icu4c/source/data/translit/ps_ps_Latn_BGN.txt

# © 2016 and later: Unicode, Inc. and others.
# License & terms of use: http://www.unicode.org/copyright.html
# Generated using tools/cldr/cldr-to-icu/build-icu-data.xml
#
# File: ps_ps_Latn_BGN.txt
# Generated from CLDR
#

#
########################################################################
# 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
#
:: [ءآابةتثجحخدذرزسشصضطظعغفقكلمنهوىي\u064E\u064F\u0650\u0651\u0652\u0654٠١٢٣٤٥٦٧٨٩ټپځڅچډړږژښگڰڼیۍې] ;
:: NFD (NFC) ;
#
#
########################################################################
#
########################################################################
#
# Define All Transformation Variables
#
########################################################################
#
$alef = ’;
$ayin = ‘;
$disambig =  \u0331 ;
#
#
# 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 ( \u0651 ) over a consonant character. In romanization
# the letter should be doubled. [The remainder of this rule deals with
# the definite article and is lexical.]
#
########################################################################
#
ب\u0651 → bb ; # ARABIC LETTER BEH + SHADDA
پ\u0651 → pp ; # ARABIC LETTER PEH + SHADDA
ت\u0651 → tt ; # ARABIC LETTER TEH + SHADDA
ټ\u0651 → ṯṯ ; # ARABIC LETTER TEH WITH RING + SHADDA
ث\u0651 → s\u0304s\u0304 ; # ARABIC LETTER THEH + SHADDA
ج\u0651 → jj ; # ARABIC LETTER JEEM + SHADDA
چ\u0651 → chch ; # ARABIC LETTER TCHEH + SHADDA
\u0651څ → tsts ; # ARABIC LETTER HAH WITH THREE DOTS ABOVE + SHADDA
\u0651ځ → dz ; # ARABIC LETTER HAH WITH HAMZA ABOVE + SHADDA
ح\u0651 → ḥḥ ; # ARABIC LETTER HAH + SHADDA
خ\u0651 → khkh ; # ARABIC LETTER KHAH + SHADDA
د\u0651 → dd ; # ARABIC LETTER DAL + SHADDA
\u0651ډ → ḏḏ ; # ARABIC LETTER DAL WITH RING + SHADDA
ذ\u0651 → z\u0304z\u0304 ; # ARABIC LETTER THAL + SHADDA
ر\u0651 → rr ; # ARABIC LETTER REH + SHADDA
\u0651ړ → ṟṟ ; # ARABIC LETTER REH WITH RING + SHADDA
ز\u0651 → zz ; # ARABIC LETTER ZAIN + SHADDA
ژ\u0651 → zhzh ; # ARABIC LETTER JEH + SHADDA
\u0651ژ → z\u035Fhz\u035Fh ; # ARABIC LETTER REH WITH DOT BELOW AND DOT ABOVE + SHADDA
س\u0651 → ss ; # ARABIC LETTER SEEN + SHADDA
ش\u0651 → shsh ; # ARABIC LETTER SHEEN + SHADDA
\u0651ښ → s\u035Fhs\u035Fh ; # ARABIC LETTER SEEN WITH DOT BELOW AND DOT ABOVE
ص\u0651 → ṣṣ ; # ARABIC LETTER SAD + SHADDA
ض\u0651 → ḍḍ ; # ARABIC LETTER DAD + SHADDA
ط\u0651 → ṭṭ ; # ARABIC LETTER TAH + SHADDA
ظ\u0651 → ẓẓ ; # ARABIC LETTER ZAH + SHADDA
ع\u0651 → $ayin $ayin ; # ARABIC LETTER AIN + SHADDA
غ\u0651 → ghgh ; # ARABIC LETTER GHAIN + SHADDA
ف\u0651 → ff ; # ARABIC LETTER FEH + SHADDA
ق\u0651 → qq ; # ARABIC LETTER QAF + SHADDA
ك\u0651 → kk ; # ARABIC LETTER KAF + SHADDA
\u0651گ → gg ; # ARABIC LETTER GAF + SHADDA
\u0651ڰ → gg ; # ARABIC LETTER GAF WITH RING + SHADDA
ل\u0651 → ll ; # ARABIC LETTER LAM + SHADDA
م\u0651 → mm ; # ARABIC LETTER MEEM + SHADDA
ن\u0651 → nn ; # ARABIC LETTER NOON + SHADDA
\u0651ڼ → ṉṉ ; # ARABIC LETTER NOON WITH RING + SHADDA
ه\u0651 → hh ; # ARABIC LETTER HEH + SHADDA
و\u0651 → ww ; # ARABIC LETTER WAW + SHADDA
\u0651ی → yy ; # ARABIC LETTER FARSI YEH + SHADDA
ى\u0651 → 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 → s\u0304 ; # 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 → z\u0304 ; # 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 → z\u035Fh ; # ARABIC LETTER REH WITH DOT BELOW AND DOT ABOVE
س → s ; # ARABIC LETTER SEEN
ش → sh ; # ARABIC LETTER SHEEN
ښ → s\u035Fh ; # 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
\u0654ی → ey ; # ARABIC LETTER FARSI YEH + HAMZA ABOVE
ی → y ; # ARABIC LETTER FARSI YEH
ى → y ; # ARABIC LETTER YEH
ې → e ; # ARABIC LETTER E
\u064Eا → ā ; # ARABIC FATHA + ALEF
\u064Eى\u0652 → ay ; # ARABIC FATHA + FARSI YEH + SUKUN
\u064Eى → á ; # ARABIC FATHA + ALEF MAKSURA
\u064E\u0652ۍ → êy ; # ARABIC FATHA + SUKUN + YEH WITH TAIL
\u064E\u0652 → ê ; # ARABIC FATHA + SUKUN
\u064E → a ; # ARABIC FATHA
\u0650\u0652ي → ey ; # ARABIC KASRA + FARSI YEH + SUKUN
\u0650ي → ī ; # ARABIC KASRA + FARSI YEH
\u0650ى → ī ; # ARABIC KASRA + YEH
\u0650و → ew ; # ARABIC KASRA + WAW
\u0650 → i ; # ARABIC KASRA
\u064Fو\u0652 → ow ; # ARABIC DAMMA + WAW + SUKUN
\u064Fوی → ūy ; # ARABIC DAMMA + WAW + FARSI YEH
\u064Fو → ū ; # ARABIC DAMMA + WAW
\u064F → u ; # ARABIC DAMMA
\u0652 →  ; # ARABIC SUKUN
#
#
########################################################################