cldr-transforms 46.0.0-BETA2

package/transforms/Arabic-Latin.txt

@@ -0,0 +1,181 @@
+
+# Generally follows UNGEGN
+#     http://www.eki.ee/wgrs/rom1_ar.pdf
+# Occasionally deviates in the direction of ISO 233
+#     http://homepage.mac.com/sirbinks/pdf/Arabic.pdf
+# a) where required for disambiguation.
+# b) with underdot instead of cedilla for letter like SAD,
+#    since those are explicitly in Unicode for transliteration.
+# c) with extra non-Arabic-language letters, like PEH
+#
+# Does *not* do assimilation of "al", nor hyphenation.
+# While it could be done, we need to determine whether a prefix "al" could
+# occur other than as the definite article (since no space is used).
+:: [[:Arabic:][:block=ARABIC:][‎ⁿ،؛؟ـً-ٕ٠-٬۰-۹﷼ښ][\u0611\u0670]] ;
+:: NFKD (NFC);
+$disambig =  ̱ ;
+$disambig2 =  ̰ ;
+$under =  ̣ ;
+$descender = ˌ;
+$notAbove = [[:^ccc=0:] & [:^ccc=230:]];
+# non-letters
+[:Nd:]{٫}[:Nd:] ↔ [:Nd:]{','}[:Nd:] ; # ARABIC DECIMAL SEPARATOR
+[:Nd:]{٬}[:Nd:] ↔ [:Nd:]{'.'}[:Nd:] ; # ARABIC THOUSANDS SEPARATOR
+٫ ↔ ',' $disambig ; # ARABIC DECIMAL SEPARATOR
+٬ ↔ '.' $disambig ; # ARABIC THOUSANDS SEPARATOR
+#  ٭ ↔ ; # ARABIC FIVE POINTED STAR // no need to transliterate
+، ↔ ',' ; # ARABIC COMMA
+؛ ↔ ';' ; # ARABIC SEMICOLON
+؟ ↔ '?' ; # ARABIC QUESTION MARK
+٪ ↔ '%' ; # ARABIC PERCENT SIGN
+۰ ↔ 0 $disambig ; # EXTENDED ARABIC-INDIC DIGIT ZERO
+۱ ↔ 1 $disambig ; # EXTENDED ARABIC-INDIC DIGIT ONE
+۲ ↔ 2 $disambig ; # EXTENDED ARABIC-INDIC DIGIT TWO
+۳ ↔ 3 $disambig ; # EXTENDED ARABIC-INDIC DIGIT THREE
+۴ ↔ 4 $disambig ; # EXTENDED ARABIC-INDIC DIGIT FOUR
+۵ ↔ 5 $disambig ; # EXTENDED ARABIC-INDIC DIGIT FIVE
+۶ ↔ 6 $disambig ; # EXTENDED ARABIC-INDIC DIGIT SIX
+۷ ↔ 7 $disambig ; # EXTENDED ARABIC-INDIC DIGIT SEVEN
+۸ ↔ 8 $disambig ; # EXTENDED ARABIC-INDIC DIGIT EIGHT
+۹ ↔ 9 $disambig ; # EXTENDED ARABIC-INDIC DIGIT NINE
+٠ ↔ 0 ; # ARABIC-INDIC DIGIT ZERO
+١ ↔ 1 ; # ARABIC-INDIC DIGIT ONE
+٢ ↔ 2 ; # ARABIC-INDIC DIGIT TWO
+٣ ↔ 3 ; # ARABIC-INDIC DIGIT THREE
+٤ ↔ 4 ; # ARABIC-INDIC DIGIT FOUR
+٥ ↔ 5 ; # ARABIC-INDIC DIGIT FIVE
+٦ ↔ 6 ; # ARABIC-INDIC DIGIT SIX
