lisp/language/tibetan.el

   1 ;;; tibetan.el --- support for Tibetan language -*- coding: iso-2022-7bit; -*-
   2
   3 ;; Copyright (C) 1997, 2002  Free Software Foundation, Inc.
   4 ;; Copyright (C) 1997, 1999, 2000
   5 ;;   National Institute of Advanced Industrial Science and Technology (AIST)
   6 ;;   Registration Number H14PRO021
   7
   8 ;; Keywords: multilingual, Tibetan
   9
  10 ;; This file is part of GNU Emacs.
  11
  12 ;; GNU Emacs is free software; you can redistribute it and/or modify
  13 ;; it under the terms of the GNU General Public License as published by
  14 ;; the Free Software Foundation; either version 2, or (at your option)
  15 ;; any later version.
  16
  17 ;; GNU Emacs is distributed in the hope that it will be useful,
  18 ;; but WITHOUT ANY WARRANTY; without even the implied warranty of
  19 ;; MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
  20 ;; GNU General Public License for more details.
  21
  22 ;; You should have received a copy of the GNU General Public License
  23 ;; along with GNU Emacs; see the file COPYING.  If not, write to the
  24 ;; Free Software Foundation, Inc., 59 Temple Place - Suite 330,
  25 ;; Boston, MA 02111-1307, USA.
  26
  27 ;; Author: Toru TOMABECHI, <Toru.Tomabechi@orient.unil.ch>
  28
  29 ;; Created: Feb. 17. 1997
  30
  31 ;; History:
  32 ;; 1997.03.13 Modification for special signs and punctuations.
  33
  34 ;;; Commentary:
  35
  36 ;;; Code:
  37
  38 ;;; Tibetan Character set.
  39 ;;; \x2130 -- \x234a is a subset of Unicode v.2 \x0f00 - \x0fb9
  40 ;;; with a slight modification. And there are some subjoined
  41 ;;; consonants which are not specified in Unicode.
  42 ;;; I hope I can add missing characters later.
  43 ;;;
  44 ;;;     00 01 02 03 04 05 06 07 08 09 0A 0B 0C 0D 0E 0F
  45 ;;;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)
  46 ;;;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,
  47 ;;;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
  48 ;;;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.
  49 ;;;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 ;
  50 ;;;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$(7!~\e(B // ;
  51 ;;;
  52 ;;;     00 01 02 03 04 05 06 07 08 09 0A 0B 0C 0D 0E 0F
  53 ;;;2220 // \e$(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$(7"/\e(B ; Base consonants
  54 ;;;2230 \e$(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$(7"?\e(B ; and
  55 ;;;2240 \e$(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$(7"O\e(B ; Vowel signs.
  56 ;;;2250 \e$(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$(7"_\e(B ; (\x2251 = vowel a)
  57 ;;;2260 \e$(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$(7"o\e(B ; Long vowels and
  58 ;;;2270 \e$(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
  59 ;;;                                                     ; atomically
  60 ;;;                                                     ; encoded.
  61 ;;;     00 01 02 03 04 05 06 07 08 09 0A 0B 0C 0D 0E 0F
  62 ;;;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
  63 ;;;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 ;
  64 ;;;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 ;
  65 ;;;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
  66 ;;;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
  67 ;;;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$(7#~\e(B // ; in Unicode.
  68 ;;;
  69 ;;;     00 01 02 03 04 05 06 07 08 09 0A 0B 0C 0D 0E 0F
  70 ;;;2420 // \e$(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
  71 ;;;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
  72 ;;;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.
  73 ;;;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
  74 ;;;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
  75 ;;;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
  76 ;;;                                                     ; when written on a
  77 ;;;     00 01 02 03 04 05 06 07 08 09 0A 0B 0C 0D 0E 0F ; file in Tibetan
  78 ;;;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.
  79 ;;;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 ;
  80 ;;;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 ;
  81 ;;;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 ;
  82 ;;;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 ;
  83 ;;;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 // ;
  84 ;;;
  85
  86
  87 (make-coding-system
  88  'tibetan-iso-8bit 2 ?Q
  89  "8-bit encoding for ASCII (MSB=0) and TIBETAN (MSB=1)."
