blob: 3b8cdcf0cd47799b10d5b8794beb1b49a462f8d3 [file] [log] [blame]
#--------------------------------------------------------------------
# Copyright (c) 1999-2004, International Business Machines
# Corporation and others. All Rights Reserved.
#--------------------------------------------------------------------
# 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:] [‎ⁿ\u060C\u061B\u061F\u0640\u064B-\u0655\u0660-\u066C\u06F0-\u06F9\uFDFC]] ;
:: NFKD (NFC);
$disambig = ̱ ;
$disambig2 = ̰ ;
$under = ̣ ;
$notAbove = [[:^ccc=0:]&[:^ccc=230:]];
# non-letters
٫ <> '.' $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
# 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 \u0308 ; # ARABIC LETTER TEH MARBUTA
ة | $1 < t ($notAbove+) \u0308 ; # 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
# 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
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
| s < c } [eiy];
| k < c ;
| i < e ;
| u < o ;
| ks < x ;
| n < ‎ⁿ;
:: (lower) ;
::NFC (NFD);
:: ( [[:Latin:] [%,.0-9;?\u02BE-\u02BF\u0302-\u0304\u0308-\u030A\u0323\u0330-\u0331\u0339\u037E]] );