+٧ ↔ 7 ; # ARABIC-INDIC DIGIT SEVEN
+٨ ↔ 8 ; # ARABIC-INDIC DIGIT EIGHT
+٩ ↔ 9 ; # ARABIC-INDIC DIGIT NINE
+؉ ↔ ‰ ;	# U+0609	ARABIC-INDIC PER MILLE SIGN
+؊ ↔ ‱ ;	 # U+060A	ARABIC-INDIC PER TEN THOUSAND SIGN
+‎۔‎ ↔ '.' ; 	# U+06D4	ARABIC FULL STOP
+# letters
+# long vowels
+َا↔ ā ; # ARABIC FATHA, ARABIC LETTER ALEF
+ُو ↔ ū ; # ARABIC DAMMA, ARABIC LETTER WAW
+ِي ↔ ī ; # ARABIC KASRA, ARABIC LETTER YEH
+# longer items moved here to prevent masking
+ث ↔ t h $disambig ; # ARABIC LETTER THEH
+ذ ↔ d h $disambig ; # ARABIC LETTER THAL
+ش ↔ s h $disambig ; # ARABIC LETTER SHEEN
+ص ↔ s $under ; # ARABIC LETTER SAD
+ض ↔ d $under ; # ARABIC LETTER DAD
+ط ↔ t $under ; # ARABIC LETTER TAH
+ظ ↔ z $under ; # ARABIC LETTER ZAH
+غ ↔ g h $disambig ; # ARABIC LETTER GHAIN
+# WARNING: special case
+# ←t, umlaut, half-ring below→ will be canonically ordered as ←t, half-ring below, umlaut→
+# so on the return, we have to skip over (but preserve) the half-ring below (or others like it)
+# ةٕ ← ẗ̹ ; # LATIN SMALL LETTER T, COMBINING RIGHT HALF RING BELOW, COMBINING DIAERESIS
+ة ↔ t ̈ ; # ARABIC LETTER TEH MARBUTA
+ة | $1 ← t ($notAbove+) ̈ ; # ARABIC LETTER TEH MARBUTA
+# non-Arabic language
+ژ ↔ z h $disambig ; # ARABIC LETTER JEH
+ڭ ↔ n $disambig g ; # ARABIC LETTER NG
+ۋ ↔ v $disambig ; # ARABIC LETTER VE
+ی ↔ y $disambig2 ; # ARABIC LETTER FARSI YEH
+ښ ↔ s $descender;
+# Arabic language
+ء ↔ ʾ ; # ARABIC LETTER HAMZA
+ا ↔ a $under; # ARABIC LETTER ALEF
+ب ↔ b ; # ARABIC LETTER BEH
+ت ↔ t ; # ARABIC LETTER TEH
+ج ↔ j ; # ARABIC LETTER JEEM
+ح ↔ h $under ; # ARABIC LETTER HAH
+خ ↔ k h $disambig ; # ARABIC LETTER KHAH
+د ↔ d ; # ARABIC LETTER DAL
+ر ↔ r ; # ARABIC LETTER REH
+ز ↔ z ; # ARABIC LETTER ZAIN
+س ↔ s ; # ARABIC LETTER SEEN
+ع ↔ ʿ ; # ARABIC LETTER AIN
+ـ → ; # ARABIC TATWEEL
+ف ↔ f ; # ARABIC LETTER FEH
+ق ↔ q ; # ARABIC LETTER QAF
+ک ↔ k $disambig ; # ARABIC LETTER KEHEH
+ك ↔ k ; # ARABIC LETTER KAF
+ل ↔ l ; # ARABIC LETTER LAM
+م ↔ m ; # ARABIC LETTER MEEM
+ن ↔ n ; # ARABIC LETTER NOON
+ه ↔ h ; # ARABIC LETTER HEH
+و ↔ w ; # ARABIC LETTER WAW
+ى ↔ y $disambig ; # ARABIC LETTER ALEF MAKSURA
