cldr-transforms 46.0.0-BETA2

package/transforms/Oriya-InterIndic.txt

@@ -0,0 +1,91 @@
+
+# Oriya-InterIndic
+#:: NFD (NFC) ;
+#ଡ଼→\uE05C;# LETTER RRA
+#ଢ଼→\uE05D;# LETTER RHA
+ୈ→\uE048;# VOWEL SIGN AI
+ୋ→\uE04B;# VOWEL SIGN O
+ୌ→\uE04C;# VOWEL SIGN AU
+ଁ→\uE001;      # SIGN CANDRABINDU
+ଂ→\uE002;      # SIGN ANUSVARA
+ଃ→\uE003;      # SIGN VISARGA
+ଅ→\uE005;      # LETTER A
+ଆ→\uE006;      # LETTER AA
+ଇ→\uE007;      # LETTER I
+ଈ→\uE008;      # LETTER II
+ଉ→\uE009;      # LETTER U
+ଊ→\uE00A;      # LETTER UU
+ଋ→\uE00B;      # LETTER VOCALIC R
+ଌ→\uE00C;      # LETTER VOCALIC L
+ଏ→\uE00F;      # LETTER E
+ଐ→\uE010;      # LETTER AI
+ଓ→\uE013;      # LETTER O
+ଔ→\uE014;      # LETTER AU
+କ→\uE015;      # LETTER KA
+ଖ→\uE016;      # LETTER KHA
+ଗ→\uE017;      # LETTER GA
+ଘ→\uE018;      # LETTER GHA
+ଙ→\uE019;      # LETTER NGA
+ଚ→\uE01A;      # LETTER CA
+ଛ→\uE01B;      # LETTER CHA
+ଜ→\uE01C;      # LETTER JA
+ଝ→\uE01D;      # LETTER JHA
+ଞ→\uE01E;      # LETTER NYA
+ଟ→\uE01F;      # LETTER TTA
+ଠ→\uE020;      # LETTER TTHA
+ଡ→\uE021;      # LETTER DDA
+ଢ→\uE022;      # LETTER DDHA
+ଣ→\uE023;      # LETTER NNA
+ତ→\uE024;      # LETTER TA
+ଥ→\uE025;      # LETTER THA
+ଦ→\uE026;      # LETTER DA
+ଧ→\uE027;      # LETTER DHA
+ନ→\uE028;      # LETTER NA
+ପ→\uE02A;      # LETTER PA
+ଫ→\uE02B;      # LETTER PHA
+ବ→\uE02C;      # LETTER BA
+ଭ→\uE02D;      # LETTER BHA
+ମ→\uE02E;      # LETTER MA
+ଯ→\uE02F;      # LETTER YA
+ର→\uE030;      # LETTER RA
+ଲ→\uE032;      # LETTER LA
+ଳ→\uE033;      # LETTER LLA
+ଵ→\uE035;      # LETTER VA
+ଶ→\uE036;      # LETTER SHA
+ଷ→\uE037;      # LETTER SSA
+ସ→\uE038;      # LETTER SA
+ହ→\uE039;      # LETTER HA
+଼→\uE03C;      # SIGN NUKTA
+ଽ→\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
+େ→\uE047;      # VOWEL SIGN E
+#
+୍→\uE04D;      # SIGN VIRAMA
+ୖ→\uE056;      # AI LENGTH MARK
+ୗ→\uE057;      # AU LENGTH MARK
+।→\uE064;      # DANDA
+॥→\uE065;      # DOUBLE DANDA
+#
+ୟ→\uE05F;      # LETTER YYA
+ୠ→\uE060;      # LETTER VOCALIC RR
+ୡ→\uE061;      # LETTER VOCALIC LL
+୦→\uE066;      # DIGIT ZERO
+୧→\uE067;      # DIGIT ONE
+୨→\uE068;      # DIGIT TWO
+୩→\uE069;      # DIGIT THREE
+୪→\uE06A;      # DIGIT FOUR
+୫→\uE06B;      # DIGIT FIVE
+୬→\uE06C;      # DIGIT SIX
+୭→\uE06D;      # DIGIT SEVEN
+୮→\uE06E;      # DIGIT EIGHT
+୯→\uE06F;      # DIGIT NINE
+୰→\uE07B;      # ISSHAR
+ୱ→\uE081;      # LETTER WA
+#    :: NFC (NFD) ;
+# eof
+