  90  '(ascii tibetan nil nil
  91    nil nil)
  92  '((safe-charsets ascii tibetan)
  93    (post-read-conversion . tibetan-post-read-conversion)
  94    (pre-write-conversion . tibetan-pre-write-conversion)))
  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
 106               . (tibetan-compose-string
 107                  (copy-sequence
 108 "Tibetan (\e4\e$(7"7r'"]\e0"7"]\e1\e4"2\e0"2\e1!;\e4%P\e0"G#!"Q\e1\e4"2\e0"2\e1!;\e(B) \e$(7!4!5!5!>\e4"7\e0"7\e1\e4$P\e0"!#C"Q\e1!;\e4"Er'"S\e0"E"S\e1\e4"G\e0"G\e1!;\e4"7\e0"7\e1\e4"2r'"[\e0"2"[\e1!;\e4"Dr'"[\e0"D"[\e1\e4"#\e0"#\e1\e4"G\e0"G\e1!>\e4"Ir'"]r'"_\e0"I"]"_\e1!;\e4"9\e0"9"Q\e1!;\e4"/r'"S\e0"/"S\e1!;\e4"5\e0"5"Q\e1\e4#2x!#9r'"[\e0"2#9"[\e1!;\e4"Hx!"Rx!"Ur'"c\e0"H"A"U"c\e1!>\e(B")))))
 109
 110
 111 ;; `\e$(7"A\e(B' is included in the pattern for subjoined consonants because we
 112 ;; treat it specially in tibetan-add-components.
 113 ;; modified by Tomabechi 1999/12/10
 114 ;; modified by Tomabechi 2000/06/08
 115 ;;          To allow infinite addition of vowels/modifiers
 116 ;;          as specified in Unicode v.3
 117 ;; \e$(7"A\e(B is removed from the class of subjoined. Tomabechi 2000/06/08
 118 ;; (for Unicode support)
 119 (defconst tibetan-composable-pattern
 120   "[\e$(7"!\e(B-\e$(7"J"K\e(B][\e$(7#!\e(B-\e$(7#J#K#L#M\e(B]*[\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]*"
 121   "Regexp matching a composable sequence of Tibetan characters.")
 122
 123 ;; Register a function to compose Tibetan characters.
 124 (aset composition-function-table (make-char 'tibetan)
 125       (list (cons tibetan-composable-pattern 'tibetan-composition-function)))
 126
 127 ;;;
 128 ;;; Definitions of conversion data.
 129 ;;;
 130
 131
 132 ;;; alists for tibetan char <-> transcription conversion
 133 ;;; longer transcription should come first
 134 (defconst tibetan-consonant-transcription-alist
 135   '(("tsh" . "\e$(7";\e(B")
 136     ("dzh" . "\e$(7"=\e(B")
 137     ("kSH" . "\e$(7"J\e(B")
 138     ("kh" . "\e$(7""\e(B")
 139     ("gh" . "\e$(7"$\e(B")
 140     ("ng" . "\e$(7"%\e(B")
 141     ("ch" . "\e$(7"'\e(B")
 142     ("ny" . "\e$(7"*\e(B")
 143     ("TH" . "\e$(7",\e(B")
 144     ("DH" . "\e$(7".\e(B")
 145     ("th" . "\e$(7"1\e(B")
 146     ("dh" . "\e$(7"3\e(B")
 147     ("ph" . "\e$(7"6\e(B")
 148     ("bh" . "\e$(7"8\e(B")
 149     ("ts" . "\e$(7":\e(B")
 150     ("dz" . "\e$(7"<\e(B")
 151     ("zh" . "\e$(7"?\e(B")
 152     ("sh" . "\e$(7"E\e(B")
 153     ("SH" . "\e$(7"F\e(B")
 154     ("k" . "\e$(7"!\e(B")
 155     ("g" . "\e$(7"#\e(B")
 156     ("c" . "\e$(7"&\e(B")
 157     ("j" . "\e$(7"(\e(B")