+ي ↔ y ; # ARABIC LETTER YEH
+ً ↔ aⁿ ; # ARABIC FATHATAN
+ٌ ↔ uⁿ ; # ARABIC DAMMATAN
+ٍ ↔ iⁿ ; # ARABIC KASRATAN
+َ ↔ a ; # ARABIC FATHA
+ُ ↔ u ; # ARABIC DAMMA
+ِ ↔ i ; # ARABIC KASRA
+ّ ↔   ̃ ; # ARABIC SHADDA
+ْ ↔   ̊ ; # ARABIC SUKUN
+# special combining marks
+ٓ ↔  ̂ ; # ARABIC MADDAH ABOVE
+ٔ ↔  ̉ ; # ARABIC HAMZA ABOVE
+ٕ ↔  ̹ ; # ARABIC HAMZA BELOW
+# Some non-Arabic language (not in UNGEGN)
+پ ↔ p ; # ARABIC LETTER PEH
+چ ↔ c h $disambig ; # ARABIC LETTER TCHEH
+ڤ ↔ v ; # ARABIC LETTER VEH
+# ڥ ↔ v $disambig ; # ARABIC LETTER FEH WITH THREE DOTS BELOW
+# ڢ ↔ f $disambig ; # ARABIC LETTER FEH WITH DOT MOVED BELOW
+گ ↔ g ; # ARABIC LETTER GAF
+# fallbacks TODO roundtrip where possible, using diacritics to distinguish
+#https://en.wikipedia.org/wiki/Sindhi_transliteration
+‎ٺ‎→ṭh;
+‎ٿ‎→th;
+‎ٽ‎→ṭ;
+‎ڙ‎→ṛ;
+‎ڦ‎→ph;
+‎ڻ‎→ṇ;
+‎ڱ‎→ṅ;
+‎ڃ‎→ñ;
+‎ڪ‎→k;
+‎ڄ‎→j̈;
+‎ۃ‎→ẖ;
+‎ڳ‎→g̤;
+‎ڍ‎→ḍh;
+‎ڌ‎→dh;
+‎ڏ‎→d̤;
+‎ڊ‎→ḍ;
+‎ڇ‎→ch;
+‎ڀ‎→bh;
+‎ٻ‎→ḇ;
+‎۽‎→'&';
+‎۾‎→'mn';
+#https://en.wiktionary.org/wiki/Wiktionary:Urdu_transliteration
+‎ھ‎ → ʱ ;
+‎ں‎ → ◌̃ ;
+‎ے‎ → ai ;
+‎ڈ‎ → ḍ ;
+‎ڑ‎ → ṛ ;
+‎ٹ‎ → ṭ ;
+#https://www.eki.ee/wgrs/rom2_ps.htm
+#https://en.wikipedia.org/wiki/Pashto_alphabet
+‎ټ‎ → ṯ ;
+‎ځ‎ → dz ;
+‎څ‎ → ts ;
+‎ډ‎ → ḏ ;
+‎ړ‎ → ṟ ;
+‎ږ‎ → z͟h ;
+‎ګ‎ → g ;
+‎ڼ‎ → ṉ ;
+‎ۍ‎ → ạy ;
+‎ې‎ → e ;
+#https://www.eki.ee/wgrs/rom1_ug.pdf
+‎ہ‎ → ḥ ;
+‎ە‎ → ĥ ;
+# Delete marks without correspondants
+[\u0611\u0670] → ;
+# fallbacks
+| s ← c } [eiy];
+| k ← c ;
+| i ← e ;
+| u ← o ;
+| ks ← x ;
+| n ← ‎ⁿ;
+:: (lower) ;
+::NFC (NFD);
+:: ( [[:Latin:] [%,.0-9;?ʾ-ʿ̂-̄̈-̣̰̊-̱̹;ˌ]] );
+

package/transforms/Armenian-Latin-BGN.json

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

package/transforms/Armenian-Latin-BGN.txt

@@ -0,0 +1,165 @@
+
+# BGN/PCGN 1981 System
+#
+# The BGN/PCGN system for Armenian was designed for use in romanizing
+# names written in the Armenian alphabet. The Roman letters and letter
+# combinations shown as equivalents to the Armenian characters reflect
+# the eastern variety of Armenian, i.e., the language spoken in
+# the Republic of Armenia.
