lisp/language/tibetan.el

   1 ;;; tibetan.el --- support for Tibetan language -*- coding: iso-2022-7bit; -*-
   2
   3 ;; Copyright (C) 1997, 2001-2012  Free Software Foundation, Inc.
   4 ;; Copyright (C) 1997, 1998, 1999, 2000, 2001, 2002, 2003, 2004, 2005,
   5 ;;   2006, 2007, 2008, 2009, 2010, 2011
   6 ;;   National Institute of Advanced Industrial Science and Technology (AIST)
   7 ;;   Registration Number H14PRO021
   8 ;; Copyright (C) 2003
   9 ;;   National Institute of Advanced Industrial Science and Technology (AIST)
  10 ;;   Registration Number H13PRO009
  11
  12 ;; Author: Toru TOMABECHI <Toru.Tomabechi@orient.unil.ch>
  13 ;; Created: Feb. 17. 1997
  14 ;; Keywords: multilingual, Tibetan, i18n
  15
  16 ;; This file is part of GNU Emacs.
  17
  18 ;; GNU Emacs is free software: you can redistribute it and/or modify
  19 ;; it under the terms of the GNU General Public License as published by
  20 ;; the Free Software Foundation, either version 3 of the License, or
  21 ;; (at your option) any later version.
  22
  23 ;; GNU Emacs is distributed in the hope that it will be useful,
  24 ;; but WITHOUT ANY WARRANTY; without even the implied warranty of
  25 ;; MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
  26 ;; GNU General Public License for more details.
  27
  28 ;; You should have received a copy of the GNU General Public License
  29 ;; along with GNU Emacs.  If not, see <http://www.gnu.org/licenses/>.
  30
  31 ;;; History:
  32
  33 ;; 1997.03.13 Modification for special signs and punctuation.
  34
  35 ;;; Commentary:
  36
  37 ;;; Code:
  38
  39 ;;; Tibetan Character set.
  40 ;;; \x2130 -- \x234a is a subset of Unicode v.2 \x0f00 - \x0fb9
  41 ;;; with a slight modification. And there are some subjoined
  42 ;;; consonants which are not specified in Unicode.
  43 ;;; I hope I can add missing characters later.
  44 ;;;
  45 ;;;     00 01 02 03 04 05 06 07 08 09 0A 0B 0C 0D 0E 0F
  46 ;;;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)
  47 ;;;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 ; Punctuation,
  48 ;;;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
  49 ;;;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.
  50 ;;;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 ;
  51 ;;;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$(7!~\e(B // ;
  52 ;;;
  53 ;;;     00 01 02 03 04 05 06 07 08 09 0A 0B 0C 0D 0E 0F
  54 ;;;2220 // \e$(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$(7"/\e(B ; Base consonants
  55 ;;;2230 \e$(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$(7"?\e(B ; and
  56 ;;;2240 \e$(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$(7"O\e(B ; Vowel signs.
  57 ;;;2250 \e$(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$(7"_\e(B ; (\x2251 = vowel a)
  58 ;;;2260 \e$(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$(7"o\e(B ; Long vowels and
  59 ;;;2270 \e$(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
  60 ;;;                                                     ; atomically
  61 ;;;                                                     ; encoded.
  62 ;;;     00 01 02 03 04 05 06 07 08 09 0A 0B 0C 0D 0E 0F
  63 ;;;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
  64 ;;;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 ;
  65 ;;;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 ;
  66 ;;;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
  67 ;;;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
  68 ;;;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$(7#~\e(B // ; in Unicode.
  69 ;;;
  70 ;;;     00 01 02 03 04 05 06 07 08 09 0A 0B 0C 0D 0E 0F
  71 ;;;2420 // \e$(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
  72 ;;;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
  73 ;;;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.
  74 ;;;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
  75 ;;;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
  76 ;;;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
  77 ;;;                                                     ; when written on a
  78 ;;;     00 01 02 03 04 05 06 07 08 09 0A 0B 0C 0D 0E 0F ; file in Tibetan
  79 ;;;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.
  80 ;;;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 ;
  81 ;;;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 ;
  82 ;;;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 ;
  83 ;;;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 ;
  84 ;;;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 // ;
  85 ;;;
  86
  87
  88 (define-coding-system 'tibetan-iso-8bit
  89   "8-bit encoding for ASCII (MSB=0) and TIBETAN (MSB=1)."
