code.delx.au - gnu-emacs/blob - lisp/language/tibetan.el

   1 ;;; tibetan.el --- support for Tibetan language -*- coding: iso-2022-7bit; -*-
   2
   3 ;; Copyright (C) 1997, 2001, 2002, 2003, 2004, 2005, 2006, 2007, 2008,
   4 ;;   2009, 2010, 2011  Free Software Foundation, Inc.
   5 ;; Copyright (C) 1997, 1998, 1999, 2000, 2001, 2002, 2003, 2004, 2005,
   6 ;;   2006, 2007, 2008, 2009, 2010, 2011
   7 ;;   National Institute of Advanced Industrial Science and Technology (AIST)
   8 ;;   Registration Number H14PRO021
   9 ;; Copyright (C) 2003
  10 ;;   National Institute of Advanced Industrial Science and Technology (AIST)
  11 ;;   Registration Number H13PRO009
  12
  13 ;; Author: Toru TOMABECHI <Toru.Tomabechi@orient.unil.ch>
  14 ;; Created: Feb. 17. 1997
  15 ;; Keywords: multilingual, Tibetan, i18n
  16
  17 ;; This file is part of GNU Emacs.
  18
  19 ;; GNU Emacs is free software: you can redistribute it and/or modify
  20 ;; it under the terms of the GNU General Public License as published by
  21 ;; the Free Software Foundation, either version 3 of the License, or
  22 ;; (at your option) any later version.
  23
  24 ;; GNU Emacs is distributed in the hope that it will be useful,
  25 ;; but WITHOUT ANY WARRANTY; without even the implied warranty of
  26 ;; MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
  27 ;; GNU General Public License for more details.
  28
  29 ;; You should have received a copy of the GNU General Public License
  30 ;; along with GNU Emacs.  If not, see <http://www.gnu.org/licenses/>.
  31
  32 ;;; History:
  33
  34 ;; 1997.03.13 Modification for special signs and punctuations.
  35
  36 ;;; Commentary:
  37
  38 ;;; Code:
  39
  40 ;;; Tibetan Character set.
  41 ;;; \x2130 -- \x234a is a subset of Unicode v.2 \x0f00 - \x0fb9
  42 ;;; with a slight modification. And there are some subjoined
  43 ;;; consonants which are not specified in Unicode.
  44 ;;; I hope I can add missing characters later.
  45 ;;;
  46 ;;;     00 01 02 03 04 05 06 07 08 09 0A 0B 0C 0D 0E 0F
  47 ;;;2120 // \e$(7!!\e(B \e$(7!"\e(B \e$(7!#\e(B \e$(7!$\e(B \e$(7!%\e(B \e$(7!&\e(B \e$(7!'\e(B \e$(7!(\e(B \e$(7!)\e(B \e$(7!*\e(B \e$(7!+\e(B \e$(7!,\e(B \e$(7!-\e(B \e$(7!.\e(B \e$(7!/\e(B ; obsolete glyphs (2123-5)
  48 ;;;2130 \e$(7!0\e(B \e$(7!1\e(B \e$(7!2\e(B \e$(7!3\e(B \e$(7!4\e(B \e$(7!5\e(B \e$(7!6\e(B \e$(7!7\e(B \e$(7!8\e(B \e$(7!9\e(B \e$(7!:\e(B \e$(7!;\e(B \e$(7!<\e(B \e$(7!=\e(B \e$(7!>\e(B \e$(7!?\e(B ; Punctuations,
  49 ;;;2140 \e$(7!@\e(B \e$(7!A\e(B \e$(7!B\e(B \e$(7!C\e(B \e$(7!D\e(B \e$(7!E\e(B \e$(7!F\e(B \e$(7!G\e(B \e$(7!H\e(B \e$(7!I\e(B \e$(7!J\e(B \e$(7!K\e(B \e$(7!L\e(B \e$(7!M\e(B \e$(7!N\e(B \e$(7!O\e(B ; Digits and
  50 ;;;2150 \e$(7!P\e(B \e$(7!Q\e(B \e$(7!R\e(B \e$(7!S\e(B \e$(7!T\e(B \e$(7!U\e(B \e$(7!V\e(B \e$(7!W\e(B \e$(7!X\e(B \e$(7!Y\e(B \e$(7!Z\e(B \e$(7![\e(B \e$(7!\\e(B \e$(7!]\e(B \e$(7!^\e(B \e$(7!_\e(B ; Special signs.
  51 ;;;2160 \e$(7!`\e(B \e$(7!a\e(B \e$(7!b\e(B \e$(7!c\e(B \e$(7!d\e(B \e$(7!e\e(B \e$(7!f\e(B \e$(7!g\e(B \e$(7!h\e(B \e$(7!i\e(B \e$(7!j\e(B \e$(7!k\e(B \e$(7!l\e(B \e$(7!m\e(B \e$(7!n\e(B \e$(7!o\e(B ;
  52 ;;;2170 \e$(7!p\e(B \e$(7!q\e(B \e$(7!r\e(B \e$(7!s\e(B \e$(7!t\e(B \e$(7!u\e(B \e$(7!v\e(B \e$(7!w\e(B \e$(7!x\e(B \e$(7!y\e(B \e$(7!z\e(B \e$(7!{\e(B \e$(7!|\e(B \e$(7!}\e(B \e$(7!~\e(B // ;
  53 ;;;
  54 ;;;     00 01 02 03 04 05 06 07 08 09 0A 0B 0C 0D 0E 0F
  55 ;;;2220 // \e$(7"!\e(B \e$(7""\e(B \e$(7"#\e(B \e$(7"$\e(B \e$(7"%\e(B \e$(7"&\e(B \e$(7"'\e(B \e$(7"(\e(B \e$(7")\e(B \e$(7"*\e(B \e$(7"+\e(B \e$(7",\e(B \e$(7"-\e(B \e$(7".\e(B \e$(7"/\e(B ; Base consonants
  56 ;;;2230 \e$(7"0\e(B \e$(7"1\e(B \e$(7"2\e(B \e$(7"3\e(B \e$(7"4\e(B \e$(7"5\e(B \e$(7"6\e(B \e$(7"7\e(B \e$(7"8\e(B \e$(7"9\e(B \e$(7":\e(B \e$(7";\e(B \e$(7"<\e(B \e$(7"=\e(B \e$(7">\e(B \e$(7"?\e(B ; and
  57 ;;;2240 \e$(7"@\e(B \e$(7"A\e(B \e$(7"B\e(B \e$(7"C\e(B \e$(7"D\e(B \e$(7"E\e(B \e$(7"F\e(B \e$(7"G\e(B \e$(7"H\e(B \e$(7"I\e(B \e$(7"J\e(B \e$(7"K\e(B \e$(7"L\e(B \e$(7"M\e(B \e$(7"N\e(B \e$(7"O\e(B ; Vowel signs.