 158     ("T" . "\e$(7"+\e(B")
 159     ("D" . "\e$(7"-\e(B")
 160     ("N" . "\e$(7"/\e(B")
 161     ("t" . "\e$(7"0\e(B")
 162     ("d" . "\e$(7"2\e(B")
 163     ("n" . "\e$(7"4\e(B")
 164     ("p" . "\e$(7"5\e(B")
 165     ("b" . "\e$(7"7\e(B")
 166     ("m" . "\e$(7"9\e(B")
 167     ("w" . "\e$(7">\e(B")
 168     ("z" . "\e$(7"@\e(B")
 169     ("'" . "\e$(7"A\e(B")
 170     ("y" . "\e$(7"B\e(B")
 171     ("r" . "\e$(7"C\e(B")
 172     ("l" . "\e$(7"D\e(B")
 173     ("s" . "\e$(7"G\e(B")
 174     ("h" . "\e$(7"H\e(B")
 175     ("H" . "\e$(7"H\e(B")
 176     ("A" . "\e$(7"I\e(B")
 177     ;; Added by Tomabechi 1999/12/10
 178     ("R" . "\e$(7"K\e(B") ;; fixed form RA
 179     ))
 180
 181
 182 (defconst tibetan-vowel-transcription-alist
 183   '(
 184     ;; Composite Vowels
 185     ;; Added by Tomabechi 2000/06/08
 186     ("frr" . "\e$(7"X\e(B")
 187     ("fll" . "\e$(7"Z\e(B")
 188     ("fa" . "\e$(7"R\e(B")
 189     ("fi" . "\e$(7"T\e(B")
 190     ("fu" . "\e$(7"V\e(B")
 191     ("fr" . "\e$(7"W\e(B")
 192     ("fl" . "\e$(7"Y\e(B")
 193     ("fI" . "\e$(7"b\e(B")
 194     ;; Normal Vowels
 195     ("ai" . "\e$(7"\\e(B")
 196     ("au" . "\e$(7"^\e(B")
 197     ("ee" . "\e$(7"\\e(B")
 198     ("oo" . "\e$(7"^\e(B")
 199     ("a" . "\e$(7"Q\e(B")                 ; invisible vowel sign (\x2251)
 200     ("i" . "\e$(7"S\e(B")
 201     ("u" . "\e$(7"U\e(B")
 202     ("e" . "\e$(7"[\e(B")
 203     ("o" . "\e$(7"]\e(B")
 204     ("E" . "\e$(7"\\e(B")
 205     ("O" . "\e$(7"^\e(B")
 206     ("I" . "\e$(7"a\e(B")
 207     ("," . "\e$(7"e\e(B")                 ; idem.
 208     ))
 209
 210 (defconst tibetan-modifier-transcription-alist
 211   '(("M" . "\e$(7"_\e(B")
 212     ("~" . "\e$(7"c\e(B")
 213     ("`" . "\e$(7"d\e(B")
 214     ("x" . "\e$(7"i\e(B")
 215     ("X" . "\e$(7"j\e(B")
 216     ("v" . "\e$(7"g\e(B")
 217     ("V" . "\e$(7"h\e(B")
 218     ("q" . "\e$(7"k\e(B")
 219     ("Q" . "\e$(7"l\e(B")
 220     ("_/" . "\e$(7!I\e(B")
 221     ("_o" . "\e$(7!g\e(B")
 222     ("_O" . "\e$(7!e\e(B")))
 223
 224 (defconst tibetan-precomposed-transcription-alist
 225   '(("phyw" . "\e$(7$G\e(B")
 226     ("tshw" . "\e$(7$)\e(B")
 227     ("rtsw" . "\e$(7%.\e(B")
 228     ("khw" . "\e$(7$"\e(B")
 229     ("nyw" . "\e$(7$%\e(B")
 230     ("tsw" . "\e$(7$(\e(B")
 231     ("zhw" . "\e$(7$*\e(B")
 232     ("shw" . "\e$(7$.\e(B")
 233     ("khy" . "\e$(7$A\e(B")
 234     ("phy" . "\e$(7$D\e(B")
 235     ("khr" . "\e$(7$Q\e(B")
 236     ("thr" . "\e$(7$T\e(B")
 237     ("phr" . "\e$(7$W\e(B")
 238     ("shr" . "\e$(7$Z\e(B")
 239     ("dzr" . "\e$(7$^\e(B")