  90   :coding-type 'iso-2022
  91   :mnemonic ?Q
  92   :designation [ascii tibetan nil nil]
  93   :charset-list '(ascii tibetan))
  94
  95 (define-coding-system-alias 'tibetan 'tibetan-iso-8bit)
  96
  97 (set-language-info-alist
  98  "Tibetan" '((charset tibetan tibetan-1-column)
  99              (coding-system tibetan-iso-8bit)
 100              (coding-priority iso-2022-7bit tibetan-iso-8bit)
 101              (input-method . "tibetan-wylie")
 102              (features tibet-util)
 103              (documentation . t)
 104              (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")))
 105
 106 ;; `\e$(7"A\e(B' is included in the pattern for subjoined consonants because we
 107 ;; treat it specially in tibetan-add-components.
 108 ;; modified by Tomabechi 1999/12/10
 109 ;; modified by Tomabechi 2000/06/08
 110 ;;          To allow infinite addition of vowels/modifiers
 111 ;;          as specified in Unicode v.3
 112 ;; \e$(7"A\e(B is removed from the class of subjoined. Tomabechi 2000/06/08
 113 ;; (for Unicode support)
 114 (defconst tibetan-composable-pattern
 115   "[\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]*"
 116   "Regexp matching a composable sequence of Tibetan characters.")
 117
 118 ;;;
 119 ;;; Definitions of conversion data.
 120 ;;;
 121
 122
 123 ;;; alists for tibetan char <-> transcription conversion
 124 ;;; longer transcription should come first
 125 (defconst tibetan-consonant-transcription-alist
 126   '(("tsh" . "\e$(7";\e(B")
 127     ("dzh" . "\e$(7"=\e(B")
 128     ("kSH" . "\e$(7"J\e(B")
 129     ("kh" . "\e$(7""\e(B")
 130     ("gh" . "\e$(7"$\e(B")
 131     ("ng" . "\e$(7"%\e(B")
 132     ("ch" . "\e$(7"'\e(B")
 133     ("ny" . "\e$(7"*\e(B")
 134     ("TH" . "\e$(7",\e(B")
 135     ("DH" . "\e$(7".\e(B")
 136     ("th" . "\e$(7"1\e(B")
 137     ("dh" . "\e$(7"3\e(B")
 138     ("ph" . "\e$(7"6\e(B")
 139     ("bh" . "\e$(7"8\e(B")
 140     ("ts" . "\e$(7":\e(B")
 141     ("dz" . "\e$(7"<\e(B")
 142     ("zh" . "\e$(7"?\e(B")
 143     ("sh" . "\e$(7"E\e(B")
 144     ("SH" . "\e$(7"F\e(B")
 145     ("k" . "\e$(7"!\e(B")
 146     ("g" . "\e$(7"#\e(B")
 147     ("c" . "\e$(7"&\e(B")
 148     ("j" . "\e$(7"(\e(B")
 149     ("T" . "\e$(7"+\e(B")
 150     ("D" . "\e$(7"-\e(B")
 151     ("N" . "\e$(7"/\e(B")
 152     ("t" . "\e$(7"0\e(B")
 153     ("d" . "\e$(7"2\e(B")
 154     ("n" . "\e$(7"4\e(B")
 155     ("p" . "\e$(7"5\e(B")
 156     ("b" . "\e$(7"7\e(B")
 157     ("m" . "\e$(7"9\e(B")
 158     ("w" . "\e$(7">\e(B")
 159     ("z" . "\e$(7"@\e(B")
 160     ("'" . "\e$(7"A\e(B")
 161     ("y" . "\e$(7"B\e(B")
 162     ("r" . "\e$(7"C\e(B")
 163     ("l" . "\e$(7"D\e(B")
 164     ("s" . "\e$(7"G\e(B")
 165     ("h" . "\e$(7"H\e(B")
 166     ("H" . "\e$(7"H\e(B")
 167     ("A" . "\e$(7"I\e(B")