package/transforms/Oriya-Kannada.json

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

package/transforms/Oriya-Kannada.txt

@@ -0,0 +1,7 @@
+
+::[ଁ-ଃଅ-ଌଏ-ଐଓ-ନପ-ରଲ-ଳଵ-ହ଼-ୃେ-ୈୋ-୍ୖ-ୗଡ଼-ଢ଼ୟ-ୡ୦-ୱ];
+::NFD;
+::Oriya-InterIndic;
+::InterIndic-Kannada;
+::NFC;
+

package/transforms/Oriya-Latin.json

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

package/transforms/Oriya-Latin.txt

@@ -0,0 +1,7 @@
+
+::[୍।-॥ଁ-ଃଅ-ଌଏ-ଐଓ-ନପ-ରଲ-ଳଵଶ-ହ଼-ୃେ-ୈୋ-୍ୖ-ୗଡ଼-ଢ଼ୟ-ୡ୦-୰ୱ];
+::NFD;
+::Oriya-InterIndic;
+::InterIndic-Latin;
+::NFC;
+

package/transforms/Oriya-Malayalam.json

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

package/transforms/Oriya-Malayalam.txt

@@ -0,0 +1,7 @@
+
+::[ଁ-ଃଅ-ଌଏ-ଐଓ-ନପ-ରଲ-ଳଵ-ହ଼-ୃେ-ୈୋ-୍ୖ-ୗଡ଼-ଢ଼ୟ-ୡ୦-ୱ];
+::NFD;
+::Oriya-InterIndic;
+::InterIndic-Malayalam;
+::NFC;
+

package/transforms/Oriya-Tamil.json

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

package/transforms/Oriya-Tamil.txt

@@ -0,0 +1,7 @@
+
+::[ଁ-ଃଅ-ଌଏ-ଐଓ-ନପ-ରଲ-ଳଵ-ହ଼-ୃେ-ୈୋ-୍ୖ-ୗଡ଼-ଢ଼ୟ-ୡ୦-ୱ];
+::NFD;
+::Oriya-InterIndic;
+::InterIndic-Tamil;
+::NFC;
+

package/transforms/Oriya-Telugu.json

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

package/transforms/Oriya-Telugu.txt

@@ -0,0 +1,7 @@
+
+::[ଁ-ଃଅ-ଌଏ-ଐଓ-ନପ-ରଲ-ଳଵ-ହ଼-ୃେ-ୈୋ-୍ୖ-ୗଡ଼-ଢ଼ୟ-ୡ୦-ୱ];
+::NFD;
+::Oriya-InterIndic;
+::InterIndic-Telugu;
+::NFC;
+

package/transforms/Oriya-ur.json

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

package/transforms/Oriya-ur.txt

@@ -0,0 +1,7 @@
+
+::[ଁ-ଃଅ-ଌଏ-ଐଓ-ନପ-ରଲ-ଳଵ-ହ଼-ୃେ-ୈୋ-୍ୖ-ୗଡ଼-ଢ଼ୟ-ୡ୦-ୱ];
+::NFD;
+::Oriya-InterIndic;
+::InterIndic-ur;
+::NFC;
+