  58 ;;;2250 \e$(7"P\e(B \e$(7"Q\e(B \e$(7"R\e(B \e$(7"S\e(B \e$(7"T\e(B \e$(7"U\e(B \e$(7"V\e(B \e$(7"W\e(B \e$(7"X\e(B \e$(7"Y\e(B \e$(7"Z\e(B \e$(7"[\e(B \e$(7"\\e(B \e$(7"]\e(B \e$(7"^\e(B \e$(7"_\e(B ; (\x2251 = vowel a)
  59 ;;;2260 \e$(7"`\e(B \e$(7"a\e(B \e$(7"b\e(B \e$(7"c\e(B \e$(7"d\e(B \e$(7"e\e(B \e$(7"f\e(B \e$(7"g\e(B \e$(7"h\e(B \e$(7"i\e(B \e$(7"j\e(B \e$(7"k\e(B \e$(7"l\e(B \e$(7"m\e(B \e$(7"n\e(B \e$(7"o\e(B ; Long vowels and
  60 ;;;2270 \e$(7"p\e(B \e$(7"q\e(B \e$(7"r\e(B \e$(7"s\e(B \e$(7"t\e(B \e$(7"u\e(B \e$(7"v\e(B \e$(7"w\e(B \e$(7"x\e(B \e$(7"y\e(B \e$(7"z\e(B \e$(7"{\e(B \e$(7"|\e(B \e$(7"}\e(B \e$(7"~\e(B // ; vocalic r, l ARE
  61 ;;;                                                     ; atomically
  62 ;;;                                                     ; encoded.
  63 ;;;     00 01 02 03 04 05 06 07 08 09 0A 0B 0C 0D 0E 0F
  64 ;;;2320 // \e$(7#!\e(B \e$(7#"\e(B \e$(7##\e(B \e$(7#$\e(B \e$(7#%\e(B \e$(7#&\e(B \e$(7#'\e(B \e$(7#(\e(B \e$(7#)\e(B \e$(7#*\e(B \e$(7#+\e(B \e$(7#,\e(B \e$(7#-\e(B \e$(7#.\e(B \e$(7#/\e(B ; Subjoined consonants
  65 ;;;2330 \e$(7#0\e(B \e$(7#1\e(B \e$(7#2\e(B \e$(7#3\e(B \e$(7#4\e(B \e$(7#5\e(B \e$(7#6\e(B \e$(7#7\e(B \e$(7#8\e(B \e$(7#9\e(B \e$(7#:\e(B \e$(7#;\e(B \e$(7#<\e(B \e$(7#=\e(B \e$(7#>\e(B \e$(7#?\e(B ;
  66 ;;;2340 \e$(7#@\e(B \e$(7#A\e(B \e$(7#B\e(B \e$(7#C\e(B \e$(7#D\e(B \e$(7#E\e(B \e$(7#F\e(B \e$(7#G\e(B \e$(7#H\e(B \e$(7#I\e(B \e$(7#J\e(B \e$(7#K\e(B \e$(7#L\e(B \e$(7#M\e(B \e$(7#N\e(B \e$(7#O\e(B ;
  67 ;;;2350 \e$(7#P\e(B \e$(7#Q\e(B \e$(7#R\e(B \e$(7#S\e(B \e$(7#T\e(B \e$(7#U\e(B \e$(7#V\e(B \e$(7#W\e(B \e$(7#X\e(B \e$(7#Y\e(B \e$(7#Z\e(B \e$(7#[\e(B \e$(7#\\e(B \e$(7#]\e(B \e$(7#^\e(B \e$(7#_\e(B ; Hereafter, the chars
  68 ;;;2360 \e$(7#`\e(B \e$(7#a\e(B \e$(7#b\e(B \e$(7#c\e(B \e$(7#d\e(B \e$(7#e\e(B \e$(7#f\e(B \e$(7#g\e(B \e$(7#h\e(B \e$(7#i\e(B \e$(7#j\e(B \e$(7#k\e(B \e$(7#l\e(B \e$(7#m\e(B \e$(7#n\e(B \e$(7#o\e(B ; are not specified
  69 ;;;2370 \e$(7#p\e(B \e$(7#q\e(B \e$(7#r\e(B \e$(7#s\e(B \e$(7#t\e(B \e$(7#u\e(B \e$(7#v\e(B \e$(7#w\e(B \e$(7#x\e(B \e$(7#y\e(B \e$(7#z\e(B \e$(7#{\e(B \e$(7#|\e(B \e$(7#}\e(B \e$(7#~\e(B // ; in Unicode.
  70 ;;;
  71 ;;;     00 01 02 03 04 05 06 07 08 09 0A 0B 0C 0D 0E 0F
  72 ;;;2420 // \e$(7$!\e(B \e$(7$"\e(B \e$(7$#\e(B \e$(7$$\e(B \e$(7$%\e(B \e$(7$&\e(B \e$(7$'\e(B \e$(7$(\e(B \e$(7$)\e(B \e$(7$*\e(B \e$(7$+\e(B \e$(7$,\e(B \e$(7$-\e(B \e$(7$.\e(B \e$(7$/\e(B ; Precomposed
  73 ;;;2430 \e$(7$0\e(B \e$(7$1\e(B \e$(7$2\e(B \e$(7$3\e(B \e$(7$4\e(B \e$(7$5\e(B \e$(7$6\e(B \e$(7$7\e(B \e$(7$8\e(B \e$(7$9\e(B \e$(7$:\e(B \e$(7$;\e(B \e$(7$<\e(B \e$(7$=\e(B \e$(7$>\e(B \e$(7$?\e(B ; consonants for
  74 ;;;2440 \e$(7$@\e(B \e$(7$A\e(B \e$(7$B\e(B \e$(7$C\e(B \e$(7$D\e(B \e$(7$E\e(B \e$(7$F\e(B \e$(7$G\e(B \e$(7$H\e(B \e$(7$I\e(B \e$(7$J\e(B \e$(7$K\e(B \e$(7$L\e(B \e$(7$M\e(B \e$(7$N\e(B \e$(7$O\e(B ; ordinary Tibetan.