 168     ;; Added by Tomabechi 1999/12/10
 169     ("R" . "\e$(7"K\e(B") ;; fixed form RA
 170     ))
 171
 172
 173 (defconst tibetan-vowel-transcription-alist
 174   '(
 175     ;; Composite Vowels
 176     ;; Added by Tomabechi 2000/06/08
 177     ("frr" . "\e$(7"X\e(B")
 178     ("fll" . "\e$(7"Z\e(B")
 179     ("fa" . "\e$(7"R\e(B")
 180     ("fi" . "\e$(7"T\e(B")
 181     ("fu" . "\e$(7"V\e(B")
 182     ("fr" . "\e$(7"W\e(B")
 183     ("fl" . "\e$(7"Y\e(B")
 184     ("fI" . "\e$(7"b\e(B")
 185     ;; Normal Vowels
 186     ("ai" . "\e$(7"\\e(B")
 187     ("au" . "\e$(7"^\e(B")
 188     ("ee" . "\e$(7"\\e(B")
 189     ("oo" . "\e$(7"^\e(B")
 190     ("a" . "\e$(7"Q\e(B")                 ; invisible vowel sign (\x2251)
 191     ("i" . "\e$(7"S\e(B")
 192     ("u" . "\e$(7"U\e(B")
 193     ("e" . "\e$(7"[\e(B")
 194     ("o" . "\e$(7"]\e(B")
 195     ("E" . "\e$(7"\\e(B")
 196     ("O" . "\e$(7"^\e(B")
 197     ("I" . "\e$(7"a\e(B")
 198     ("," . "\e$(7"e\e(B")                 ; idem.
 199     ))
 200
 201 (defconst tibetan-modifier-transcription-alist
 202   '(("M" . "\e$(7"_\e(B")
 203     ("~" . "\e$(7"c\e(B")
 204     ("`" . "\e$(7"d\e(B")
 205     ("x" . "\e$(7"i\e(B")
 206     ("X" . "\e$(7"j\e(B")
 207     ("v" . "\e$(7"g\e(B")
 208     ("V" . "\e$(7"h\e(B")
 209     ("q" . "\e$(7"k\e(B")
 210     ("Q" . "\e$(7"l\e(B")
 211     ("_/" . "\e$(7!I\e(B")
 212     ("_o" . "\e$(7!g\e(B")
 213     ("_O" . "\e$(7!e\e(B")))
 214
 215 (defconst tibetan-precomposed-transcription-alist
 216   '(("phyw" . "\e$(7$G\e(B")
 217     ("tshw" . "\e$(7$)\e(B")
 218     ("rtsw" . "\e$(7%.\e(B")
 219     ("khw" . "\e$(7$"\e(B")
 220     ("nyw" . "\e$(7$%\e(B")
 221     ("tsw" . "\e$(7$(\e(B")
 222     ("zhw" . "\e$(7$*\e(B")
 223     ("shw" . "\e$(7$.\e(B")
 224     ("khy" . "\e$(7$A\e(B")
 225     ("phy" . "\e$(7$D\e(B")
 226     ("khr" . "\e$(7$Q\e(B")
 227     ("thr" . "\e$(7$T\e(B")
 228     ("phr" . "\e$(7$W\e(B")
 229     ("shr" . "\e$(7$Z\e(B")
 230     ("dzr" . "\e$(7$^\e(B")
 231     ("grw" . "\e$(7$_\e(B")
 232     ("rng" . "\e$(7%#\e(B")
 233     ("rny" . "\e$(7%%\e(B")
 234     ("rts" . "\e$(7%+\e(B")
 235     ("rdz" . "\e$(7%,\e(B")
 236     ("rgw" . "\e$(7%-\e(B")
 237     ("rky" . "\e$(7%0\e(B")
 238     ("rgy" . "\e$(7%1\e(B")
 239     ("rmy" . "\e$(7%2\e(B")
 240     ("lng" . "\e$(7%B\e(B")
 241     ("sng" . "\e$(7%R\e(B")
 242     ("sny" . "\e$(7%S\e(B")
 243     ("sts" . "\e$(7%Z\e(B")
 244     ("sky" . "\e$(7%`\e(B")
 245     ("sgy" . "\e$(7%a\e(B")
 246     ("spy" . "\e$(7%b\e(B")
 247     ("sby" . "\e$(7%c\e(B")
 248     ("smy" . "\e$(7%d\e(B")
 249     ("skr" . "\e$(7%p\e(B")
 250     ("sgr" . "\e$(7%q\e(B")
 251     ("snr" . "\e$(7%r\e(B")
 252     ("spr" . "\e$(7%s\e(B")
 253     ("sbr" . "\e$(7%t\e(B")
 254     ("smr" . "\e$(7%u\e(B")
 255     ("kw" . "\e$(7$!\e(B")