package/transforms/Pashto-Latin-BGN.json

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

package/transforms/Pashto-Latin-BGN.txt

@@ -0,0 +1,231 @@
+
+#
+########################################################################
+# BGN/PCGN 1968 System
+#
+# This system was adopted in 1968 for the romanization of Pashto
+# geographic names in Afghanistan. Persian names in Afghanistan are
+# romanized in accordance with the Romanization System for Persian
+# (BGN/PCGN 1958 System), shown on pages 87-92).
+#
+# Originally prepared by Michael Everson <everson@evertype.com>
+########################################################################
+#
+# MINIMAL FILTER: Pashto-Latin
+#
+:: [ءآابةتثجحخدذرزسشصضطظعغفقكلمنهوىئَُِّْ٠١٢٣٤٥٦٧٨٩ټپځڅچډړږژښگڰڼیۍې] ;
+:: NFD (NFC) ;
+#
+#
+########################################################################
+#
+########################################################################
+#
+# Define All Transformation Variables
+#
+########################################################################
+#
+$alef = ’;
+$ayin = ‘;
+$disambig =  ̱ ;
+#
+#
+# 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:]] ;
+#
+#
+########################################################################
+[: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
+٠ ↔ 0 $disambig ; # ARABIC-INDIC DIGIT ZERO
+١ ↔ 1 $disambig ; # ARABIC-INDIC DIGIT ONE
+٢ ↔ 2 $disambig ; # ARABIC-INDIC DIGIT TWO
+٣ ↔ 3 $disambig ; # ARABIC-INDIC DIGIT THREE
+٤ ↔ 4 $disambig ; # ARABIC-INDIC DIGIT FOUR
+٥ ↔ 5 $disambig ; # ARABIC-INDIC DIGIT FIVE
+٦ ↔ 6 $disambig ; # ARABIC-INDIC DIGIT SIX
+٧ ↔ 7 $disambig ; # ARABIC-INDIC DIGIT SEVEN
+٨ ↔ 8 $disambig ; # ARABIC-INDIC DIGIT EIGHT
+٩ ↔ 9 $disambig ; # ARABIC-INDIC DIGIT NINE
+۰ ↔ 0 ; # EXTENDED ARABIC-INDIC DIGIT ZERO
+۱ ↔ 1 ; # EXTENDED ARABIC-INDIC DIGIT ONE
+۲ ↔ 2 ; # EXTENDED ARABIC-INDIC DIGIT TWO
+۳ ↔ 3 ; # EXTENDED ARABIC-INDIC DIGIT THREE
+۴ ↔ 4 ; # EXTENDED ARABIC-INDIC DIGIT FOUR
+۵ ↔ 5 ; # EXTENDED ARABIC-INDIC DIGIT FIVE
+۶ ↔ 6 ; # EXTENDED ARABIC-INDIC DIGIT SIX
+۷ ↔ 7 ; # EXTENDED ARABIC-INDIC DIGIT SEVEN
+۸ ↔ 8 ; # EXTENDED ARABIC-INDIC DIGIT EIGHT
+۹ ↔ 9 ; # EXTENDED ARABIC-INDIC DIGIT NINE
+#
+########################################################################
+#
+# Rules moved to front to avoid masking
+#
+########################################################################
+#
+########################################################################
+#
+# BGN Page 89 Rule 4
+#
+# The character sequences كه , زه , سه , and گه may be romanized k·h, z·h,
+# s·h, and g·h in order to differentiate those romanizations from the
+# digraphs kh, zh, sh, and gh.
