ছ→\uE01B; # LETTER CHA
+জ→\uE01C; # LETTER JA
+ঝ→\uE01D; # LETTER JHA
+ঞ→\uE01E; # LETTER NYA
+ট→\uE01F; # LETTER TTA
+ঠ→\uE020; # LETTER TTHA
+ড→\uE021; # LETTER DDA
+ঢ→\uE022; # LETTER DDHA
+ণ→\uE023; # LETTER NNA
+ত→\uE024; # LETTER TA
+থ→\uE025; # LETTER THA
+দ→\uE026; # LETTER DA
+ধ→\uE027; # LETTER DHA
+ন→\uE028; # LETTER NA
+প→\uE02A; # LETTER PA
+ফ→\uE02B; # LETTER PHA
+ব→\uE02C; # LETTER BA
+ভ→\uE02D; # LETTER BHA
+ম→\uE02E; # LETTER MA
+য→\uE02F; # LETTER YA
+র→\uE030; # LETTER RA
+ল→\uE032; # LETTER LA
+শ→\uE036; # LETTER SHA
+ষ→\uE037; # LETTER SSA
+স→\uE038; # LETTER SA
+হ→\uE039; # LETTER HA
+়→\uE03C; # SIGN NUKTA
+ঽ→\uE03D; # SIGN AVAGRAHA
+া→\uE03E; # VOWEL SIGN AA
+ি→\uE03F; # VOWEL SIGN I
+ী→\uE040; # VOWEL SIGN II
+ু→\uE041; # VOWEL SIGN U
+ূ→\uE042; # VOWEL SIGN UU
+ৃ→\uE043; # VOWEL SIGN VOCALIC R
+ৄ→\uE044; # VOWEL SIGN VOCALIC RR
+ে→\uE047; # VOWEL SIGN E
+ৈ→\uE048; # VOWEL SIGN AI
+ো→\uE04B;
+ৌ→\uE04C;
+#
+্→\uE04D; # SIGN VIRAMA
+ৎ→\uE083; # Khanda-ta
+ৗ→\uE057; # AU LENGTH MARK
+#
+ৠ→\uE060; # LETTER VOCALIC RR
+ৡ→\uE061; # LETTER VOCALIC LL
+ৢ→\uE062; # VOWEL SIGN VOCALIC L
+ৣ→\uE063; # VOWEL SIGN VOCALIC LL
+০→\uE066; # DIGIT ZERO
+১→\uE067; # DIGIT ONE
+২→\uE068; # DIGIT TWO
+৩→\uE069; # DIGIT THREE
+৪→\uE06A; # DIGIT FOUR
+৫→\uE06B; # DIGIT FIVE
+৬→\uE06C; # DIGIT SIX
+৭→\uE06D; # DIGIT SEVEN
+৮→\uE06E; # DIGIT EIGHT
+৯→\uE06F; # DIGIT NINE
+ৰ→\uE071; # Bengali-InterIndic: LETTER RA WITH MIDDLE DIAGONAL
+ৱ→\uE072; # Bengali-InterIndic: LETTER RA WITH LOWER DIAGONAL
+৲→\uE073; # Bengali-InterIndic: RUPEE MARK
+৳→\uE074; # Bengali-InterIndic: RUPEE SIGN
+৴→\uE075; # Bengali-InterIndic: CURRENCY NUMERATOR ONE
+৵→\uE076; # Bengali-InterIndic: CURRENCY NUMERATOR TWO
+৶→\uE077; # Bengali-InterIndic: CURRENCY NUMERATOR THREE
+৷→\uE078; # Bengali-InterIndic: CURRENCY NUMERATOR FOUR
+৸→\uE079; # Bengali-InterIndic: CURRENCY NUMERATOR ONE LESS THAN THE DENOMINATOR
+৹→\uE07A; # Bengali-InterIndic: CURRENCY DENOMINATOR SIXTEEN
+৺→\uE07B; # ISSHAR
+।→\uE064;        # DANDA
+॥→\uE065;        # DOUBLE DANDA
+#    :: NFC (NFD) ;
+# eof
+

package/transforms/Bengali-Kannada.json

@@ -0,0 +1,9 @@
+{
+  "_visibility": "external",
+  "_alias": "Bengali-Kannada",
+  "_aliasBcp47": "und-Knda-t-und-beng",
+  "_source": "Beng",
+  "_direction": "forward",
+  "_target": "Knda",
+  "_rulesFile": "Bengali-Kannada.txt"
+}

package/transforms/Bengali-Kannada.txt

@@ -0,0 +1,7 @@
+
+::[।-॥ঁ-ঃঅ-ঌএ-ঐও-নপ-রলশ-হ়-ৄে-ৈো-্ৗড়-ঢ়য়-ৣ০-৺ৎ];
+::NFD;
+::Bengali-InterIndic;
+::InterIndic-Kannada;
+::NFC;
+