  75 ;;;2450 \e$(7$P\e(B \e$(7$Q\e(B \e$(7$R\e(B \e$(7$S\e(B \e$(7$T\e(B \e$(7$U\e(B \e$(7$V\e(B \e$(7$W\e(B \e$(7$X\e(B \e$(7$Y\e(B \e$(7$Z\e(B \e$(7$[\e(B \e$(7$\\e(B \e$(7$]\e(B \e$(7$^\e(B \e$(7$_\e(B ; They are decomposed
  76 ;;;2460 \e$(7$`\e(B \e$(7$a\e(B \e$(7$b\e(B \e$(7$c\e(B \e$(7$d\e(B \e$(7$e\e(B \e$(7$f\e(B \e$(7$g\e(B \e$(7$h\e(B \e$(7$i\e(B \e$(7$j\e(B \e$(7$k\e(B \e$(7$l\e(B \e$(7$m\e(B \e$(7$n\e(B \e$(7$o\e(B ; into base and
  77 ;;;2470 \e$(7$p\e(B \e$(7$q\e(B \e$(7$r\e(B \e$(7$s\e(B \e$(7$t\e(B \e$(7$u\e(B \e$(7$v\e(B \e$(7$w\e(B \e$(7$x\e(B \e$(7$y\e(B \e$(7$z\e(B \e$(7${\e(B \e$(7$|\e(B \e$(7$}\e(B \e$(7$~\e(B // ; subjoined consonants
  78 ;;;                                                     ; when written on a
  79 ;;;     00 01 02 03 04 05 06 07 08 09 0A 0B 0C 0D 0E 0F ; file in Tibetan
  80 ;;;2520 // \e$(7%!\e(B \e$(7%"\e(B \e$(7%#\e(B \e$(7%$\e(B \e$(7%%\e(B \e$(7%&\e(B \e$(7%'\e(B \e$(7%(\e(B \e$(7%)\e(B \e$(7%*\e(B \e$(7%+\e(B \e$(7%,\e(B \e$(7%-\e(B \e$(7%.\e(B \e$(7%/\e(B ; coding system.
  81 ;;;2530 \e$(7%0\e(B \e$(7%1\e(B \e$(7%2\e(B \e$(7%3\e(B \e$(7%4\e(B \e$(7%5\e(B \e$(7%6\e(B \e$(7%7\e(B \e$(7%8\e(B \e$(7%9\e(B \e$(7%:\e(B \e$(7%;\e(B \e$(7%<\e(B \e$(7%=\e(B \e$(7%>\e(B \e$(7%?\e(B ;
  82 ;;;2540 \e$(7%@\e(B \e$(7%A\e(B \e$(7%B\e(B \e$(7%C\e(B \e$(7%D\e(B \e$(7%E\e(B \e$(7%F\e(B \e$(7%G\e(B \e$(7%H\e(B \e$(7%I\e(B \e$(7%J\e(B \e$(7%K\e(B \e$(7%L\e(B \e$(7%M\e(B \e$(7%N\e(B \e$(7%O\e(B ;
  83 ;;;2550 \e$(7%P\e(B \e$(7%Q\e(B \e$(7%R\e(B \e$(7%S\e(B \e$(7%T\e(B \e$(7%U\e(B \e$(7%V\e(B \e$(7%W\e(B \e$(7%X\e(B \e$(7%Y\e(B \e$(7%Z\e(B \e$(7%[\e(B \e$(7%\\e(B \e$(7%]\e(B \e$(7%^\e(B \e$(7%_\e(B ;
  84 ;;;2560 \e$(7%`\e(B \e$(7%a\e(B \e$(7%b\e(B \e$(7%c\e(B \e$(7%d\e(B \e$(7%e\e(B \e$(7%f\e(B \e$(7%g\e(B \e$(7%h\e(B \e$(7%i\e(B \e$(7%j\e(B \e$(7%k\e(B \e$(7%l\e(B \e$(7%m\e(B \e$(7%n\e(B \e$(7%o\e(B ;
  85 ;;;2570 \e$(7%p\e(B \e$(7%q\e(B \e$(7%r\e(B \e$(7%s\e(B \e$(7%t\e(B \e$(7%u\e(B \e$(7%v\e(B \e$(7%w\e(B \e$(7%x\e(B \e$(7%y\e(B \e$(7%z\e(B \e$(7%{\e(B \e$(7%|\e(B \e$(7%}\e(B \e$(7%~\e(B // ;
  86 ;;;
  87
  88
  89 (define-coding-system 'tibetan-iso-8bit
  90   "8-bit encoding for ASCII (MSB=0) and TIBETAN (MSB=1)."
  91   :coding-type 'iso-2022
  92   :mnemonic ?Q
  93   :designation [ascii tibetan nil nil]
  94   :charset-list '(ascii tibetan))
  95
  96 (define-coding-system-alias 'tibetan 'tibetan-iso-8bit)
  97
  98 (set-language-info-alist
  99  "Tibetan" '((charset tibetan tibetan-1-column)
 100              (coding-system tibetan-iso-8bit)
 101              (coding-priority iso-2022-7bit tibetan-iso-8bit)
 102              (input-method . "tibetan-wylie")
 103              (features tibet-util)
 104              (documentation . t)
 105              (sample-text . "Tibetan (\e$(7"7"]"2!;"G#!"Q"2!;\e(B) \e$(7!4!5!5!>"7"!#C"Q!;"E"S"G!;"7"2"[!;"D"["#"G!>"I"]"_!;"9"Q!;"/"S!;"5"Q"2#9"[!;"H"A"U"c!>\e(B")))
 106
 107 ;; `\e$(7"A\e(B' is included in the pattern for subjoined consonants because we
 108 ;; treat it specially in tibetan-add-components.