 256     ("gw" . "\e$(7$#\e(B")
 257     ("cw" . "\e$(7$$\e(B")
 258     ("tw" . "\e$(7$&\e(B")
 259     ("dw" . "\e$(7$'\e(B")
 260     ("zw" . "\e$(7$+\e(B")
 261     ("rw" . "\e$(7$,\e(B")
 262     ("lw" . "\e$(7$-\e(B")
 263     ("sw" . "\e$(7$/\e(B")
 264     ("hw" . "\e$(7$0\e(B")
 265     ("ky" . "\e$(7$@\e(B")
 266     ("gy" . "\e$(7$B\e(B")
 267     ("py" . "\e$(7$C\e(B")
 268     ("by" . "\e$(7$E\e(B")
 269     ("my" . "\e$(7$F\e(B")
 270     ("kr" . "\e$(7$P\e(B")
 271     ("gr" . "\e$(7$R\e(B")
 272     ("tr" . "\e$(7$S\e(B")
 273     ("dr" . "\e$(7$U\e(B")
 274     ("pr" . "\e$(7$V\e(B")
 275     ("brk" . "\e$(7"7%!\e(B")
 276     ("brg" . "\e$(7"7%"\e(B")
 277     ("brng" . "\e$(7"7%#\e(B")
 278     ("brj" . "\e$(7"7%$\e(B")
 279     ("brny" . "\e$(7"7%%\e(B")
 280     ("brt" .  "\e$(7"7%&\e(B")
 281     ("brd" . "\e$(7"7%'\e(B")
 282     ("brn" . "\e$(7"7%(\e(B")
 283     ("brts" . "\e$(7"7%+\e(B")
 284     ("brdz" . "\e$(7"7%,\e(B")
 285     ("brl" . "\e$(7"7$d\e(B")
 286     ("br" . "\e$(7$X\e(B")
 287     ("mr" . "\e$(7$Y\e(B")
 288     ("sr" . "\e$(7$[\e(B")
 289     ("hr" . "\e$(7$\\e(B")
 290     ("jr" . "\e$(7$]\e(B")
 291     ("kl" . "\e$(7$`\e(B")
 292     ("gl" . "\e$(7$a\e(B")
 293     ("blt" . "\e$(7"7%E\e(B")
 294     ("bld" . "\e$(7"7%F\e(B")
 295     ("bl" . "\e$(7$b\e(B")
 296     ("zl" . "\e$(7$c\e(B")
 297     ("rl" . "\e$(7$d\e(B")
 298     ("sl" . "\e$(7$e\e(B")
 299     ("rk" . "\e$(7%!\e(B")
 300     ("rg" . "\e$(7%"\e(B")
 301     ("rj" . "\e$(7%$\e(B")
 302     ("rt" . "\e$(7%&\e(B")
 303     ("rd" . "\e$(7%'\e(B")
 304     ("rn" . "\e$(7%(\e(B")
 305     ("rb" . "\e$(7%)\e(B")
 306     ("rm" . "\e$(7%*\e(B")
 307     ("lk" . "\e$(7%@\e(B")
 308     ("lg" . "\e$(7%A\e(B")
 309     ("lc" . "\e$(7%C\e(B")
 310     ("lj" . "\e$(7%D\e(B")
 311     ("lt" . "\e$(7%E\e(B")
 312     ("ld" . "\e$(7%F\e(B")
 313     ("ln" . "\e$(7!!\e(B")                        ; dummy \x2121
 314     ("lp" . "\e$(7%G\e(B")
 315     ("lb" . "\e$(7%H\e(B")
 316     ("lh" . "\e$(7%I\e(B")
 317     ("sk" . "\e$(7%P\e(B")
 318     ("sg" . "\e$(7%Q\e(B")
 319     ("st" . "\e$(7%T\e(B")
 320     ("sd" . "\e$(7%U\e(B")
 321     ("sn" . "\e$(7%V\e(B")
 322     ("sp" . "\e$(7%W\e(B")
 323     ("sb" . "\e$(7%X\e(B")
 324     ("sm" . "\e$(7%Y\e(B"))
 325   )
 326
 327
 328 (defconst tibetan-subjoined-transcription-alist
 329   (sort '(("+k"  . "\e$(7#!\e(B")
 330           ("+kh" . "\e$(7#"\e(B")
 331           ("+g"  . "\e$(7##\e(B")
 332           ("+gh" . "\e$(7#$\e(B")
 333           ("+ng" . "\e$(7#%\e(B")
 334           ("+c"  . "\e$(7#&\e(B")
 335           ("+ch" . "\e$(7#'\e(B")
 336           ("+j"  . "\e$(7#(\e(B")
 337           ("+ny"  . "\e$(7#*\e(B")
 338           ("+T"  . "\e$(7#+\e(B")
 339           ("+TH" . "\e$(7#,\e(B")