+#
+# The Armenian Alphabet as defined by the BGN (Page 11):
+#
+#   ԱԲԳԴԵԶԷԸԹԺԻԼԽԾԿՀՁՂՃՄՅՆՇՈՉՊՋՌՍՎՐՑՓՔՕՖ
+#   աբգդեզէըթժիլխծկհձղճմյնշոչպջռսվտրցփքևօֆ
+#
+# Originally prepared by Michael Everson <everson@evertype.com>
+#
+# https://www.gov.uk/government/uploads/system/uploads/attachment_data/file/614615/ROMANIZATION_SYSTEM_FOR_ARMENIAN.PDF
+::[ԱԲԳԴԵԶԷԸԹԺԻԼԽԾԿՀՁՂՃՄՅՆՇՈՉՊՋՌՍՎՏՐՑՒՓՔՕՖաբգդեզէըթժիլխծկհձղճմյնշոչպջռսվտրցւփքօֆև։];
+::NFC;
+$upperConsonants = [ԲԳԴԶԹԺԼԽԾԿՀՁՂՃՄՅՆՇՉՊՋՌՍՎՐՑՓՔՖ] ;
+$lowerConsonants = [բգդզթժլխծկհձղճմյնշչպջռսվտրցփքֆ] ;
+$consonants = [$upperConsonants $lowerConsonants] ;
+$upperVowels = [ԱԵԷԸԻՈՕՒ] ;
+$lowerVowels = [աեէըիոևօւ] ;
+$vowels = [$upperVowels $lowerVowels] ;
+$lower = [$lowerConsonants $lowerVowels] ;
+$aspirate = ’ ;
+# 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:]] ;
+# BGN/PCGN 1981, Note 5: “The characters ԵՎ , եւ and և should be romanized
+# yev initially and after the vowel characters ա, ե, է, ի, ո, ու, and օ.
+# In all other instances these characters should be romanized ev.”
+#
+# BGN/PCGN 1981, Note 3: “In Soviet-era sources this upper-case digraph
+# character is found as Ե ի.”
+$YEV = [{ԵՎ} {ԵՒ} {ԵԻ}];
+$Yev = [{Եվ} {Եւ} {Եի}];
+$yev = [{եվ} {եւ} և {եի}];
+$yev_vowels = [ա ե է ի ո {ու} օ Ա Ե Է Ի Ո {ՈՒ} {Ու} Օ];
+[$wordBoundary $yev_vowels] {$YEV} → YEV;
+[$wordBoundary $yev_vowels] {$Yev} → Yev;
+[$wordBoundary $yev_vowels] {$yev} → yev;
+$YEV → EV;
+$Yev → Ev;
+$yev → ev;
+::null;
+# BGN Page 12 Rule 1: The character ե should be romanized ye initially,
+# after the vowel characters ա, ե, է, ը, ի, ո, ւ, and օ.
+# In all other instances, it should be romanized e.
