blob: 8cf10e0901fba8b1697b72c7139595701ed6c76b [file] [log] [blame]
#--------------------------------------------------------------------
# Copyright (c) 1999-2001, International Business Machines
# Corporation and others. All Rights Reserved.
#--------------------------------------------------------------------
# $Source: /xsrl/Nsvn/icu/icu4j/src/com/ibm/icu/impl/data/Attic/Transliterator_ThaiLogical_Latin.txt,v $
# $Date: 2002/07/15 01:26:18 $
# $Revision: 1.3 $
#--------------------------------------------------------------------
# Thai-Latin
# This set of rules follows ISO 11940
# see http://homepage.mac.com/sirbinks/pdf/Thai.r2.pdf
# except that that does not mention an implicit vowel, so we use ọ
#
# The transcription is fairly ugly, so we ought to also do the UNGEGN version
# see: http://www.eki.ee/wgrs/rom1_th.pdf
# and probably make that the main variant.
# Note: this is an internal file. The NFD/NFC is handled externally, in the index
# The insertion of spaces between words, the reversal of the vowels
# and the conversion of space to semicolon are done *outside* of these rules.
# So as far as these rules are concerned, the vowels are in logical order!
# insert implicit vowel (and remove it going the other way)
$consonant = [ก-ฮ];
$vowel = [ะ-ฺเ-ไ็];
{ ( $consonant ) } [^$vowel ] > | $1  ;
 > ọ ;
< ọ ;
# Consonants
ข <> k̄h ; # THAI CHARACTER KHO KHAI
ฃ <> ḳ̄h ; # THAI CHARACTER KHO KHUAT
ค <> kʹh ; # THAI CHARACTER KHO KHWAI
ฅ <> ḳh ; # THAI CHARACTER KHO KHON
ฆ <> kh ; # THAI CHARACTER KHO RAKHANG
ง <> ng ; # THAI CHARACTER NGO NGU
ฉ <> c̄h ; # THAI CHARACTER CHO CHING
ช <> ch ; # THAI CHARACTER CHO CHANG
ฌ <> c̣h ; # THAI CHARACTER CHO CHOE
ญ <> ỵ ; # THAI CHARACTER YO YING
ฎ <> ḍ ; # THAI CHARACTER DO CHADA
ฐ <> ṭ̄h ; # THAI CHARACTER THO THAN
ฑ <> ṯh ; # THAI CHARACTER THO NANGMONTHO
ฒ <> tʹh ; # THAI CHARACTER THO PHUTHAO
ณ <> ṇ ; # THAI CHARACTER NO NEN
ด <> d ; # THAI CHARACTER DO DEK
ถ <> t̄h ; # THAI CHARACTER THO THUNG
ธ <> ṭh ; # THAI CHARACTER THO THONG
น <> n ; # THAI CHARACTER NO NU
บ <> b ; # THAI CHARACTER BO BAIMAI
ผ <> p̄h ; # THAI CHARACTER PHO PHUNG
ฝ <> f̄ ; # THAI CHARACTER FO FA
ภ <> p̣h ; # THAI CHARACTER PHO SAMPHAO
ม <> m ; # THAI CHARACTER MO MA
ย <> y ; # THAI CHARACTER YO YAK
ร <> r ; # THAI CHARACTER RO RUA
ฤ <> v ; # THAI CHARACTER RU
ฦ <> ł ; # THAI CHARACTER LU
ว <> w ; # THAI CHARACTER WO WAEN
ศ <> ṣ̄ ; # THAI CHARACTER SO SALA
ษ <> s̄ʹ ; # THAI CHARACTER SO RUSI
ส <> s̄ ; # THAI CHARACTER SO SUA
ห <> h̄ ; # THAI CHARACTER HO HIP
ฬ <> ḷ ; # THAI CHARACTER LO CHULA
# items moved down so that longer ones come first
ฏ <> ṭ ; # THAI CHARACTER TO PATAK
ท <> th ; # THAI CHARACTER THO THAHAN
พ <> ph ; # THAI CHARACTER PHO PHAN
ล <> l ; # THAI CHARACTER LO LING
ป <> p ; # THAI CHARACTER PO PLA
ฟ <> f ; # THAI CHARACTER FO FAN
ต <> t ; # THAI CHARACTER TO TAO
จ <> c ; # THAI CHARACTER CHO CHAN
ก <> k ; # THAI CHARACTER KO KAI
อ <> x ; # THAI CHARACTER O ANG
ซ <> s ; # THAI CHARACTER SO SO
ฮ <> ḥ ; # THAI CHARACTER HO NOKHUK
# vowels
ั <> ạ ; # THAI CHARACTER MAI HAN-AKAT
า <> ā ; # THAI CHARACTER SARA AA
ำ <> å ; # THAI CHARACTER SARA AM
ะ <> a ; # THAI CHARACTER SARA A
ี <> ī ; # THAI CHARACTER SARA II
ื <> ụ̄ ; # THAI CHARACTER SARA UEE
ึ <> ụ ; # THAI CHARACTER SARA UE
ู <> ū ; # THAI CHARACTER SARA UU
ุ <> u ; # THAI CHARACTER SARA U
ฯ <> ‡ ; # THAI CHARACTER PAIYANNOI
# ฿ <> XXX ; # THAI CURRENCY SYMBOL BAHT
เ <> e ; # THAI CHARACTER SARA E
แ <> æ ; # THAI CHARACTER SARA AE
โ <> o ; # THAI CHARACTER SARA O
ใ <> ı ; # THAI CHARACTER SARA AI MAIMUAN
ไ <> ị ; # THAI CHARACTER SARA AI MAIMALAI
ๅ <> ɨ ; # THAI CHARACTER LAKKHANGYAO
็ <> ̆ ; # THAI CHARACTER MAITAIKHU
่ <> ̀ ; # THAI CHARACTER MAI EK
้ <> ̂ ; # THAI CHARACTER MAI THO
๊ <> ́ ; # THAI CHARACTER MAI TRI
๋ <> ̌ ; # THAI CHARACTER MAI CHATTAWA
์ <> ̒ ; # THAI CHARACTER THANTHAKHAT
ํ <> ̥̊ ; # THAI CHARACTER NIKHAHIT
๎ <> '~' ; # THAI CHARACTER YAMAKKAN
๏ <> § ; # THAI CHARACTER FONGMAN
๐ <> 0 ; # THAI DIGIT ZERO
๑ <> 1 ; # THAI DIGIT ONE
๒ <> 2 ; # THAI DIGIT TWO
๓ <> 3 ; # THAI DIGIT THREE
๔ <> 4 ; # THAI DIGIT FOUR
๕ <> 5 ; # THAI DIGIT FIVE
๖ <> 6 ; # THAI DIGIT SIX
๗ <> 7 ; # THAI DIGIT SEVEN
๘ <> 8 ; # THAI DIGIT EIGHT
๙ <> 9 ; # THAI DIGIT NINE
๚ <> '||' ; # THAI CHARACTER ANGKHANKHU
๛ <> » ; # THAI CHARACTER KHOMUT
ๆ <> « ; # THAI CHARACTER MAIYAMOK
# moved down to make shorter first
ฺ <>̥ ; # THAI CHARACTER PHINTHU
ิ <> i ; # THAI CHARACTER SARA I
# fallbacks
| k < g ;
| k < h ;
| c < j ;
| k < q ;
| s < z ;
:: (lower);