 340           ("+D"  . "\e$(7#-\e(B")
 341           ("+DH" . "\e$(7#.\e(B")
 342           ("+N"  . "\e$(7#/\e(B")
 343           ("+t"  . "\e$(7#0\e(B")
 344           ("+th" . "\e$(7#1\e(B")
 345           ("+d"  . "\e$(7#2\e(B")
 346           ("+dh" . "\e$(7#3\e(B")
 347           ("+n"  . "\e$(7#4\e(B")
 348           ("+p"  . "\e$(7#5\e(B")
 349           ("+ph" . "\e$(7#6\e(B")
 350           ("+b"  . "\e$(7#7\e(B")
 351           ("+bh" . "\e$(7#8\e(B")
 352           ("+m"  . "\e$(7#9\e(B")
 353           ("+ts" . "\e$(7#:\e(B")
 354           ("+tsh" . "\e$(7#;\e(B")
 355           ("+dz" . "\e$(7#<\e(B")
 356           ("+dzh" . "\e$(7#=\e(B")
 357           ("+w"  . "\e$(7#>\e(B")
 358           ("+zh" . "\e$(7#?\e(B")
 359           ("+z"  . "\e$(7#@\e(B")
 360           ("+'"  . "\e$(7#A\e(B")
 361           ("+y"  . "\e$(7#B\e(B")
 362           ("+r"  . "\e$(7#C\e(B")
 363           ("+l"  . "\e$(7#D\e(B")
 364           ("+sh" . "\e$(7#E\e(B")
 365           ("+SH" . "\e$(7#F\e(B")
 366           ("+s"  . "\e$(7#G\e(B")
 367           ("+h"  . "\e$(7#H\e(B")
 368           ("+A"  . "\e$(7#I\e(B")
 369           ("+kSH" . "\e$(7#J\e(B")
 370           ;; Added by Tomabechi 1999/12/10
 371           ("+W" . "\e$(7#K\e(B") ;; fixed form subscribed WA
 372           ("+Y" . "\e$(7#L\e(B") ;; fixed form subscribed YA
 373           ("+R" . "\e$(7#M\e(B") ;; fixed form subscribed RA
 374           )
 375         (lambda (x y) (> (length (car x)) (length (car y))))))
 376
 377 ;;;
 378 ;;; alist for Tibetan base consonant <-> subjoined consonant conversion.
 379 ;;;
 380 (defconst tibetan-base-to-subjoined-alist
 381   '(("\e$(7"!\e(B" . "\e$(7#!\e(B")
 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$(7"0\e(B" . "\e$(7#0\e(B")
 396     ("\e$(7"1\e(B" . "\e$(7#1\e(B")
 397     ("\e$(7"2\e(B" . "\e$(7#2\e(B")
 398     ("\e$(7"3\e(B" . "\e$(7#3\e(B")
 399     ("\e$(7"4\e(B" . "\e$(7#4\e(B")
 400     ("\e$(7"5\e(B" . "\e$(7#5\e(B")
 401     ("\e$(7"6\e(B" . "\e$(7#6\e(B")
 402     ("\e$(7"7\e(B" . "\e$(7#7\e(B")
 403     ("\e$(7"8\e(B" . "\e$(7#8\e(B")
 404     ("\e$(7"9\e(B" . "\e$(7#9\e(B")
 405     ("\e$(7":\e(B" . "\e$(7#:\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$(7"A\e(B" . "\e$(7#A\e(B")
 413     ("\e$(7"B\e(B" . "\e$(7#B\e(B")
 414     ("\e$(7"C\e(B" . "\e$(7#C\e(B")
 415     ("\e$(7"D\e(B" . "\e$(7#D\e(B")
 416     ("\e$(7"E\e(B" . "\e$(7#E\e(B")
 417     ("\e$(7"F\e(B" . "\e$(7#F\e(B")
 418     ("\e$(7"G\e(B" . "\e$(7#G\e(B")
 419     ("\e$(7"H\e(B" . "\e$(7#H\e(B")
 420     ("\e$(7"I\e(B" . "\e$(7#I\e(B")
 421     ("\e$(7"J\e(B" . "\e$(7#J\e(B")
 422     ;; Added by Tomabechi 1999/12/10
 423     ("\e$(7"K\e(B" . "\e$(7#M\e(B") ;; Fixed form RA (224B->234D)
 424     ))
 425
 426 ;;; alist for Tibetan composite vowels (long i, vocalic r, etc.)