+$upperVowels {Ե → YE ; # ARMENIAN CAPITAL LETTER ECH
+$lowerVowels {Ե → Ye ; # ARMENIAN CAPITAL LETTER ECH
+$wordBoundary {Ե → Ye ; # ARMENIAN CAPITAL LETTER ECH
+Ե → E ; # ARMENIAN CAPITAL LETTER ECH
+$vowels {ե → ye ; # ARMENIAN SMALL LETTER ECH
+$wordBoundary {ե → ye ; # ARMENIAN SMALL LETTER ECH
+ե → e ; # ARMENIAN SMALL LETTER ECH
+::null;
+Ա → A ; # ARMENIAN CAPITAL LETTER AYB
+ա → a ; # ARMENIAN SMALL LETTER AYB
+Բ → B ; # ARMENIAN CAPITAL LETTER BEN
+բ → b ; # ARMENIAN SMALL LETTER BEN
+Գ → G ; # ARMENIAN CAPITAL LETTER GIM
+գ → g ; # ARMENIAN SMALL LETTER GIM
+Դ → D ; # ARMENIAN CAPITAL LETTER DA
+դ → d ; # ARMENIAN SMALL LETTER DA
+Զ → Z ; # ARMENIAN CAPITAL LETTER ZA
+զ → z ; # ARMENIAN SMALL LETTER ZA
+Է → E ; # ARMENIAN CAPITAL LETTER EH
+է → e ; # ARMENIAN SMALL LETTER EH
+Ը → Y ; # ARMENIAN CAPITAL LETTER ET
+ը → y ; # ARMENIAN SMALL LETTER ET
+Թ → T $aspirate ; # ARMENIAN CAPITAL LETTER TO
+թ → t $aspirate ; # ARMENIAN SMALL LETTER TO
+Ժ} $lower → Zh ; # ARMENIAN CAPITAL LETTER ZHE
+Ժ → ZH ; # ARMENIAN CAPITAL LETTER ZHE
+ժ → zh ; # ARMENIAN SMALL LETTER ZHE
+Ի → I ; # ARMENIAN CAPITAL LETTER INI
+ի → i ; # ARMENIAN SMALL LETTER INI
+Լ → L ; # ARMENIAN CAPITAL LETTER LIWN
+լ → l ; # ARMENIAN SMALL LETTER LIWN
+Խ} $lower → Kh ; # ARMENIAN CAPITAL LETTER XEH
+Խ → KH ; # ARMENIAN CAPITAL LETTER XEH
+խ → kh ; # ARMENIAN SMALL LETTER XEH
+Ծ} $lower → Ts ; # ARMENIAN CAPITAL LETTER CA
+Ծ → TS ; # ARMENIAN CAPITAL LETTER CA
+ծ → ts ; # ARMENIAN SMALL LETTER CA
+Կ → K ; # ARMENIAN CAPITAL LETTER KEN
+կ → k ; # ARMENIAN SMALL LETTER KEN
+Հ → H ; # ARMENIAN CAPITAL LETTER HO
+հ → h ; # ARMENIAN SMALL LETTER HO
+Ձ} $lower → Dz ; # ARMENIAN CAPITAL LETTER JA
+Ձ → DZ ; # ARMENIAN CAPITAL LETTER JA
+ձ → dz ; # ARMENIAN SMALL LETTER JA
+Ղ} $lower → Gh ; # ARMENIAN CAPITAL LETTER GHAD
+Ղ → GH ; # ARMENIAN CAPITAL LETTER GHAD
+ղ → gh ; # ARMENIAN SMALL LETTER GHAD
+Ճ} $lower → Ch ; # ARMENIAN CAPITAL LETTER CHEH
+Ճ → CH ; # ARMENIAN CAPITAL LETTER CHEH
+ճ → ch ; # ARMENIAN SMALL LETTER CHEH
+Մ → M ; # ARMENIAN CAPITAL LETTER MEN
+մ → m ; # ARMENIAN SMALL LETTER MEN
+Յ → Y ; # ARMENIAN CAPITAL LETTER YI
+յ → y ; # ARMENIAN SMALL LETTER YI
+Ն → N ; # ARMENIAN CAPITAL LETTER NOW
+ն → n ; # ARMENIAN SMALL LETTER NOW
+Շ} $lower → Sh ; # ARMENIAN CAPITAL LETTER SHA
+Շ → SH ; # ARMENIAN CAPITAL LETTER SHA
+շ → sh ; # ARMENIAN SMALL LETTER SHA
+# Transliteration Case 34:
+Ու} $lower → U ; # ARMENIAN CAPITAL LETTER VO + SMALL YIWN
+ՈՒ → U ; # ARMENIAN CAPITAL LETTER VO + CAPITAL YIWN
+ու → u ; # ARMENIAN SMALL LETTER VO + SMALL YIWN
+# BGN Page 12 Rule 2:
+#
+# The character ո should be romanized vo initially except in the
+# word ով, which should be romanized ov. In all other instances, it
+# should be romanized o.