+#
+########################################################################
+#
+كه → k·h ; # ARABIC LETTER KAF + HEH
+زه → z·h ; # ARABIC LETTER ZAIN + HEH
+سه → s·h ; # ARABIC LETTER SEEN + HEH
+گه → g·h ; # ARABIC LETTER GAF + HEH
+#
+#
+########################################################################
+#
+# End Rule 4
+#
+########################################################################
+#
+########################################################################
+#
+# BGN Page 91 Rule 7
+#
+# Doubles consonant sounds are represented in Arabic script by
+# placing a shaddah ( ّ ) over a consonant character. In romanization
+# the letter should be doubled. [The remainder of this rule deals with
+# the definite article and is lexical.]
+#
+########################################################################
+#
+بّ → bb ; # ARABIC LETTER BEH + SHADDA
+پّ → pp ; # ARABIC LETTER PEH + SHADDA
+تّ → tt ; # ARABIC LETTER TEH + SHADDA
+ټّ → ṯṯ ; # ARABIC LETTER TEH WITH RING + SHADDA
+ثّ → s̄s̄ ; # ARABIC LETTER THEH + SHADDA
+جّ → jj ; # ARABIC LETTER JEEM + SHADDA
+چّ → chch ; # ARABIC LETTER TCHEH + SHADDA
+ّڅ → tsts ; # ARABIC LETTER HAH WITH THREE DOTS ABOVE + SHADDA
+ّځ → dz ; # ARABIC LETTER HAH WITH HAMZA ABOVE + SHADDA
+حّ → ḥḥ ; # ARABIC LETTER HAH + SHADDA
+خّ → khkh ; # ARABIC LETTER KHAH + SHADDA
+دّ → dd ; # ARABIC LETTER DAL + SHADDA
+ّډ → ḏḏ ; # ARABIC LETTER DAL WITH RING + SHADDA
+ذّ → z̄z̄ ; # ARABIC LETTER THAL + SHADDA
+رّ → rr ; # ARABIC LETTER REH + SHADDA
+ّړ → ṟṟ ; # ARABIC LETTER REH WITH RING + SHADDA
+زّ → zz ; # ARABIC LETTER ZAIN + SHADDA
+ژّ → zhzh ; # ARABIC LETTER JEH + SHADDA
+ّژ → z͟hz͟h ; # ARABIC LETTER REH WITH DOT BELOW AND DOT ABOVE + SHADDA
+سّ → ss ; # ARABIC LETTER SEEN + SHADDA
+شّ → shsh ; # ARABIC LETTER SHEEN + SHADDA
+ّښ → s͟hs͟h ; # ARABIC LETTER SEEN WITH DOT BELOW AND DOT ABOVE
+صّ → ṣṣ ; # ARABIC LETTER SAD + SHADDA
+ضّ → ḍḍ ; # ARABIC LETTER DAD + SHADDA
+طّ → ṭṭ ; # ARABIC LETTER TAH + SHADDA
+ظّ → ẓẓ ; # ARABIC LETTER ZAH + SHADDA
+عّ → $ayin $ayin ; # ARABIC LETTER AIN + SHADDA
+غّ → ghgh ; # ARABIC LETTER GHAIN + SHADDA
+فّ → ff ; # ARABIC LETTER FEH + SHADDA
+قّ → qq ; # ARABIC LETTER QAF + SHADDA
+كّ → kk ; # ARABIC LETTER KAF + SHADDA
+ّگ → gg ; # ARABIC LETTER GAF + SHADDA
+ّڰ → gg ; # ARABIC LETTER GAF WITH RING + SHADDA
+لّ → ll ; # ARABIC LETTER LAM + SHADDA
+مّ → mm ; # ARABIC LETTER MEEM + SHADDA
+نّ → nn ; # ARABIC LETTER NOON + SHADDA
+ّڼ → ṉṉ ; # ARABIC LETTER NOON WITH RING + SHADDA
+هّ → hh ; # ARABIC LETTER HEH + SHADDA
+وّ → ww ; # ARABIC LETTER WAW + SHADDA
+ّی → yy ; # ARABIC LETTER FARSI YEH + SHADDA
+ىّ → yy ; # ARABIC LETTER YEH + SHADDA