 240     ("grw" . "\e$(7$_\e(B")
 241     ("rng" . "\e$(7%#\e(B")
 242     ("rny" . "\e$(7%%\e(B")
 243     ("rts" . "\e$(7%+\e(B")
 244     ("rdz" . "\e$(7%,\e(B")
 245     ("rgw" . "\e$(7%-\e(B")
 246     ("rky" . "\e$(7%0\e(B")
 247     ("rgy" . "\e$(7%1\e(B")
 248     ("rmy" . "\e$(7%2\e(B")
 249     ("lng" . "\e$(7%B\e(B")
 250     ("sng" . "\e$(7%R\e(B")
 251     ("sny" . "\e$(7%S\e(B")
 252     ("sts" . "\e$(7%Z\e(B")
 253     ("sky" . "\e$(7%`\e(B")
 254     ("sgy" . "\e$(7%a\e(B")
 255     ("spy" . "\e$(7%b\e(B")
 256     ("sby" . "\e$(7%c\e(B")
 257     ("smy" . "\e$(7%d\e(B")
 258     ("skr" . "\e$(7%p\e(B")
 259     ("sgr" . "\e$(7%q\e(B")
 260     ("snr" . "\e$(7%r\e(B")
 261     ("spr" . "\e$(7%s\e(B")
 262     ("sbr" . "\e$(7%t\e(B")
 263     ("smr" . "\e$(7%u\e(B")
 264     ("kw" . "\e$(7$!\e(B")
 265     ("gw" . "\e$(7$#\e(B")
 266     ("cw" . "\e$(7$$\e(B")
 267     ("tw" . "\e$(7$&\e(B")
 268     ("dw" . "\e$(7$'\e(B")
 269     ("zw" . "\e$(7$+\e(B")
 270     ("rw" . "\e$(7$,\e(B")
 271     ("lw" . "\e$(7$-\e(B")
 272     ("sw" . "\e$(7$/\e(B")
 273     ("hw" . "\e$(7$0\e(B")
 274     ("ky" . "\e$(7$@\e(B")
 275     ("gy" . "\e$(7$B\e(B")
 276     ("py" . "\e$(7$C\e(B")
 277     ("by" . "\e$(7$E\e(B")
 278     ("my" . "\e$(7$F\e(B")
 279     ("kr" . "\e$(7$P\e(B")
 280     ("gr" . "\e$(7$R\e(B")
 281     ("tr" . "\e$(7$S\e(B")
 282     ("dr" . "\e$(7$U\e(B")
 283     ("pr" . "\e$(7$V\e(B")
 284     ("brk" . "\e$(7"7%!\e(B")
 285     ("brg" . "\e$(7"7%"\e(B")
 286     ("brng" . "\e$(7"7%#\e(B")
 287     ("brj" . "\e$(7"7%$\e(B")
 288     ("brny" . "\e$(7"7%%\e(B")
 289     ("brt" .  "\e$(7"7%&\e(B")
 290     ("brd" . "\e$(7"7%'\e(B")
 291     ("brn" . "\e$(7"7%(\e(B")
 292     ("brts" . "\e$(7"7%+\e(B")
 293     ("brdz" . "\e$(7"7%,\e(B")
 294     ("brl" . "\e$(7"7$d\e(B")
 295     ("br" . "\e$(7$X\e(B")
 296     ("mr" . "\e$(7$Y\e(B")
 297     ("sr" . "\e$(7$[\e(B")
 298     ("hr" . "\e$(7$\\e(B")
 299     ("jr" . "\e$(7$]\e(B")
 300     ("kl" . "\e$(7$`\e(B")
 301     ("gl" . "\e$(7$a\e(B")
 302     ("blt" . "\e$(7"7%E\e(B")
 303     ("bld" . "\e$(7"7%F\e(B")
 304     ("bl" . "\e$(7$b\e(B")
 305     ("zl" . "\e$(7$c\e(B")
 306     ("rl" . "\e$(7$d\e(B")
 307     ("sl" . "\e$(7$e\e(B")
 308     ("rk" . "\e$(7%!\e(B")
 309     ("rg" . "\e$(7%"\e(B")
 310     ("rj" . "\e$(7%$\e(B")
 311     ("rt" . "\e$(7%&\e(B")
 312     ("rd" . "\e$(7%'\e(B")
 313     ("rn" . "\e$(7%(\e(B")
 314     ("rb" . "\e$(7%)\e(B")
 315     ("rm" . "\e$(7%*\e(B")
 316     ("lk" . "\e$(7%@\e(B")
 317     ("lg" . "\e$(7%A\e(B")
 318     ("lc" . "\e$(7%C\e(B")