 427 ;;; New variable. created by Tomabechi 2000/06/08
 428 (defconst tibetan-composite-vowel-alist
 429   '(;; LONG A
 430     ;; ("\e$(7"R\e(B" . ((bc . tc) ?\e$(7"R\e(B))
 431     ;; LONG I
 432     ("\e$(7"T\e(B" . (?\e$(7"R\e(B (tc . bc) ?\e$(7"S\e(B))
 433     ;; LONG U
 434     ("\e$(7"V\e(B" . (?\e$(7"R\e(B (bc . tc) ?\e$(7"U\e(B))
 435     ;; VOCALIC R
 436     ("\e$(7"W\e(B" . (?\e$(7#C\e(B (tc . bc) ?\e$(7"a\e(B))
 437     ;; LONG VOCALIC R
 438     ("\e$(7"X\e(B" . (?\e$(7#C\e(B (bc . tc) ?\e$(7"R\e(B (tc . bc) ?\e$(7"a\e(B))
 439     ;; VOCALIC L
 440     ("\e$(7"Y\e(B" . (?\e$(7#D\e(B (tc . bc) ?\e$(7"a\e(B))
 441     ;;\e$(7!;\e(BLONG VOCALIC L
 442     ("\e$(7"Z\e(B" . (?\e$(7#D\e(B (bc . tc) ?\e$(7"R\e(B (tc . bc) ?\e$(7"a\e(B))
 443     ;; LONG REVERSE I
 444     ("\e$(7"b\e(B" . (?\e$(7"R\e(B (tc . bc) ?\e$(7"a\e(B))
 445     ))
 446
 447
 448
 449 ;;;
 450 ;;; alist for Tibetan consonantic components <-> precomposed glyph conversion.
 451 ;;; (includes some punctuation conversion rules)
 452 ;;;
 453 (defconst tibetan-precomposition-rule-alist
 454   `(("\e$(7"6#B#>\e(B" . "\e$(7$G\e(B")
 455     ("\e$(7"##C#>\e(B" . "\e$(7$_\e(B")
 456     ("\e$(7";#>\e(B" . "\e$(7$)\e(B")
 457     ("\e$(7"C#:#>\e(B" . "\e$(7%.\e(B")
 458     ("\e$(7"C###>\e(B" . "\e$(7%-\e(B")
 459     ("\e$(7"C#!#B\e(B" . "\e$(7%0\e(B")
 460     ("\e$(7"C###B\e(B" . "\e$(7%1\e(B")
 461     ("\e$(7"C#9#B\e(B" . "\e$(7%2\e(B")
 462     ("\e$(7"G#!#B\e(B" . "\e$(7%`\e(B")
 463     ("\e$(7"G###B\e(B" . "\e$(7%a\e(B")
 464     ("\e$(7"G#5#B\e(B" . "\e$(7%b\e(B")
 465     ("\e$(7"G#7#B\e(B" . "\e$(7%c\e(B")
 466     ("\e$(7"G#9#B\e(B" . "\e$(7%d\e(B")
 467     ("\e$(7"G#!#C\e(B" . "\e$(7%p\e(B")
 468     ("\e$(7"G###C\e(B" . "\e$(7%q\e(B")
 469     ("\e$(7"G#4#C\e(B" . "\e$(7%r\e(B")
 470     ("\e$(7"G#5#C\e(B" . "\e$(7%s\e(B")
 471     ("\e$(7"G#7#C\e(B" . "\e$(7%t\e(B")
 472     ("\e$(7"G#9#C\e(B" . "\e$(7%u\e(B")
 473     ("\e$(7""#>\e(B" . "\e$(7$"\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#>\e(B" . "\e$(7$.\e(B")
 478     ("\e$(7""#B\e(B" . "\e$(7$A\e(B")
 479     ("\e$(7"6#B\e(B" . "\e$(7$D\e(B")
 480     ("\e$(7""#C\e(B" . "\e$(7$Q\e(B")
 481     ("\e$(7"1#C\e(B" . "\e$(7$T\e(B")
 482     ("\e$(7"6#C\e(B" . "\e$(7$W\e(B")
 483     ("\e$(7"E#C\e(B" . "\e$(7$Z\e(B")
 484     ("\e$(7"<#C\e(B" . "\e$(7$^\e(B")
 485     ("\e$(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"D#%\e(B" . "\e$(7%B\e(B")
 490     ("\e$(7"G#%\e(B" . "\e$(7%R\e(B")
 491     ("\e$(7"G#*\e(B" . "\e$(7%S\e(B")