+#
+#
+########################################################################
+#
+# End Rule 7
+#
+########################################################################
+#
+########################################################################
+#
+# Start of Transformations
+#
+########################################################################
+#
+$wordBoundary{ء →  ; # ARABIC LETTER HAMZA
+ء → $alef ; # ARABIC LETTER HAMZA
+$wordBoundary{ا →  ; # ARABIC LETTER ALEF
+آ → $alef ā ; # ARABIC LETTER ALEF WITH MADDA ABOVE
+ب → b ; # ARABIC LETTER BEH
+پ → p ; # ARABIC LETTER PEH
+ت → t ; # ARABIC LETTER TEH
+ټ → ṯ ; # ARABIC LETTER TEH WITH RING
+ة → h ; # ARABIC LETTER TEH MARBUTA
+ث → s̄ ; # ARABIC LETTER THEH
+ج → j ; # ARABIC LETTER JEEM
+چ → ch ; # ARABIC LETTER TCHEH
+څ → ts ; # ARABIC LETTER HAH WITH THREE DOTS ABOVE
+ځ → dz ; # ARABIC LETTER HAH WITH HAMZA ABOVE
+ح → ḥ ; # ARABIC LETTER HAH
+خ → kh ; # ARABIC LETTER KHAH
+د → d ; # ARABIC LETTER DAL
+ډ → ḏ ; # ARABIC LETTER DAL WITH RING
+ذ → z̄ ; # ARABIC LETTER THAL
+ر → r ; # ARABIC LETTER REH
+ړ → ṟ ; # ARABIC LETTER REH WITH RING
+ز → z ; # ARABIC LETTER ZAIN
+ژ → zh ; # ARABIC LETTER JEH
+ږ → z͟h ; # ARABIC LETTER REH WITH DOT BELOW AND DOT ABOVE
+س → s ; # ARABIC LETTER SEEN
+ش → sh ; # ARABIC LETTER SHEEN
+ښ → s͟h ; # ARABIC LETTER SEEN WITH DOT BELOW AND DOT ABOVE
+ص → ṣ ; # ARABIC LETTER SAD
+ض → ẕ ; # ARABIC LETTER DAD
+ط → ṭ ; # ARABIC LETTER TAH
+ظ → ẓ ; # ARABIC LETTER ZAH
+ع → $ayin ; # ARABIC LETTER AIN
+غ → gh ; # ARABIC LETTER GHAIN
+ف → f ; # ARABIC LETTER FEH
+ق → q ; # ARABIC LETTER QAF
+ك → k ; # ARABIC LETTER KAF
+گ → g ; # ARABIC LETTER GAF
+ڰ → g ; # ARABIC LETTER GAF WITH RING
+ل → l ; # ARABIC LETTER LAM
+م → m ; # ARABIC LETTER MEEM
+ن → n ; # ARABIC LETTER NOON
+ڼ → ṉ ; # ARABIC LETTER NOON WITH RING
+و → w ; # ARABIC LETTER WAW
+ه → h ; # ARABIC LETTER HEH
+ٔی → ey ; # ARABIC LETTER FARSI YEH + HAMZA ABOVE
+ی → y ; # ARABIC LETTER FARSI YEH
+ى → y ; # ARABIC LETTER YEH
+ې → e ; # ARABIC LETTER E
+َا → ā ; # ARABIC FATHA + ALEF
+َىْ → ay ; # ARABIC FATHA + FARSI YEH + SUKUN
+َى → á ; # ARABIC FATHA + ALEF MAKSURA
+َْۍ → êy ; # ARABIC FATHA + SUKUN + YEH WITH TAIL
+َْ → ê ; # ARABIC FATHA + SUKUN
+َ → a ; # ARABIC FATHA
+ِْي → ey ; # ARABIC KASRA + FARSI YEH + SUKUN
+ِي → ī ; # ARABIC KASRA + FARSI YEH
+ِى → ī ; # ARABIC KASRA + YEH
+ِو → ew ; # ARABIC KASRA + WAW
+ِ → i ; # ARABIC KASRA
+ُوْ → ow ; # ARABIC DAMMA + WAW + SUKUN
+ُوی → ūy ; # ARABIC DAMMA + WAW + FARSI YEH
+ُو → ū ; # ARABIC DAMMA + WAW
+ُ → u ; # ARABIC DAMMA
+ْ →  ; # ARABIC SUKUN
+#
+#
+########################################################################
+