 319     ("lj" . "\e$(7%D\e(B")
 320     ("lt" . "\e$(7%E\e(B")
 321     ("ld" . "\e$(7%F\e(B")
 322     ("ln" . "\e$(7!!\e(B")                        ; dummy \x2121
 323     ("lp" . "\e$(7%G\e(B")
 324     ("lb" . "\e$(7%H\e(B")
 325     ("lh" . "\e$(7%I\e(B")
 326     ("sk" . "\e$(7%P\e(B")
 327     ("sg" . "\e$(7%Q\e(B")
 328     ("st" . "\e$(7%T\e(B")
 329     ("sd" . "\e$(7%U\e(B")
 330     ("sn" . "\e$(7%V\e(B")
 331     ("sp" . "\e$(7%W\e(B")
 332     ("sb" . "\e$(7%X\e(B")
 333     ("sm" . "\e$(7%Y\e(B"))
 334   )
 335
 336
 337 (defconst tibetan-subjoined-transcription-alist
 338   (sort '(("+k"  . "\e$(7#!\e(B")
 339           ("+kh" . "\e$(7#"\e(B")
 340           ("+g"  . "\e$(7##\e(B")
 341           ("+gh" . "\e$(7#$\e(B")
 342           ("+ng" . "\e$(7#%\e(B")
 343           ("+c"  . "\e$(7#&\e(B")
 344           ("+ch" . "\e$(7#'\e(B")
 345           ("+j"  . "\e$(7#(\e(B")
 346           ("+ny"  . "\e$(7#*\e(B")
 347           ("+T"  . "\e$(7#+\e(B")
 348           ("+TH" . "\e$(7#,\e(B")
 349           ("+D"  . "\e$(7#-\e(B")
 350           ("+DH" . "\e$(7#.\e(B")
 351           ("+N"  . "\e$(7#/\e(B")
 352           ("+t"  . "\e$(7#0\e(B")
 353           ("+th" . "\e$(7#1\e(B")
 354           ("+d"  . "\e$(7#2\e(B")
 355           ("+dh" . "\e$(7#3\e(B")
 356           ("+n"  . "\e$(7#4\e(B")
 357           ("+p"  . "\e$(7#5\e(B")
 358           ("+ph" . "\e$(7#6\e(B")
 359           ("+b"  . "\e$(7#7\e(B")
 360           ("+bh" . "\e$(7#8\e(B")
 361           ("+m"  . "\e$(7#9\e(B")
 362           ("+ts" . "\e$(7#:\e(B")
 363           ("+tsh" . "\e$(7#;\e(B")
 364           ("+dz" . "\e$(7#<\e(B")
 365           ("+dzh" . "\e$(7#=\e(B")
 366           ("+w"  . "\e$(7#>\e(B")
 367           ("+zh" . "\e$(7#?\e(B")
 368           ("+z"  . "\e$(7#@\e(B")
 369           ("+'"  . "\e$(7#A\e(B")
 370           ("+y"  . "\e$(7#B\e(B")
 371           ("+r"  . "\e$(7#C\e(B")
 372           ("+l"  . "\e$(7#D\e(B")
 373           ("+sh" . "\e$(7#E\e(B")
 374           ("+SH" . "\e$(7#F\e(B")
 375           ("+s"  . "\e$(7#G\e(B")
 376           ("+h"  . "\e$(7#H\e(B")
 377           ("+A"  . "\e$(7#I\e(B")
 378           ("+kSH" . "\e$(7#J\e(B")
 379           ;; Added by Tomabechi 1999/12/10
 380           ("+W" . "\e$(7#K\e(B") ;; fixed form subscribed WA
 381           ("+Y" . "\e$(7#L\e(B") ;; fixed form subscribed YA
 382           ("+R" . "\e$(7#M\e(B") ;; fixed form subscribed RA
 383           )
 384         (lambda (x y) (> (length (car x)) (length (car y))))))
 385
 386 ;;;
 387 ;;; alist for Tibetan base consonant <-> subjoined consonant conversion.