 492     ("\e$(7"G#:\e(B" . "\e$(7%Z\e(B")
 493     ("\e$(7"!#>\e(B" . "\e$(7$!\e(B")
 494     ("\e$(7"##>\e(B" . "\e$(7$#\e(B")
 495     ("\e$(7"&#>\e(B" . "\e$(7$$\e(B")
 496     ("\e$(7"0#>\e(B" . "\e$(7$&\e(B")
 497     ("\e$(7"2#>\e(B" . "\e$(7$'\e(B")
 498     ("\e$(7"@#>\e(B" . "\e$(7$+\e(B")
 499     ("\e$(7"C#>\e(B" . "\e$(7$,\e(B")
 500     ("\e$(7"D#>\e(B" . "\e$(7$-\e(B")
 501     ("\e$(7"G#>\e(B" . "\e$(7$/\e(B")
 502     ("\e$(7"H#>\e(B" . "\e$(7$0\e(B")
 503     ("\e$(7"!#B\e(B" . "\e$(7$@\e(B")
 504     ("\e$(7"##B\e(B" . "\e$(7$B\e(B")
 505     ("\e$(7"5#B\e(B" . "\e$(7$C\e(B")
 506     ("\e$(7"7#B\e(B" . "\e$(7$E\e(B")
 507     ("\e$(7"9#B\e(B" . "\e$(7$F\e(B")
 508     ("\e$(7"!#C\e(B" . "\e$(7$P\e(B")
 509     ("\e$(7"##C\e(B" . "\e$(7$R\e(B")
 510     ("\e$(7"0#C\e(B" . "\e$(7$S\e(B")
 511     ("\e$(7"2#C\e(B" . "\e$(7$U\e(B")
 512     ("\e$(7"5#C\e(B" . "\e$(7$V\e(B")
 513     ("\e$(7"7#C\e(B" . "\e$(7$X\e(B")
 514     ("\e$(7"9#C\e(B" . "\e$(7$Y\e(B")
 515     ("\e$(7"G#C\e(B" . "\e$(7$[\e(B")
 516     ("\e$(7"H#C\e(B" . "\e$(7$\\e(B")
 517     ("\e$(7"(#C\e(B" . "\e$(7$]\e(B")
 518     ("\e$(7"!#D\e(B" . "\e$(7$`\e(B")
 519     ("\e$(7"##D\e(B" . "\e$(7$a\e(B")
 520     ("\e$(7"7#D\e(B" . "\e$(7$b\e(B")
 521     ("\e$(7"@#D\e(B" . "\e$(7$c\e(B")
 522     ("\e$(7"C#D\e(B" . "\e$(7$d\e(B")
 523     ("\e$(7"G#D\e(B" . "\e$(7$e\e(B")
 524     ("\e$(7"C#!\e(B" . "\e$(7%!\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#0\e(B" . "\e$(7%&\e(B")
 528     ("\e$(7"C#2\e(B" . "\e$(7%'\e(B")
 529     ("\e$(7"C#4\e(B" . "\e$(7%(\e(B")
 530     ("\e$(7"C#7\e(B" . "\e$(7%)\e(B")
 531     ("\e$(7"C#9\e(B" . "\e$(7%*\e(B")
 532     ("\e$(7"D#!\e(B" . "\e$(7%@\e(B")
 533     ("\e$(7"D##\e(B" . "\e$(7%A\e(B")
 534     ("\e$(7"D#4\e(B" . "\e$(7!!\e(B") ; dummy 0x2121 added 2000/06/08 for transition l -> lng
 535     ("\e$(7"D#&\e(B" . "\e$(7%C\e(B")
 536     ("\e$(7"D#(\e(B" . "\e$(7%D\e(B")
 537     ("\e$(7"D#0\e(B" . "\e$(7%E\e(B")
 538     ("\e$(7"D#2\e(B" . "\e$(7%F\e(B")
 539     ("\e$(7"D#5\e(B" . "\e$(7%G\e(B")
 540     ("\e$(7"D#7\e(B" . "\e$(7%H\e(B")
 541     ("\e$(7"D#H\e(B" . "\e$(7%I\e(B")
 542     ("\e$(7"G#!\e(B" . "\e$(7%P\e(B")
 543     ("\e$(7"G##\e(B" . "\e$(7%Q\e(B")
 544     ("\e$(7"G#0\e(B" . "\e$(7%T\e(B")
 545     ("\e$(7"G#2\e(B" . "\e$(7%U\e(B")
 546     ("\e$(7"G#4\e(B" . "\e$(7%V\e(B")
 547     ("\e$(7"G#5\e(B" . "\e$(7%W\e(B")
 548     ("\e$(7"G#7\e(B" . "\e$(7%X\e(B")
 549     ("\e$(7"G#9\e(B" . "\e$(7%Y\e(B")))
 550
 551 (defconst tibetan-regexp