package/transforms/Persian-Latin-BGN.json

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

package/transforms/Persian-Latin-BGN.txt

@@ -0,0 +1,203 @@
+
+#
+########################################################################
+# BGN/PCGN 1956 System
+#
+# This system was adopted by the BGN in 1946 and by the PCGN in 1958.
+# It is used for the romanization of geographic names in Iran and
+# for Persian-language names in Afghanistan.
+#
+# Originally prepared by Michael Everson <everson@evertype.com>
+########################################################################
+#
+# MINIMAL FILTER: Persian-Latin
+#
+:: [[:arabic:][:block=ARABIC:][ءآابةتثجحخدذرزسشصضطظعغفقكلمنهویيَُِّْ٠١٢٣٤٥٦٧٨٩پچژگی]] ;
+:: NFKD (NFC) ;
+#
+#
+########################################################################
+#
+########################################################################
+#
+# Define All Transformation Variables
+#
+########################################################################
+#
+$alef = ’;
+$ayin = ‘;
+$disambig =  ̱ ;
+#
+#
+# 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:]] ;
+#
+#
+########################################################################
+# 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
+٠ ↔ 0 $disambig ; # ARABIC-INDIC DIGIT ZERO
+١ ↔ 1 $disambig ; # ARABIC-INDIC DIGIT ONE
+٢ ↔ 2 $disambig ; # ARABIC-INDIC DIGIT TWO
+٣ ↔ 3 $disambig ; # ARABIC-INDIC DIGIT THREE
+٤ ↔ 4 $disambig ; # ARABIC-INDIC DIGIT FOUR
+٥ ↔ 5 $disambig ; # ARABIC-INDIC DIGIT FIVE
+٦ ↔ 6 $disambig ; # ARABIC-INDIC DIGIT SIX
+٧ ↔ 7 $disambig ; # ARABIC-INDIC DIGIT SEVEN
+٨ ↔ 8 $disambig ; # ARABIC-INDIC DIGIT EIGHT
+٩ ↔ 9 $disambig ; # ARABIC-INDIC DIGIT NINE
+۰ ↔ 0 ; # EXTENDED ARABIC-INDIC DIGIT ZERO
+۱ ↔ 1 ; # EXTENDED ARABIC-INDIC DIGIT ONE
+۲ ↔ 2 ; # EXTENDED ARABIC-INDIC DIGIT TWO
+۳ ↔ 3 ; # EXTENDED ARABIC-INDIC DIGIT THREE
+۴ ↔ 4 ; # EXTENDED ARABIC-INDIC DIGIT FOUR
+۵ ↔ 5 ; # EXTENDED ARABIC-INDIC DIGIT FIVE
+۶ ↔ 6 ; # EXTENDED ARABIC-INDIC DIGIT SIX
+۷ ↔ 7 ; # EXTENDED ARABIC-INDIC DIGIT SEVEN
+۸ ↔ 8 ; # EXTENDED ARABIC-INDIC DIGIT EIGHT
+۹ ↔ 9 ; # EXTENDED ARABIC-INDIC DIGIT NINE
+#
+########################################################################
+#
+# Rules moved to front to avoid masking
+#
+########################################################################
+#
+########################################################################
+#
+# BGN Page 89 Rule 4
+#
+# The character sequences كه , زه , سه , and گه may be romanized k·h, z·h,
+# s·h, and g·h in order to differentiate those romanizations from the
+# digraphs kh, zh, sh, and gh.
+#
+########################################################################
+#
+كه → k·h ; # ARABIC LETTER KAF + HEH
+زه → z·h ; # ARABIC LETTER ZAIN + HEH
+سه → s·h ; # ARABIC LETTER SEEN + HEH
+گه → g·h ; # ARABIC LETTER GAF + HEH
+#
+#
+########################################################################
+#
+# End Rule 4
+#
+########################################################################
+#
+########################################################################
+#
+# BGN Page 91 Rule 7
+#
+# Doubles consonant sounds are represented in Arabic script by
+# placing a shaddah ( ّ ) over a consonant character. In romanization
+# the letter should be doubled. [The remainder of this rule deals with
+# the definite article and is lexical.]
+#
+########################################################################
+#