 388 ;;;
 389 (defconst tibetan-base-to-subjoined-alist
 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$(7"'\e(B" . "\e$(7#'\e(B")
 397     ("\e$(7"(\e(B" . "\e$(7#(\e(B")
 398     ("\e$(7"*\e(B" . "\e$(7#*\e(B")
 399     ("\e$(7"+\e(B" . "\e$(7#+\e(B")
 400     ("\e$(7",\e(B" . "\e$(7#,\e(B")
 401     ("\e$(7"-\e(B" . "\e$(7#-\e(B")
 402     ("\e$(7".\e(B" . "\e$(7#.\e(B")
 403     ("\e$(7"/\e(B" . "\e$(7#/\e(B")
 404     ("\e$(7"0\e(B" . "\e$(7#0\e(B")
 405     ("\e$(7"1\e(B" . "\e$(7#1\e(B")
 406     ("\e$(7"2\e(B" . "\e$(7#2\e(B")
 407     ("\e$(7"3\e(B" . "\e$(7#3\e(B")
 408     ("\e$(7"4\e(B" . "\e$(7#4\e(B")
 409     ("\e$(7"5\e(B" . "\e$(7#5\e(B")
 410     ("\e$(7"6\e(B" . "\e$(7#6\e(B")
 411     ("\e$(7"7\e(B" . "\e$(7#7\e(B")
 412     ("\e$(7"8\e(B" . "\e$(7#8\e(B")
 413     ("\e$(7"9\e(B" . "\e$(7#9\e(B")
 414     ("\e$(7":\e(B" . "\e$(7#:\e(B")
 415     ("\e$(7";\e(B" . "\e$(7#;\e(B")
 416     ("\e$(7"<\e(B" . "\e$(7#<\e(B")
 417     ("\e$(7"=\e(B" . "\e$(7#=\e(B")
 418     ("\e$(7">\e(B" . "\e$(7#>\e(B")
 419     ("\e$(7"?\e(B" . "\e$(7#?\e(B")
 420     ("\e$(7"@\e(B" . "\e$(7#@\e(B")
 421     ("\e$(7"A\e(B" . "\e$(7#A\e(B")
 422     ("\e$(7"B\e(B" . "\e$(7#B\e(B")
 423     ("\e$(7"C\e(B" . "\e$(7#C\e(B")
 424     ("\e$(7"D\e(B" . "\e$(7#D\e(B")
 425     ("\e$(7"E\e(B" . "\e$(7#E\e(B")
 426     ("\e$(7"F\e(B" . "\e$(7#F\e(B")
 427     ("\e$(7"G\e(B" . "\e$(7#G\e(B")
 428     ("\e$(7"H\e(B" . "\e$(7#H\e(B")
 429     ("\e$(7"I\e(B" . "\e$(7#I\e(B")
 430     ("\e$(7"J\e(B" . "\e$(7#J\e(B")
 431     ;; Added by Tomabechi 1999/12/10
 432     ("\e$(7"K\e(B" . "\e$(7#M\e(B") ;; Fixed form RA (224B->234D)
 433     ))
 434
 435 ;;; alist for Tibetan composite vowels (long i, vocalic r, etc.)