 109 ;; modified by Tomabechi 1999/12/10
 110 ;; modified by Tomabechi 2000/06/08
 111 ;;          To allow infinite addition of vowels/modifiers
 112 ;;          as specified in Unicode v.3
 113 ;; \e$(7"A\e(B is removed from the class of subjoined. Tomabechi 2000/06/08
 114 ;; (for Unicode support)
 115 (defconst tibetan-composable-pattern
 116   "[\e$(7"!\e(B-\e$(7"J"K\e(B][\e$(7#!\e(B-\e$(7#J#K#L#M\e(B]*[\e$,1FP\e$(7"Q"R"S\e(B-\e$(7"^"a"b"e\e(B]*[\e$(7"_"c"d"g\e(B-\e$(7"l!I!e!g\e(B]*"
 117   "Regexp matching a composable sequence of Tibetan characters.")
 118
 119 ;;;
 120 ;;; Definitions of conversion data.
 121 ;;;
 122
 123
 124 ;;; alists for tibetan char <-> transcription conversion
 125 ;;; longer transcription should come first
 126 (defconst tibetan-consonant-transcription-alist
 127   '(("tsh" . "\e$(7";\e(B")
 128     ("dzh" . "\e$(7"=\e(B")
 129     ("kSH" . "\e$(7"J\e(B")
 130     ("kh" . "\e$(7""\e(B")
 131     ("gh" . "\e$(7"$\e(B")
 132     ("ng" . "\e$(7"%\e(B")
 133     ("ch" . "\e$(7"'\e(B")
 134     ("ny" . "\e$(7"*\e(B")
 135     ("TH" . "\e$(7",\e(B")
 136     ("DH" . "\e$(7".\e(B")
 137     ("th" . "\e$(7"1\e(B")
 138     ("dh" . "\e$(7"3\e(B")
 139     ("ph" . "\e$(7"6\e(B")
 140     ("bh" . "\e$(7"8\e(B")
 141     ("ts" . "\e$(7":\e(B")
 142     ("dz" . "\e$(7"<\e(B")
 143     ("zh" . "\e$(7"?\e(B")
 144     ("sh" . "\e$(7"E\e(B")
 145     ("SH" . "\e$(7"F\e(B")
 146     ("k" . "\e$(7"!\e(B")
 147     ("g" . "\e$(7"#\e(B")
 148     ("c" . "\e$(7"&\e(B")
 149     ("j" . "\e$(7"(\e(B")
 150     ("T" . "\e$(7"+\e(B")
 151     ("D" . "\e$(7"-\e(B")
 152     ("N" . "\e$(7"/\e(B")
 153     ("t" . "\e$(7"0\e(B")
 154     ("d" . "\e$(7"2\e(B")
 155     ("n" . "\e$(7"4\e(B")
 156     ("p" . "\e$(7"5\e(B")
 157     ("b" . "\e$(7"7\e(B")
 158     ("m" . "\e$(7"9\e(B")
 159     ("w" . "\e$(7">\e(B")
 160     ("z" . "\e$(7"@\e(B")
 161     ("'" . "\e$(7"A\e(B")
 162     ("y" . "\e$(7"B\e(B")
 163     ("r" . "\e$(7"C\e(B")
 164     ("l" . "\e$(7"D\e(B")
 165     ("s" . "\e$(7"G\e(B")
 166     ("h" . "\e$(7"H\e(B")
 167     ("H" . "\e$(7"H\e(B")
 168     ("A" . "\e$(7"I\e(B")
 169     ;; Added by Tomabechi 1999/12/10
 170     ("R" . "\e$(7"K\e(B") ;; fixed form RA
 171     ))
 172
 173
 174 (defconst tibetan-vowel-transcription-alist
 175   '(
 176     ;; Composite Vowels
 177     ;; Added by Tomabechi 2000/06/08
 178     ("frr" . "\e$(7"X\e(B")
 179     ("fll" . "\e$(7"Z\e(B")
 180     ("fa" . "\e$(7"R\e(B")
 181     ("fi" . "\e$(7"T\e(B")
 182     ("fu" . "\e$(7"V\e(B")
 183     ("fr" . "\e$(7"W\e(B")
 184     ("fl" . "\e$(7"Y\e(B")
 185     ("fI" . "\e$(7"b\e(B")
 186     ;; Normal Vowels
 187     ("ai" . "\e$(7"\\e(B")
 188     ("au" . "\e$(7"^\e(B")
 189     ("ee" . "\e$(7"\\e(B")
 190     ("oo" . "\e$(7"^\e(B")
 191     ("a" . "\e$(7"Q\e(B")                 ; invisible vowel sign (\x2251)
 192     ("i" . "\e$(7"S\e(B")
 193     ("u" . "\e$(7"U\e(B")
 194     ("e" . "\e$(7"[\e(B")
 195     ("o" . "\e$(7"]\e(B")
 196     ("E" . "\e$(7"\\e(B")
 197     ("O" . "\e$(7"^\e(B")
 198     ("I" . "\e$(7"a\e(B")
 199     ("," . "\e$(7"e\e(B")                 ; idem.