+ՈՎ → OV ;
+Ով → Ov ;
+ով → ov ;
+$wordBoundary{Ո}$lower → Vo ; # ARMENIAN CAPITAL LETTER VO
+$wordBoundary{Ո → VO ; # ARMENIAN CAPITAL LETTER VO
+Ո →  O ; # ARMENIAN CAPITAL LETTER ECH
+$wordBoundary{ո → vo ; # ARMENIAN SMALL LETTER VO
+ո →  o ; # ARMENIAN SMALL LETTER VO
+Չ} $lower → Ch $aspirate ; # ARMENIAN CAPITAL LETTER CHA
+Չ → CH $aspirate ; # ARMENIAN CAPITAL LETTER CHA
+չ → ch $aspirate ; # ARMENIAN SMALL LETTER CHA
+Պ → P ; # ARMENIAN CAPITAL LETTER PEH
+պ → p ; # ARMENIAN SMALL LETTER PEH
+Ջ → J ; # ARMENIAN CAPITAL LETTER JHEH
+ջ → j ; # ARMENIAN SMALL LETTER JHEH
+Ռ} $lower → Rr ; # ARMENIAN CAPITAL LETTER RA
+Ռ → RR ; # ARMENIAN CAPITAL LETTER RA
+ռ → rr ; # ARMENIAN SMALL LETTER RA
+Ս → S ; # ARMENIAN CAPITAL LETTER SEH
+ս → s ; # ARMENIAN SMALL LETTER SEH
+Վ → V ; # ARMENIAN CAPITAL LETTER VEW
+վ → v ; # ARMENIAN SMALL LETTER VEW
+Տ → T ; # ARMENIAN CAPITAL LETTER TIWN
+տ → t ; # ARMENIAN SMALL LETTER TIWN
+Ր → R ; # ARMENIAN CAPITAL LETTER REH
+ր → r ; # ARMENIAN SMALL LETTER REH
+Ց} $lower → Ts $aspirate ; # ARMENIAN CAPITAL LETTER CHEH
+Ց → TS $aspirate ; # ARMENIAN CAPITAL LETTER CO
+ց → ts $aspirate ; # ARMENIAN SMALL LETTER CO
+########################################################################
+#
+# The BGN does not show YIWN on its own.
+#
+#Ւ → W ; # ARMENIAN CAPITAL LETTER YIWN
+#ւ → w ; # ARMENIAN SMALL LETTER YIWN
+#
+########################################################################
+Փ → P $aspirate ; # ARMENIAN CAPITAL LETTER PIWR
+փ → p $aspirate ; # ARMENIAN SMALL LETTER PIWR
+Ք → K $aspirate ; # ARMENIAN CAPITAL LETTER KEH
+ք → k $aspirate ; # ARMENIAN SMALL LETTER KEH
+Օ → O ; # ARMENIAN CAPITAL LETTER OH
+օ → o ; # ARMENIAN SMALL LETTER OH
+Ֆ → F ; # ARMENIAN CAPITAL LETTER FEH
+ֆ → f ; # ARMENIAN SMALL LETTER FEH
+։ → \. ; # ARMENIAN FULL STOP
+

package/transforms/Azerbaijani-Latin-BGN.json

@@ -0,0 +1,9 @@
+{
+  "_visibility": "external",
+  "_alias": "Azerbaijani-Latin/BGN",
+  "_aliasBcp47": "az-t-az-cyrl-m0-bgn",
+  "_source": "az_Cyrl",
+  "_direction": "forward",
+  "_target": "az",
+  "_rulesFile": "Azerbaijani-Latin-BGN.txt"
+}

package/transforms/Azerbaijani-Latin-BGN.txt

@@ -0,0 +1,183 @@
+
+#
+########################################################################
+# BGN/PCGN 1993 Agreement
+#
+# Azerbaijani is the official language of Azerbaijan. In 1991, the
+# Azerbaijani government adopted a Roman alphabet to replace the
+# existing Cyrillic alphabet. The Azerbaijani Cyrillic alphabet
+# contains nine letters not present in the Russian alphabet:
+# Ғғ, Әә, Јј, Ҝҝ, Өө, Үү, Һһ, Ҹҹ, and ’. Four obsolete letters
+# Йй, Ээ, Юю and Яя are also given.