 436 ;;; New varialble. created by Tomabechi 2000/06/08
 437 (defconst tibetan-composite-vowel-alist
 438   '(;; LONG A
 439     ;; ("\e$(7"R\e(B" . ((bc . tc) ?\e$(7"R\e(B))
 440     ;; LONG I
 441     ("\e$(7"T\e(B" . (?\e$(7"R\e(B (tc . bc) ?\e$(7"S\e(B))
 442     ;; LONG U
 443     ("\e$(7"V\e(B" . (?\e$(7"R\e(B (bc . tc) ?\e$(7"U\e(B))
 444     ;; VOCALIC R
 445     ("\e$(7"W\e(B" . (?\e$(7#C\e(B (tc . bc) ?\e$(7"a\e(B))
 446     ;; LONG VOCALIC R
 447     ("\e$(7"X\e(B" . (?\e$(7#C\e(B (bc . tc) ?\e$(7"R\e(B (tc . bc) ?\e$(7"a\e(B))
 448     ;; VOCALIC L
 449     ("\e$(7"Y\e(B" . (?\e$(7#D\e(B (tc . bc) ?\e$(7"a\e(B))
 450     ;;\e$(7!;\e(BLONG VOCALIC L
 451     ("\e$(7"Z\e(B" . (?\e$(7#D\e(B (bc . tc) ?\e$(7"R\e(B (tc . bc) ?\e$(7"a\e(B))
 452     ;; LONG REVERSE I
 453     ("\e$(7"b\e(B" . (?\e$(7"R\e(B (tc . bc) ?\e$(7"a\e(B))
 454     ))
 455
 456
 457
 458 ;;;
 459 ;;; alist for Tibetan consonantic components <-> precomposed glyph conversion.
 460 ;;; (includes some punctuation conversion rules)
 461 ;;;
 462 (defconst tibetan-precomposition-rule-alist
 463   `(("\e$(7"6#B#>\e(B" . "\e$(7$G\e(B")
 464     ("\e$(7"##C#>\e(B" . "\e$(7$_\e(B")
 465     ("\e$(7";#>\e(B" . "\e$(7$)\e(B")
 466     ("\e$(7"C#:#>\e(B" . "\e$(7%.\e(B")
 467     ("\e$(7"C###>\e(B" . "\e$(7%-\e(B")
 468     ("\e$(7"C#!#B\e(B" . "\e$(7%0\e(B")
 469     ("\e$(7"C###B\e(B" . "\e$(7%1\e(B")
 470     ("\e$(7"C#9#B\e(B" . "\e$(7%2\e(B")
 471     ("\e$(7"G#!#B\e(B" . "\e$(7%`\e(B")
 472     ("\e$(7"G###B\e(B" . "\e$(7%a\e(B")
 473     ("\e$(7"G#5#B\e(B" . "\e$(7%b\e(B")
 474     ("\e$(7"G#7#B\e(B" . "\e$(7%c\e(B")
 475     ("\e$(7"G#9#B\e(B" . "\e$(7%d\e(B")
 476     ("\e$(7"G#!#C\e(B" . "\e$(7%p\e(B")
 477     ("\e$(7"G###C\e(B" . "\e$(7%q\e(B")
 478     ("\e$(7"G#4#C\e(B" . "\e$(7%r\e(B")
 479     ("\e$(7"G#5#C\e(B" . "\e$(7%s\e(B")
 480     ("\e$(7"G#7#C\e(B" . "\e$(7%t\e(B")
 481     ("\e$(7"G#9#C\e(B" . "\e$(7%u\e(B")
 482     ("\e$(7""#>\e(B" . "\e$(7$"\e(B")
 483     ("\e$(7"*#>\e(B" . "\e$(7$%\e(B")
 484     ("\e$(7":#>\e(B" . "\e$(7$(\e(B")
 485     ("\e$(7"?#>\e(B" . "\e$(7$*\e(B")
 486     ("\e$(7"E#>\e(B" . "\e$(7$.\e(B")