 200     ))
 201
 202 (defconst tibetan-modifier-transcription-alist
 203   '(("M" . "\e$(7"_\e(B")
 204     ("~" . "\e$(7"c\e(B")
 205     ("`" . "\e$(7"d\e(B")
 206     ("x" . "\e$(7"i\e(B")
 207     ("X" . "\e$(7"j\e(B")
 208     ("v" . "\e$(7"g\e(B")
 209     ("V" . "\e$(7"h\e(B")
 210     ("q" . "\e$(7"k\e(B")
 211     ("Q" . "\e$(7"l\e(B")
 212     ("_/" . "\e$(7!I\e(B")
 213     ("_o" . "\e$(7!g\e(B")
 214     ("_O" . "\e$(7!e\e(B")))
 215
 216 (defconst tibetan-precomposed-transcription-alist
 217   '(("phyw" . "\e$(7$G\e(B")
 218     ("tshw" . "\e$(7$)\e(B")
 219     ("rtsw" . "\e$(7%.\e(B")
 220     ("khw" . "\e$(7$"\e(B")
 221     ("nyw" . "\e$(7$%\e(B")
 222     ("tsw" . "\e$(7$(\e(B")
 223     ("zhw" . "\e$(7$*\e(B")
 224     ("shw" . "\e$(7$.\e(B")
 225     ("khy" . "\e$(7$A\e(B")
 226     ("phy" . "\e$(7$D\e(B")
 227     ("khr" . "\e$(7$Q\e(B")
 228     ("thr" . "\e$(7$T\e(B")
 229     ("phr" . "\e$(7$W\e(B")
 230     ("shr" . "\e$(7$Z\e(B")
 231     ("dzr" . "\e$(7$^\e(B")
 232     ("grw" . "\e$(7$_\e(B")
 233     ("rng" . "\e$(7%#\e(B")
 234     ("rny" . "\e$(7%%\e(B")
 235     ("rts" . "\e$(7%+\e(B")
 236     ("rdz" . "\e$(7%,\e(B")
 237     ("rgw" . "\e$(7%-\e(B")
 238     ("rky" . "\e$(7%0\e(B")
 239     ("rgy" . "\e$(7%1\e(B")
 240     ("rmy" . "\e$(7%2\e(B")
 241     ("lng" . "\e$(7%B\e(B")
 242     ("sng" . "\e$(7%R\e(B")
 243     ("sny" . "\e$(7%S\e(B")
 244     ("sts" . "\e$(7%Z\e(B")
 245     ("sky" . "\e$(7%`\e(B")
 246     ("sgy" . "\e$(7%a\e(B")
 247     ("spy" . "\e$(7%b\e(B")
 248     ("sby" . "\e$(7%c\e(B")
 249     ("smy" . "\e$(7%d\e(B")
 250     ("skr" . "\e$(7%p\e(B")
 251     ("sgr" . "\e$(7%q\e(B")
 252     ("snr" . "\e$(7%r\e(B")
 253     ("spr" . "\e$(7%s\e(B")
 254     ("sbr" . "\e$(7%t\e(B")
 255     ("smr" . "\e$(7%u\e(B")
 256     ("kw" . "\e$(7$!\e(B")
 257     ("gw" . "\e$(7$#\e(B")
 258     ("cw" . "\e$(7$$\e(B")
 259     ("tw" . "\e$(7$&\e(B")
 260     ("dw" . "\e$(7$'\e(B")
 261     ("zw" . "\e$(7$+\e(B")
 262     ("rw" . "\e$(7$,\e(B")
 263     ("lw" . "\e$(7$-\e(B")
 264     ("sw" . "\e$(7$/\e(B")
 265     ("hw" . "\e$(7$0\e(B")
 266     ("ky" . "\e$(7$@\e(B")
 267     ("gy" . "\e$(7$B\e(B")
 268     ("py" . "\e$(7$C\e(B")
 269     ("by" . "\e$(7$E\e(B")
 270     ("my" . "\e$(7$F\e(B")
 271     ("kr" . "\e$(7$P\e(B")
 272     ("gr" . "\e$(7$R\e(B")
 273     ("tr" . "\e$(7$S\e(B")
 274     ("dr" . "\e$(7$U\e(B")
 275     ("pr" . "\e$(7$V\e(B")
 276     ("brk" . "\e$(7"7%!\e(B")
 277     ("brg" . "\e$(7"7%"\e(B")
 278     ("brng" . "\e$(7"7%#\e(B")
 279     ("brj" . "\e$(7"7%$\e(B")
 280     ("brny" . "\e$(7"7%%\e(B")
 281     ("brt" .  "\e$(7"7%&\e(B")
 282     ("brd" . "\e$(7"7%'\e(B")
 283     ("brn" . "\e$(7"7%(\e(B")
 284     ("brts" . "\e$(7"7%+\e(B")
 285     ("brdz" . "\e$(7"7%,\e(B")
 286     ("brl" . "\e$(7"7$d\e(B")
 287     ("br" . "\e$(7$X\e(B")
 288     ("mr" . "\e$(7$Y\e(B")
 289     ("sr" . "\e$(7$[\e(B")
 290     ("hr" . "\e$(7$\\e(B")
 291     ("jr" . "\e$(7$]\e(B")
 292     ("kl" . "\e$(7$`\e(B")
 293     ("gl" . "\e$(7$a\e(B")
 294     ("blt" . "\e$(7"7%E\e(B")
 295     ("bld" . "\e$(7"7%F\e(B")
 296     ("bl" . "\e$(7$b\e(B")
 297     ("zl" . "\e$(7$c\e(B")
 298     ("rl" . "\e$(7$d\e(B")
 299     ("sl" . "\e$(7$e\e(B")
 300     ("rk" . "\e$(7%!\e(B")
 301     ("rg" . "\e$(7%"\e(B")
 302     ("rj" . "\e$(7%$\e(B")
 303     ("rt" . "\e$(7%&\e(B")
 304     ("rd" . "\e$(7%'\e(B")
 305     ("rn" . "\e$(7%(\e(B")
 306     ("rb" . "\e$(7%)\e(B")
 307     ("rm" . "\e$(7%*\e(B")
 308     ("lk" . "\e$(7%@\e(B")
 309     ("lg" . "\e$(7%A\e(B")
 310     ("lc" . "\e$(7%C\e(B")
 311     ("lj" . "\e$(7%D\e(B")
 312     ("lt" . "\e$(7%E\e(B")
 313     ("ld" . "\e$(7%F\e(B")
 314     ("ln" . "\e$(7!!\e(B")                        ; dummy \x2121
 315     ("lp" . "\e$(7%G\e(B")
 316     ("lb" . "\e$(7%H\e(B")
 317     ("lh" . "\e$(7%I\e(B")
 318     ("sk" . "\e$(7%P\e(B")
 319     ("sg" . "\e$(7%Q\e(B")
 320     ("st" . "\e$(7%T\e(B")
 321     ("sd" . "\e$(7%U\e(B")
 322     ("sn" . "\e$(7%V\e(B")
 323     ("sp" . "\e$(7%W\e(B")
 324     ("sb" . "\e$(7%X\e(B")