+#
+# The Azerbaijani Alphabet as defined by the BGN (Page 13):
+#
+#   АБВГҒДЕӘЖЗИЫЈКҜЛМНОӨПРСТУҮФХҺЧҸШЙЭЮЯ
+#   абвгғдеәжзиыјкҝлмноөпрстуүфхһчҹш’йэюя
+#
+# Originally prepared by Michael Everson <everson@evertype.com>
+########################################################################
+#
+# MINIMAL FILTER: AzerbaijaniCyrl-Latin
+#
+:: [АБВГҒДЕӘЖЗИЫЈКҜЛМНОӨПРСТУҮФХҺЧҸШЙЭЮЯабвгғдеәжзиыјкҝлмноөпрстуүфхһчҹш’йэюя] ;
+:: NFD (NFC) ;
+#
+#
+########################################################################
+#
+########################################################################
+#
+# Define All Transformation Variables
+#
+########################################################################
+#
+$prime  = ’ ;
+$wordBoundary =  [^[:L:][:M:][:N:]] ;
+$upperConsonants = [БВГҒДЖЗЈКҜЛМНПРСТФХҺЧҸШЙ] ;
+$lowerConsonants = [бвгғджзјкҝлмнпрстфхһчҹш’й] ;
+$consonants = [$upperConsonants $lowerConsonants] ;
+$upperVowels = [АЕӘИЫОӨУҮЭЮЯ] ;
+$lowerVowels = [аеәиыоөуүэюя] ;
+$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
+#
+#
+#
+########################################################################
+#
+########################################################################
+#
+# 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 → Q ; # CYRILLIC CAPITAL LETTER GHE
+г → q ; # CYRILLIC SMALL LETTER GHE
+Ғ → Ğ ; # CYRILLIC CAPITAL LETTER GHE WITH STROKE
+ғ → ğ ; # CYRILLIC SMALL LETTER GHE WITH STROKE
+Д → D ; # CYRILLIC CAPITAL LETTER DE
+д → d ; # CYRILLIC SMALL LETTER DE
+Е → E ; # CYRILLIC CAPITAL LETTER DE
+е → e ; # CYRILLIC SMALL LETTER DE
+#
+########################################################################
+#
+# BGN Page 14 Note 1
+#
+# The special letter Ə ə, known as schwa, should be reproduced in that
+# form whenever encountered. In those instances when it cannot be
+# reproduced, however, the letter Ä ä may be substituted for it.
+#
+########################################################################
+#
+Ә → Ə; # CYRILLIC CAPITAL LETTER SCHWA
+ә → ə; # CYRILLIC SMALL LETTER SCHWA
+#
+#
+# Alternative rule when schwa is not available. To apply uncomment the
+# following by removing the '#' mark at the start of the line and insert
+# before the two rule lines above.
