212 lines
7.6 KiB
Plaintext
212 lines
7.6 KiB
Plaintext
# © 2016 and later: Unicode, Inc. and others.
|
||
# License & terms of use: http://www.unicode.org/copyright.html#License
|
||
#
|
||
# File: ar_ar_Latn_BGN.txt
|
||
# Generated from CLDR
|
||
#
|
||
|
||
#
|
||
########################################################################
|
||
# BGN/PCGN 1956 System
|
||
#
|
||
# This system was adopted by the BGN in 1946 and by the PCGN
|
||
# in 1956 and has been applied in the systematic romanization
|
||
# of geographic names in Bahrain, Egypt, Iraq, Jordan,
|
||
# Kuwait, Lebanon, Libya, Oman, Qatar, Saudi Arabia, Sudan,
|
||
# Syria, Tunisia, the United Arab Emirates, and Yemen, all
|
||
# of which has been covered by published BGN engineers.
|
||
#
|
||
# Originally prepared by Michael Everson <everson@evertype.com>
|
||
########################################################################
|
||
#
|
||
# MINIMAL FILTER: Arabic-Latin
|
||
#
|
||
:: [[:arabic:][:block=ARABIC:][ءآابةتثجحخدذرزسشصضطظعغفقكلمنهوىي\u064B\u064C\u064D\u064E\u064F\u0650\u0651\u0652٠١٢٣٤٥٦٧٨٩ٱ]] ;
|
||
:: NFKD (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:]] ;
|
||
#
|
||
#
|
||
########################################################################
|
||
# 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
|
||
#
|
||
########################################################################
|
||
#
|
||
# Rules moved to front to avoid masking
|
||
#
|
||
########################################################################
|
||
#
|
||
########################################################################
|
||
#
|
||
# BGN Page 8 Rule 5
|
||
#
|
||
# The character sequences ت , كه , ته , and سه may be romanized t·h, k·h,
|
||
# d·h, and s·h in order to differentiate those romanizations from the
|
||
# digraphs th, kh, dh, and sh.
|
||
#
|
||
########################################################################
|
||
#
|
||
ته → t·h ; # ARABIC LETTER TEH + HEH
|
||
كه → k·h ; # ARABIC LETTER KAF + HEH
|
||
ده → d·h ; # ARABIC LETTER DAL + HEH
|
||
سه → s·h ; # ARABIC LETTER SEEN + HEH
|
||
#
|
||
#
|
||
########################################################################
|
||
#
|
||
# End Rule 5
|
||
#
|
||
########################################################################
|
||
########################################################################
|
||
#
|
||
#
|
||
# BGN Page 8 Rule 9
|
||
#
|
||
# 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
|
||
ت\u0651 → tt ; # ARABIC LETTER TEH + SHADDA
|
||
ث\u0651 → thth ; # ARABIC LETTER THEH + SHADDA
|
||
ج\u0651 → jj ; # ARABIC LETTER JEEM + SHADDA
|
||
ح\u0651 → ḥḥ ; # ARABIC LETTER HAH + SHADDA
|
||
خ\u0651 → khkh ; # ARABIC LETTER KHAH + SHADDA
|
||
د\u0651 → dd ; # ARABIC LETTER DAL + SHADDA
|
||
ذ\u0651 → dhdh ; # ARABIC LETTER THAL + SHADDA
|
||
ر\u0651 → rr ; # ARABIC LETTER REH + SHADDA
|
||
ز\u0651 → zz ; # ARABIC LETTER ZAIN + SHADDA
|
||
س\u0651 → ss ; # ARABIC LETTER SEEN + SHADDA
|
||
ش\u0651 → shsh ; # ARABIC LETTER SHEEN + SHADDA
|
||
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
|
||
ل\u0651 → ll ; # ARABIC LETTER LAM + SHADDA
|
||
م\u0651 → mm ; # ARABIC LETTER MEEM + SHADDA
|
||
ن\u0651 → nn ; # ARABIC LETTER NOON + SHADDA
|
||
ه\u0651 → hh ; # ARABIC LETTER HEH + SHADDA
|
||
و\u0651 → ww ; # ARABIC LETTER WAW + SHADDA
|
||
ى\u0651 → yy ; # ARABIC LETTER YEH + SHADDA
|
||
#
|
||
#
|
||
########################################################################
|
||
#
|
||
# End Rule 9
|
||
#
|
||
########################################################################
|
||
#
|
||
########################################################################
|
||
#
|
||
# Start of Transformations
|
||
#
|
||
########################################################################
|
||
#
|
||
$wordBoundary{ء → ; # ARABIC LETTER HAMZA
|
||
ء → $alef ; # ARABIC LETTER HAMZA
|
||
$wordBoundary{ا → ; # ARABIC LETTER ALEF
|
||
ٱ → $alef ; # ARABIC LETTER ALEF WASLA
|
||
$wordBoundary{آ → ā ; # ARABIC LETTER ALEF WITH MADDA ABOVE
|
||
آ → $alef ā ; # ARABIC LETTER ALEF WITH MADDA ABOVE
|
||
ب → b ; # ARABIC LETTER BEH
|
||
ت → t ; # ARABIC LETTER TEH
|
||
ة → h ; # ARABIC LETTER TEH MARBUTA
|
||
ث → th ; # ARABIC LETTER THEH
|
||
ج → j ; # ARABIC LETTER JEEM
|
||
ح → ḩ ; # ARABIC LETTER HAH
|
||
خ → kh ; # ARABIC LETTER KHAH
|
||
د → d ; # ARABIC LETTER DAL
|
||
ذ → dh ; # ARABIC LETTER THAL
|
||
ر → r ; # ARABIC LETTER REH
|
||
ز → z ; # ARABIC LETTER ZAIN
|
||
س → s ; # ARABIC LETTER SEEN
|
||
ش → sh ; # ARABIC LETTER SHEEN
|
||
ص → ş ; # ARABIC LETTER SAD
|
||
ض → ḑ ; # ARABIC LETTER DAD
|
||
ط → ţ ; # ARABIC LETTER TAH
|
||
ظ → z\u0327 ; # 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 $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 ; # ARABIC LETTER YEH
|
||
\u064Eا → ā ; # ARABIC FATHA + ALEF
|
||
\u064Eى → á ; # ARABIC FATHA + ALEF MAKSURA
|
||
\u064Eي\u0652 → ay ; # ARABIC FATHA + YEH + SUKUN
|
||
\u064Eو\u0652 → aw ; # ARABIC FATHA + WAW + SUKUN
|
||
\u064E → a ; # ARABIC FATHA
|
||
\u0650ي → ī ; # ARABIC KASRA + YEH
|
||
\u0650 → i ; # ARABIC KASRA
|
||
\u064Fو → ū ; # ARABIC DAMMA + WAW
|
||
\u064F → u ; # ARABIC DAMMA
|
||
\u0652 → ; # ARABIC SUKUN
|
||
\u064B → aⁿ ; # ARABIC FATHATAN
|
||
\u064D → iⁿ ; # ARABIC KASRATAN
|
||
\u064C → uⁿ ; # ARABIC DAMMATAN
|
||
::NFC (NFD) ;
|
||
#
|
||
#
|
||
########################################################################
|
||
|