 325     ("sm" . "\e$(7%Y\e(B"))
 326   )
 327
 328
 329 (defconst tibetan-subjoined-transcription-alist
 330   (sort '(("+k"  . "\e$(7#!\e(B")
 331           ("+kh" . "\e$(7#"\e(B")
 332           ("+g"  . "\e$(7##\e(B")
 333           ("+gh" . "\e$(7#$\e(B")
 334           ("+ng" . "\e$(7#%\e(B")
 335           ("+c"  . "\e$(7#&\e(B")
 336           ("+ch" . "\e$(7#'\e(B")
 337           ("+j"  . "\e$(7#(\e(B")
 338           ("+ny"  . "\e$(7#*\e(B")
 339           ("+T"  . "\e$(7#+\e(B")
 340           ("+TH" . "\e$(7#,\e(B")
 341           ("+D"  . "\e$(7#-\e(B")
 342           ("+DH" . "\e$(7#.\e(B")
 343           ("+N"  . "\e$(7#/\e(B")
 344           ("+t"  . "\e$(7#0\e(B")
 345           ("+th" . "\e$(7#1\e(B")
 346           ("+d"  . "\e$(7#2\e(B")
 347           ("+dh" . "\e$(7#3\e(B")
 348           ("+n"  . "\e$(7#4\e(B")
 349           ("+p"  . "\e$(7#5\e(B")
 350           ("+ph" . "\e$(7#6\e(B")
 351           ("+b"  . "\e$(7#7\e(B")
 352           ("+bh" . "\e$(7#8\e(B")
 353           ("+m"  . "\e$(7#9\e(B")
 354           ("+ts" . "\e$(7#:\e(B")
 355           ("+tsh" . "\e$(7#;\e(B")
 356           ("+dz" . "\e$(7#<\e(B")
 357           ("+dzh" . "\e$(7#=\e(B")
 358           ("+w"  . "\e$(7#>\e(B")
 359           ("+zh" . "\e$(7#?\e(B")
 360           ("+z"  . "\e$(7#@\e(B")
 361           ("+'"  . "\e$(7#A\e(B")
 362           ("+y"  . "\e$(7#B\e(B")
 363           ("+r"  . "\e$(7#C\e(B")
 364           ("+l"  . "\e$(7#D\e(B")
 365           ("+sh" . "\e$(7#E\e(B")
 366           ("+SH" . "\e$(7#F\e(B")
 367           ("+s"  . "\e$(7#G\e(B")
 368           ("+h"  . "\e$(7#H\e(B")
 369           ("+A"  . "\e$(7#I\e(B")
 370           ("+kSH" . "\e$(7#J\e(B")
 371           ;; Added by Tomabechi 1999/12/10
 372           ("+W" . "\e$(7#K\e(B") ;; fixed form subscribed WA
 373           ("+Y" . "\e$(7#L\e(B") ;; fixed form subscribed YA
 374           ("+R" . "\e$(7#M\e(B") ;; fixed form subscribed RA
 375           )
 376         (lambda (x y) (> (length (car x)) (length (car y))))))
 377
 378 ;;;
 379 ;;; alist for Tibetan base consonant <-> subjoined consonant conversion.
 380 ;;;
 381 (defconst tibetan-base-to-subjoined-alist
 382   '(("\e$(7"!\e(B" . "\e$(7#!\e(B")
 383     ("\e$(7""\e(B" . "\e$(7#"\e(B")
 384     ("\e$(7"#\e(B" . "\e$(7##\e(B")
 385     ("\e$(7"$\e(B" . "\e$(7#$\e(B")
 386     ("\e$(7"%\e(B" . "\e$(7#%\e(B")
 387     ("\e$(7"&\e(B" . "\e$(7#&\e(B")
 388     ("\e$(7"'\e(B" . "\e$(7#'\e(B")
 389     ("\e$(7"(\e(B" . "\e$(7#(\e(B")
 390     ("\e$(7"*\e(B" . "\e$(7#*\e(B")
 391     ("\e$(7"+\e(B" . "\e$(7#+\e(B")
 392     ("\e$(7",\e(B" . "\e$(7#,\e(B")
 393     ("\e$(7"-\e(B" . "\e$(7#-\e(B")
 394     ("\e$(7".\e(B" . "\e$(7#.\e(B")
 395     ("\e$(7"/\e(B" . "\e$(7#/\e(B")
 396     ("\e$(7"0\e(B" . "\e$(7#0\e(B")
 397     ("\e$(7"1\e(B" . "\e$(7#1\e(B")
 398     ("\e$(7"2\e(B" . "\e$(7#2\e(B")
 399     ("\e$(7"3\e(B" . "\e$(7#3\e(B")
 400     ("\e$(7"4\e(B" . "\e$(7#4\e(B")
 401     ("\e$(7"5\e(B" . "\e$(7#5\e(B")
 402     ("\e$(7"6\e(B" . "\e$(7#6\e(B")
 403     ("\e$(7"7\e(B" . "\e$(7#7\e(B")
 404     ("\e$(7"8\e(B" . "\e$(7#8\e(B")
 405     ("\e$(7"9\e(B" . "\e$(7#9\e(B")
 406     ("\e$(7":\e(B" . "\e$(7#:\e(B")
 407     ("\e$(7";\e(B" . "\e$(7#;\e(B")
 408     ("\e$(7"<\e(B" . "\e$(7#<\e(B")
 409     ("\e$(7"=\e(B" . "\e$(7#=\e(B")
 410     ("\e$(7">\e(B" . "\e$(7#>\e(B")
 411     ("\e$(7"?\e(B" . "\e$(7#?\e(B")
 412     ("\e$(7"@\e(B" . "\e$(7#@\e(B")
 413     ("\e$(7"A\e(B" . "\e$(7#A\e(B")
 414     ("\e$(7"B\e(B" . "\e$(7#B\e(B")
 415     ("\e$(7"C\e(B" . "\e$(7#C\e(B")
 416     ("\e$(7"D\e(B" . "\e$(7#D\e(B")
 417     ("\e$(7"E\e(B" . "\e$(7#E\e(B")
 418     ("\e$(7"F\e(B" . "\e$(7#F\e(B")
 419     ("\e$(7"G\e(B" . "\e$(7#G\e(B")
 420     ("\e$(7"H\e(B" . "\e$(7#H\e(B")
 421     ("\e$(7"I\e(B" . "\e$(7#I\e(B")
 422     ("\e$(7"J\e(B" . "\e$(7#J\e(B")
 423     ;; Added by Tomabechi 1999/12/10
 424     ("\e$(7"K\e(B" . "\e$(7#M\e(B") ;; Fixed form RA (224B->234D)
 425     ))
 426
 427 ;;; alist for Tibetan composite vowels (long i, vocalic r, etc.)