+بّ → bb ; # ARABIC LETTER BEH + SHADDA
+پّ → pp ; # ARABIC LETTER PEH + SHADDA
+تّ → tt ; # ARABIC LETTER TEH + SHADDA
+ثّ → s̄s̄ ; # ARABIC LETTER THEH + SHADDA
+جّ → jj ; # ARABIC LETTER JEEM + SHADDA
+چّ → chch ; # ARABIC LETTER TCHEH + SHADDA
+حّ → ḥḥ ; # ARABIC LETTER HAH + SHADDA
+خّ → khkh ; # ARABIC LETTER KHAH + SHADDA
+دّ → dd ; # ARABIC LETTER DAL + SHADDA
+ذّ → z̄z̄ ; # ARABIC LETTER THAL + SHADDA
+رّ → rr ; # ARABIC LETTER REH + SHADDA
+زّ → zz ; # ARABIC LETTER ZAIN + SHADDA
+ژّ → zhzh ; # ARABIC LETTER JEH + SHADDA
+سّ → ss ; # ARABIC LETTER SEEN + SHADDA
+شّ → shsh ; # ARABIC LETTER SHEEN + SHADDA
+صّ → ṣṣ ; # ARABIC LETTER SAD + SHADDA
+ضّ → ḍḍ ; # ARABIC LETTER DAD + SHADDA
+طّ → ṭṭ ; # ARABIC LETTER TAH + SHADDA
+ظّ → ẓẓ ; # ARABIC LETTER ZAH + SHADDA
+عّ → $ayin $ayin ; # ARABIC LETTER AIN + SHADDA
+غّ → ghgh ; # ARABIC LETTER GHAIN + SHADDA
+فّ → ff ; # ARABIC LETTER FEH + SHADDA
+قّ → qq ; # ARABIC LETTER QAF + SHADDA
+كّ → kk ; # ARABIC LETTER KAF + SHADDA
+لّ → ll ; # ARABIC LETTER LAM + SHADDA
+مّ → mm ; # ARABIC LETTER MEEM + SHADDA
+نّ → nn ; # ARABIC LETTER NOON + SHADDA
+هّ → hh ; # ARABIC LETTER HEH + SHADDA
+وّ → ww ; # ARABIC LETTER WAW + SHADDA
+یّ → yy ; # ARABIC LETTER FARSI YEH + SHADDA
+#
+#
+########################################################################
+#
+# End Rule 7
+#
+########################################################################
+#
+########################################################################
+#
+# Start of Transformations
+#
+########################################################################
+#
+$wordBoundary{ء →  ; # ARABIC LETTER HAMZA
+ء → $alef ; # ARABIC LETTER HAMZA
+$wordBoundary{ا →  ; # ARABIC LETTER ALEF
+آ → $alef ā ; # ARABIC FATHA ALEF WITH MADDA ABOVE
+ب → b ; # ARABIC LETTER BEH
+پ → p ; # ARABIC LETTER PEH
+ت → t ; # ARABIC LETTER TEH
+ة → h ; # ARABIC LETTER TEH MARBUTA
+ث → s̄ ; # ARABIC LETTER THEH
+ج → j ; # ARABIC LETTER JEEM
+چ → ch ; # ARABIC LETTER TCHEH
+ح → ḥ ; # ARABIC LETTER HAH
+خ → kh ; # ARABIC LETTER KHAH
+د → d ; # ARABIC LETTER DAL
+ذ → z̄ ; # ARABIC LETTER THAL
+ر → r ; # ARABIC LETTER REH
+ز → z ; # ARABIC LETTER ZAIN
+ژ → zh ; # ARABIC LETTER JEH
+س → s ; # ARABIC LETTER SEEN
+ش → sh ; # ARABIC LETTER SHEEN
+ص → ṣ ; # ARABIC LETTER SAD
+ض → ẕ ; # ARABIC LETTER DAD
+ط → ṭ ; # ARABIC LETTER TAH
+ظ → ẓ ; # ARABIC LETTER ZAH
+ع → $ayin ; # ARABIC LETTER AIN
+غ → gh ; # ARABIC LETTER GHAIN
+ف → f ; # ARABIC LETTER FEH
+ق → q ; # ARABIC LETTER QAF
+ک ↔ k ; # ARABIC LETTER KEHEH
+ك ↔ k $disambig ; # ARABIC LETTER KAF
+گ → g ; # ARABIC LETTER GAF
+ل → l ; # ARABIC LETTER LAM
+م → m ; # ARABIC LETTER MEEM
+ن → n ; # ARABIC LETTER NOON
+ه → h ; # ARABIC LETTER HEH
+و → v ; # ARABIC LETTER WAW
+ی → y ; # ARABIC LETTER FARSI YEH
+َا → ā ; # ARABIC FATHA + ALEF
+َی → á ; # ARABIC FATHA + FARSI YEH
+َوْ → ow ; # ARABIC FATHA + WAW + SUKUN
+َ → a ; # ARABIC FATHA
+ِي → ī ; # ARABIC KASRA + YEH
+ِ → e ; # ARABIC KASRA
+ُو → ū ; # ARABIC DAMMA + WAW
+ُ → o ; # ARABIC DAMMA
+ْ →  ; # ARABIC SUKUN
+::NFC (NFD) ;
+#
+#
+########################################################################
+

package/transforms/Pinyin-NumericPinyin.json

@@ -0,0 +1,7 @@
+{
+  "_visibility": "external",
+  "_source": "Pinyin",
+  "_direction": "both",
+  "_target": "NumericPinyin",
+  "_rulesFile": "Pinyin-NumericPinyin.txt"
+}

package/transforms/Pinyin-NumericPinyin.txt

@@ -0,0 +1,8 @@
+
+# Only intended for internal use
+̄ ↔ 1;
+́ ↔ 2;
+̌ ↔ 3;
+̀ ↔ 4;
+← 5;
+