+#
+# Ә → Ä; # CYRILLIC CAPITAL LETTER SCHWA
+# ә → ä; # CYRILLIC SMALL LETTER SCHWA
+#
+########################################################################
+#
+# End BGN Page 14 Note 1
+#
+########################################################################
+Ж → J ; # CYRILLIC CAPITAL LETTER ZHE
+ж → j ; # CYRILLIC SMALL LETTER ZHE
+З → Z ; # CYRILLIC CAPITAL LETTER ZE
+з → z ; # CYRILLIC SMALL LETTER ZE
+И → İ ; # CYRILLIC CAPITAL LETTER I
+и → i ; # CYRILLIC SMALL LETTER I
+Ы → I ; # CYRILLIC CAPITAL LETTER YERU
+ы → ı ; # CYRILLIC SMALL LETTER YERU
+Ј → Y ; # CYRILLIC CAPITAL LETTER JE
+ј → y ; # CYRILLIC SMALL LETTER JE
+К → K ; # CYRILLIC CAPITAL LETTER KA
+к → k ; # CYRILLIC SMALL LETTER KA
+Ҝ → G ; # CYRILLIC CAPITAL LETTER KA WITH VERTICAL STROKE
+ҝ → g ; # CYRILLIC SMALL LETTER KA WITH VERTICAL STROKE
+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
+Ө → Ö ; # CYRILLIC CAPITAL LETTER BARRED O
+ө → ö ; # CYRILLIC SMALL LETTER BARRED 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
+С → S ; # CYRILLIC CAPITAL LETTER ES
+с → s ; # CYRILLIC SMALL LETTER ES
+Т → T ; # CYRILLIC CAPITAL LETTER TE
+т → t ; # CYRILLIC SMALL LETTER TE
+У → U ; # CYRILLIC CAPITAL LETTER U
+у → u ; # CYRILLIC SMALL LETTER U
+Ү → Ü ; # CYRILLIC CAPITAL LETTER STRAIGHT U
+ү → ü ; # CYRILLIC SMALL LETTER STRAIGHT U
+Ф → F ; # CYRILLIC CAPITAL LETTER EF
+ф → f ; # CYRILLIC SMALL LETTER EF
+Х → X ; # CYRILLIC CAPITAL LETTER HA
+х → x ; # CYRILLIC SMALL LETTER HA
+Һ → H ; # CYRILLIC CAPITAL LETTER SHHA
+һ → h ; # CYRILLIC SMALL LETTER SHHA
+Ч → Ç ; # CYRILLIC CAPITAL LETTER CHE
+ч → ç ; # CYRILLIC SMALL LETTER CHE
+Ҹ → C ; # CYRILLIC CAPITAL LETTER CHE WITH VERTICAL STROKE
+ҹ → c ; # CYRILLIC SMALL LETTER CHE WITH VERTICAL STROKE
+Ш → Ş ; # CYRILLIC CAPITAL LETTER SHA
+ш → ş ; # CYRILLIC SMALL LETTER SHA
+#
+########################################################################
+#
+# BGN Page 13 Rule 33, maps the symbol onto itself and
+# is ignored here for computational efficiency.
+#
+# $prime → $prime ; # RIGHT SINGLE QUOTATION MARK
+#
+########################################################################
+#
+########################################################################
+#
+# BGN Page 14 Note 2:
+#
+# The obsolete characters й, э, ю, and я should be romanized ẏ, ė,
+# yu̇, and yȧ.
+#
+########################################################################
+#
+Й → Ẏ ; # CYRILLIC CAPITAL LETTER HARD SIGN
+й → ẏ ; # CYRILLIC SMALL LETTER HARD SIGN
+Э → Ė ; # CYRILLIC CAPITAL LETTER SOFT SIGN
+э → ė ; # CYRILLIC SMALL LETTER SOFT SIGN
+Ю} $lower → Yu̇ ; # CYRILLIC CAPITAL LETTER YU
+Ю → YU̇ ; # CYRILLIC CAPITAL LETTER YU
+ю → yu̇ ;  # CYRILLIC SMALL LETTER YU
+Я} $lower → Yȧ ; # CYRILLIC CAPITAL LETTER YA
+Я → YȦ ; # CYRILLIC CAPITAL LETTER YA
+я → yȧ ; # CYRILLIC SMALL LETTER YA
+#
+#
+########################################################################
+#
+# End BGN Page 14 Note 2.
+#
+########################################################################
+

package/transforms/Belarusian-Latin-BGN.json

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