 428 ;;; New varialble. created by Tomabechi 2000/06/08
 429 (defconst tibetan-composite-vowel-alist
 430   '(;; LONG A
 431     ;; ("\e$(7"R\e(B" . ((bc . tc) ?\e$(7"R\e(B))
 432     ;; LONG I
 433     ("\e$(7"T\e(B" . (?\e$(7"R\e(B (tc . bc) ?\e$(7"S\e(B))
 434     ;; LONG U
 435     ("\e$(7"V\e(B" . (?\e$(7"R\e(B (bc . tc) ?\e$(7"U\e(B))
 436     ;; VOCALIC R
 437     ("\e$(7"W\e(B" . (?\e$(7#C\e(B (tc . bc) ?\e$(7"a\e(B))
 438     ;; LONG VOCALIC R
 439     ("\e$(7"X\e(B" . (?\e$(7#C\e(B (bc . tc) ?\e$(7"R\e(B (tc . bc) ?\e$(7"a\e(B))
 440     ;; VOCALIC L
 441     ("\e$(7"Y\e(B" . (?\e$(7#D\e(B (tc . bc) ?\e$(7"a\e(B))
 442     ;;\e$(7!;\e(BLONG VOCALIC L
 443     ("\e$(7"Z\e(B" . (?\e$(7#D\e(B (bc . tc) ?\e$(7"R\e(B (tc . bc) ?\e$(7"a\e(B))
 444     ;; LONG REVERSE I
 445     ("\e$(7"b\e(B" . (?\e$(7"R\e(B (tc . bc) ?\e$(7"a\e(B))
 446     ))
 447
 448
 449
 450 ;;;
 451 ;;; alist for Tibetan consonantic components <-> precomposed glyph conversion.
 452 ;;; (includes some punctuation conversion rules)
 453 ;;;
 454 (defconst tibetan-precomposition-rule-alist
 455   `(("\e$(7"6#B#>\e(B" . "\e$(7$G\e(B")
 456     ("\e$(7"##C#>\e(B" . "\e$(7$_\e(B")
 457     ("\e$(7";#>\e(B" . "\e$(7$)\e(B")
 458     ("\e$(7"C#:#>\e(B" . "\e$(7%.\e(B")
 459     ("\e$(7"C###>\e(B" . "\e$(7%-\e(B")
 460     ("\e$(7"C#!#B\e(B" . "\e$(7%0\e(B")
 461     ("\e$(7"C###B\e(B" . "\e$(7%1\e(B")
 462     ("\e$(7"C#9#B\e(B" . "\e$(7%2\e(B")
 463     ("\e$(7"G#!#B\e(B" . "\e$(7%`\e(B")
 464     ("\e$(7"G###B\e(B" . "\e$(7%a\e(B")
 465     ("\e$(7"G#5#B\e(B" . "\e$(7%b\e(B")
 466     ("\e$(7"G#7#B\e(B" . "\e$(7%c\e(B")
 467     ("\e$(7"G#9#B\e(B" . "\e$(7%d\e(B")
 468     ("\e$(7"G#!#C\e(B" . "\e$(7%p\e(B")
 469     ("\e$(7"G###C\e(B" . "\e$(7%q\e(B")
 470     ("\e$(7"G#4#C\e(B" . "\e$(7%r\e(B")
 471     ("\e$(7"G#5#C\e(B" . "\e$(7%s\e(B")
 472     ("\e$(7"G#7#C\e(B" . "\e$(7%t\e(B")
 473     ("\e$(7"G#9#C\e(B" . "\e$(7%u\e(B")
 474     ("\e$(7""#>\e(B" . "\e$(7$"\e(B")
 475     ("\e$(7"*#>\e(B" . "\e$(7$%\e(B")
 476     ("\e$(7":#>\e(B" . "\e$(7$(\e(B")
 477     ("\e$(7"?#>\e(B" . "\e$(7$*\e(B")
 478     ("\e$(7"E#>\e(B" . "\e$(7$.\e(B")
 479     ("\e$(7""#B\e(B" . "\e$(7$A\e(B")
 480     ("\e$(7"6#B\e(B" . "\e$(7$D\e(B")
 481     ("\e$(7""#C\e(B" . "\e$(7$Q\e(B")
 482     ("\e$(7"1#C\e(B" . "\e$(7$T\e(B")
 483     ("\e$(7"6#C\e(B" . "\e$(7$W\e(B")
 484     ("\e$(7"E#C\e(B" . "\e$(7$Z\e(B")
 485     ("\e$(7"<#C\e(B" . "\e$(7$^\e(B")
 486     ("\e$(7"C#%\e(B" . "\e$(7%#\e(B")
 487     ("\e$(7"C#*\e(B" . "\e$(7%%\e(B")
 488     ("\e$(7"C#:\e(B" . "\e$(7%+\e(B")
 489     ("\e$(7"C#<\e(B" . "\e$(7%,\e(B")
 490     ("\e$(7"D#%\e(B" . "\e$(7%B\e(B")
 491     ("\e$(7"G#%\e(B" . "\e$(7%R\e(B")
 492     ("\e$(7"G#*\e(B" . "\e$(7%S\e(B")
 493     ("\e$(7"G#:\e(B" . "\e$(7%Z\e(B")
 494     ("\e$(7"!#>\e(B" . "\e$(7$!\e(B")
 495     ("\e$(7"##>\e(B" . "\e$(7$#\e(B")
 496     ("\e$(7"&#>\e(B" . "\e$(7$$\e(B")
 497     ("\e$(7"0#>\e(B" . "\e$(7$&\e(B")
 498     ("\e$(7"2#>\e(B" . "\e$(7$'\e(B")
 499     ("\e$(7"@#>\e(B" . "\e$(7$+\e(B")
 500     ("\e$(7"C#>\e(B" . "\e$(7$,\e(B")
 501     ("\e$(7"D#>\e(B" . "\e$(7$-\e(B")
 502     ("\e$(7"G#>\e(B" . "\e$(7$/\e(B")
 503     ("\e$(7"H#>\e(B" . "\e$(7$0\e(B")
 504     ("\e$(7"!#B\e(B" . "\e$(7$@\e(B")
 505     ("\e$(7"##B\e(B" . "\e$(7$B\e(B")
 506     ("\e$(7"5#B\e(B" . "\e$(7$C\e(B")
 507     ("\e$(7"7#B\e(B" . "\e$(7$E\e(B")
 508     ("\e$(7"9#B\e(B" . "\e$(7$F\e(B")
 509     ("\e$(7"!#C\e(B" . "\e$(7$P\e(B")
 510     ("\e$(7"##C\e(B" . "\e$(7$R\e(B")
 511     ("\e$(7"0#C\e(B" . "\e$(7$S\e(B")
 512     ("\e$(7"2#C\e(B" . "\e$(7$U\e(B")
 513     ("\e$(7"5#C\e(B" . "\e$(7$V\e(B")
 514     ("\e$(7"7#C\e(B" . "\e$(7$X\e(B")
 515     ("\e$(7"9#C\e(B" . "\e$(7$Y\e(B")
 516     ("\e$(7"G#C\e(B" . "\e$(7$[\e(B")
 517     ("\e$(7"H#C\e(B" . "\e$(7$\\e(B")
 518     ("\e$(7"(#C\e(B" . "\e$(7$]\e(B")
 519     ("\e$(7"!#D\e(B" . "\e$(7$`\e(B")
 520     ("\e$(7"##D\e(B" . "\e$(7$a\e(B")