 552   (let ((l (list tibetan-precomposed-transcription-alist
 553                  tibetan-consonant-transcription-alist
 554                  tibetan-vowel-transcription-alist
 555                  tibetan-modifier-transcription-alist
 556                  tibetan-subjoined-transcription-alist))
 557         (separator "\\|")
 558         tail pattern)
 559     (while l
 560       (setq tail (car l) l (cdr l))
 561       (while tail
 562         (setq pattern (cons separator (cons (car (car tail)) pattern))
 563               tail (cdr tail))))
 564     (apply 'concat (nreverse (cdr pattern))))
 565   "Regexp matching a Tibetan transcription of a composable Tibetan sequence.
 566 The result of matching is to be used for indexing alists at conversion
 567 from a roman transcription to the corresponding Tibetan character.")
 568
 569 (defvar tibetan-precomposed-regexp
 570   (purecopy
 571   (let ((l tibetan-precomposed-transcription-alist)
 572         temp)
 573     (setq temp "^\\(")
 574     (setq temp
 575           (concat temp (car (car l))))
 576     (setq l (cdr l))
 577     (while l
 578       (setq temp
 579             (concat temp "\\|" (car (car l))))
 580       (setq l (cdr l)))
 581     (concat temp "\\)")))
 582   "Regexp string to match a romanized Tibetan complex consonant.
 583 The result of matching is to be used for indexing alists when the input key
 584 from an input method is converted to the corresponding precomposed glyph.")
 585
 586 (defvar tibetan-precomposition-rule-regexp
 587   (purecopy
 588   (let ((l tibetan-precomposition-rule-alist)
 589         temp)
 590     (setq temp "\\(")
 591     (setq temp (concat temp (car (car l))))
 592     (setq l (cdr l))
 593     (while l
 594       (setq temp (concat temp "\\|" (car (car l))))
 595       (setq l (cdr l)))
 596     (concat temp "\\)")))
 597   "Regexp string to match a sequence of Tibetan consonantic components, i.e.,
 598 one base consonant and one or more subjoined consonants.
 599 The result of matching is to be used for indexing alist when the component
 600 sequence is converted to the corresponding precomposed glyph.
 601 This also matches some punctuation characters which need conversion.")
 602
 603 (defvar tibetan-decomposed nil)
 604 (defvar tibetan-decomposed-temp nil)
 605
 606 ;; For automatic composition.
 607 (set-char-table-range
 608  composition-function-table '(#xF00 . #xFD1)
 609  (list (vector tibetan-composable-pattern 0 'font-shape-gstring)))
 610
 611 (provide 'tibetan)
 612
 613 ;;; tibetan.el ends here