package/transforms/Bengali-Latin.json

@@ -0,0 +1,9 @@
+{
+  "_visibility": "external",
+  "_alias": "Bengali-Latin",
+  "_aliasBcp47": "und-Latn-t-und-beng",
+  "_source": "Beng",
+  "_direction": "forward",
+  "_target": "Latn",
+  "_rulesFile": "Bengali-Latin.txt"
+}

package/transforms/Bengali-Latin.txt

@@ -0,0 +1,7 @@
+
+::[[:script=bengali:][।-॥ঁ-ঃঅ-ঌএ-ঐও-নপ-রলশ-হ়-ৄে-ৈো-্ৗড়-ঢ়য়-ৣ০-৺ৎ]];
+::NFD;
+::Bengali-InterIndic;
+::InterIndic-Latin;
+::NFC;
+

package/transforms/Bengali-Malayalam.json

@@ -0,0 +1,9 @@
+{
+  "_visibility": "external",
+  "_alias": "Bengali-Malayalam",
+  "_aliasBcp47": "und-Mlym-t-und-beng",
+  "_source": "Beng",
+  "_direction": "forward",
+  "_target": "Mlym",
+  "_rulesFile": "Bengali-Malayalam.txt"
+}

package/transforms/Bengali-Malayalam.txt

@@ -0,0 +1,7 @@
+
+::[।-॥ঁ-ঃঅ-ঌএ-ঐও-নপ-রলশ-হ়-ৄে-ৈো-্ৗড়-ঢ়য়-ৣ০-৺ৎ];
+::NFD;
+::Bengali-InterIndic;
+::InterIndic-Malayalam;
+::NFC;
+

package/transforms/Bengali-Oriya.json

@@ -0,0 +1,9 @@
+{
+  "_visibility": "external",
+  "_alias": "Bengali-Oriya",
+  "_aliasBcp47": "und-Orya-t-und-beng",
+  "_source": "Beng",
+  "_direction": "forward",
+  "_target": "Orya",
+  "_rulesFile": "Bengali-Oriya.txt"
+}

package/transforms/Bengali-Oriya.txt

@@ -0,0 +1,7 @@
+
+::[।-॥ঁ-ঃঅ-ঌএ-ঐও-নপ-রলশ-হ়-ৄে-ৈো-্ৗড়-ঢ়য়-ৣ০-৺ৎ];
+::NFD;
+::Bengali-InterIndic;
+::InterIndic-Oriya;
+::NFC;
+

package/transforms/Bengali-Tamil.json

@@ -0,0 +1,9 @@
+{
+  "_visibility": "external",
+  "_alias": "Bengali-Tamil",
+  "_aliasBcp47": "und-Taml-t-und-beng",
+  "_source": "Beng",
+  "_direction": "forward",
+  "_target": "Taml",
+  "_rulesFile": "Bengali-Tamil.txt"
+}

package/transforms/Bengali-Tamil.txt

@@ -0,0 +1,7 @@
+
+::[।-॥ঁ-ঃঅ-ঌএ-ঐও-নপ-রলশ-হ়-ৄে-ৈো-্ৗড়-ঢ়য়-ৣ০-৺ৎ];
+::NFD;
+::Bengali-InterIndic;
+::InterIndic-Tamil;
+::NFC;
+

package/transforms/Bengali-Telugu.json

@@ -0,0 +1,9 @@
+{
+  "_visibility": "external",
+  "_alias": "Bengali-Telugu",
+  "_aliasBcp47": "und-Telu-t-und-beng",
+  "_source": "Beng",
+  "_direction": "forward",
+  "_target": "Telu",
+  "_rulesFile": "Bengali-Telugu.txt"
+}

package/transforms/Bengali-Telugu.txt

@@ -0,0 +1,7 @@
+
+::[।-॥ঁ-ঃঅ-ঌএ-ঐও-নপ-রলশ-হ়-ৄে-ৈো-্ৗড়-ঢ়য়-ৣ০-৺ৎ];
+::NFD;
+::Bengali-InterIndic;
+::InterIndic-Telugu;
+::NFC;
+

package/transforms/Bengali-ur.json

@@ -0,0 +1,8 @@
+{
+  "_visibility": "external",
+  "_aliasBcp47": "ur-t-und-beng",
+  "_source": "Beng",
+  "_direction": "forward",
+  "_target": "ur",
+  "_rulesFile": "Bengali-ur.txt"
+}

package/transforms/Bengali-ur.txt

@@ -0,0 +1,7 @@
+
+::[।-॥ঁ-ঃঅ-ঌএ-ঐও-নপ-রলশ-হ়-ৄে-ৈো-্ৗড়-ঢ়য়-ৣ০-৺ৎ];
+::NFD;
+::Bengali-InterIndic;
+::InterIndic-ur;
+::NFC;
+

package/transforms/Bulgarian-Latin-BGN.json

@@ -0,0 +1,9 @@
+{
+  "_visibility": "external",
+  "_alias": "Bulgarian-Latin/BGN",
+  "_aliasBcp47": "bg-Latn-t-bg-m0-bgn",
+  "_source": "bg",
+  "_direction": "forward",
+  "_target": "bg_Latn",
+  "_rulesFile": "Bulgarian-Latin-BGN.txt"
+}