 521     ("\e$(7"7#D\e(B" . "\e$(7$b\e(B")
 522     ("\e$(7"@#D\e(B" . "\e$(7$c\e(B")
 523     ("\e$(7"C#D\e(B" . "\e$(7$d\e(B")
 524     ("\e$(7"G#D\e(B" . "\e$(7$e\e(B")
 525     ("\e$(7"C#!\e(B" . "\e$(7%!\e(B")
 526     ("\e$(7"C##\e(B" . "\e$(7%"\e(B")
 527     ("\e$(7"C#(\e(B" . "\e$(7%$\e(B")
 528     ("\e$(7"C#0\e(B" . "\e$(7%&\e(B")
 529     ("\e$(7"C#2\e(B" . "\e$(7%'\e(B")
 530     ("\e$(7"C#4\e(B" . "\e$(7%(\e(B")
 531     ("\e$(7"C#7\e(B" . "\e$(7%)\e(B")
 532     ("\e$(7"C#9\e(B" . "\e$(7%*\e(B")
 533     ("\e$(7"D#!\e(B" . "\e$(7%@\e(B")
 534     ("\e$(7"D##\e(B" . "\e$(7%A\e(B")
 535     ("\e$(7"D#4\e(B" . "\e$(7!!\e(B") ; dummy 0x2121 added 2000/06/08 for transition l -> lng
 536     ("\e$(7"D#&\e(B" . "\e$(7%C\e(B")
 537     ("\e$(7"D#(\e(B" . "\e$(7%D\e(B")
 538     ("\e$(7"D#0\e(B" . "\e$(7%E\e(B")
 539     ("\e$(7"D#2\e(B" . "\e$(7%F\e(B")
 540     ("\e$(7"D#5\e(B" . "\e$(7%G\e(B")
 541     ("\e$(7"D#7\e(B" . "\e$(7%H\e(B")
 542     ("\e$(7"D#H\e(B" . "\e$(7%I\e(B")
 543     ("\e$(7"G#!\e(B" . "\e$(7%P\e(B")
 544     ("\e$(7"G##\e(B" . "\e$(7%Q\e(B")
 545     ("\e$(7"G#0\e(B" . "\e$(7%T\e(B")
 546     ("\e$(7"G#2\e(B" . "\e$(7%U\e(B")
 547     ("\e$(7"G#4\e(B" . "\e$(7%V\e(B")
 548     ("\e$(7"G#5\e(B" . "\e$(7%W\e(B")
 549     ("\e$(7"G#7\e(B" . "\e$(7%X\e(B")
 550     ("\e$(7"G#9\e(B" . "\e$(7%Y\e(B")))
 551
 552 (defconst tibetan-regexp
 553   (let ((l (list tibetan-precomposed-transcription-alist
 554                  tibetan-consonant-transcription-alist
 555                  tibetan-vowel-transcription-alist
 556                  tibetan-modifier-transcription-alist
 557                  tibetan-subjoined-transcription-alist))
 558         (separator "\\|")
 559         tail pattern)
 560     (while l
 561       (setq tail (car l) l (cdr l))
 562       (while tail
 563         (setq pattern (cons separator (cons (car (car tail)) pattern))
 564               tail (cdr tail))))
 565     (apply 'concat (nreverse (cdr pattern))))
 566   "Regexp matching a Tibetan transcription of a composable Tibetan sequence.
 567 The result of matching is to be used for indexing alists at conversion
 568 from a roman transcription to the corresponding Tibetan character.")
 569
 570 (defvar tibetan-precomposed-regexp
 571   (purecopy
 572   (let ((l tibetan-precomposed-transcription-alist)
 573         temp)
 574     (setq temp "^\\(")
 575     (setq temp
 576           (concat temp (car (car l))))
 577     (setq l (cdr l))
 578     (while l
 579       (setq temp
 580             (concat temp "\\|" (car (car l))))
 581       (setq l (cdr l)))
 582     (concat temp "\\)")))
 583   "Regexp string to match a romanized Tibetan complex consonant.
 584 The result of matching is to be used for indexing alists when the input key
 585 from an input method is converted to the corresponding precomposed glyph.")
 586
 587 (defvar tibetan-precomposition-rule-regexp
 588   (purecopy
 589   (let ((l tibetan-precomposition-rule-alist)
 590         temp)
 591     (setq temp "\\(")
 592     (setq temp (concat temp (car (car l))))
 593     (setq l (cdr l))
 594     (while l
 595       (setq temp (concat temp "\\|" (car (car l))))
 596       (setq l (cdr l)))
 597     (concat temp "\\)")))
 598   "Regexp string to match a sequence of Tibetan consonantic components, i.e.,
 599 one base consonant and one or more subjoined consonants.
 600 The result of matching is to be used for indexing alist when the component
 601 sequence is converted to the corresponding precomposed glyph.
 602 This also matches some punctuation characters which need conversion.")
 603
 604 (defvar tibetan-decomposed nil)
 605 (defvar tibetan-decomposed-temp nil)
 606
 607 ;; For automatic composition.
 608 (set-char-table-range
 609  composition-function-table '(#xF00 . #xFD1)
 610  (list (vector tibetan-composable-pattern 0 'font-shape-gstring)))
 611
 612 (provide 'tibetan)
 613
 614 ;;; tibetan.el ends here