 487     ("\e$(7""#B\e(B" . "\e$(7$A\e(B")
 488     ("\e$(7"6#B\e(B" . "\e$(7$D\e(B")
 489     ("\e$(7""#C\e(B" . "\e$(7$Q\e(B")
 490     ("\e$(7"1#C\e(B" . "\e$(7$T\e(B")
 491     ("\e$(7"6#C\e(B" . "\e$(7$W\e(B")
 492     ("\e$(7"E#C\e(B" . "\e$(7$Z\e(B")
 493     ("\e$(7"<#C\e(B" . "\e$(7$^\e(B")
 494     ("\e$(7"C#%\e(B" . "\e$(7%#\e(B")
 495     ("\e$(7"C#*\e(B" . "\e$(7%%\e(B")
 496     ("\e$(7"C#:\e(B" . "\e$(7%+\e(B")
 497     ("\e$(7"C#<\e(B" . "\e$(7%,\e(B")
 498     ("\e$(7"D#%\e(B" . "\e$(7%B\e(B")
 499     ("\e$(7"G#%\e(B" . "\e$(7%R\e(B")
 500     ("\e$(7"G#*\e(B" . "\e$(7%S\e(B")
 501     ("\e$(7"G#:\e(B" . "\e$(7%Z\e(B")
 502     ("\e$(7"!#>\e(B" . "\e$(7$!\e(B")
 503     ("\e$(7"##>\e(B" . "\e$(7$#\e(B")
 504     ("\e$(7"&#>\e(B" . "\e$(7$$\e(B")
 505     ("\e$(7"0#>\e(B" . "\e$(7$&\e(B")
 506     ("\e$(7"2#>\e(B" . "\e$(7$'\e(B")
 507     ("\e$(7"@#>\e(B" . "\e$(7$+\e(B")
 508     ("\e$(7"C#>\e(B" . "\e$(7$,\e(B")
 509     ("\e$(7"D#>\e(B" . "\e$(7$-\e(B")
 510     ("\e$(7"G#>\e(B" . "\e$(7$/\e(B")
 511     ("\e$(7"H#>\e(B" . "\e$(7$0\e(B")
 512     ("\e$(7"!#B\e(B" . "\e$(7$@\e(B")
 513     ("\e$(7"##B\e(B" . "\e$(7$B\e(B")
 514     ("\e$(7"5#B\e(B" . "\e$(7$C\e(B")
 515     ("\e$(7"7#B\e(B" . "\e$(7$E\e(B")
 516     ("\e$(7"9#B\e(B" . "\e$(7$F\e(B")
 517     ("\e$(7"!#C\e(B" . "\e$(7$P\e(B")
 518     ("\e$(7"##C\e(B" . "\e$(7$R\e(B")
 519     ("\e$(7"0#C\e(B" . "\e$(7$S\e(B")
 520     ("\e$(7"2#C\e(B" . "\e$(7$U\e(B")
 521     ("\e$(7"5#C\e(B" . "\e$(7$V\e(B")
 522     ("\e$(7"7#C\e(B" . "\e$(7$X\e(B")
 523     ("\e$(7"9#C\e(B" . "\e$(7$Y\e(B")
 524     ("\e$(7"G#C\e(B" . "\e$(7$[\e(B")
 525     ("\e$(7"H#C\e(B" . "\e$(7$\\e(B")
 526     ("\e$(7"(#C\e(B" . "\e$(7$]\e(B")
 527     ("\e$(7"!#D\e(B" . "\e$(7$`\e(B")
 528     ("\e$(7"##D\e(B" . "\e$(7$a\e(B")
 529     ("\e$(7"7#D\e(B" . "\e$(7$b\e(B")
 530     ("\e$(7"@#D\e(B" . "\e$(7$c\e(B")
 531     ("\e$(7"C#D\e(B" . "\e$(7$d\e(B")
 532     ("\e$(7"G#D\e(B" . "\e$(7$e\e(B")
 533     ("\e$(7"C#!\e(B" . "\e$(7%!\e(B")
 534     ("\e$(7"C##\e(B" . "\e$(7%"\e(B")
 535     ("\e$(7"C#(\e(B" . "\e$(7%$\e(B")
 536     ("\e$(7"C#0\e(B" . "\e$(7%&\e(B")
 537     ("\e$(7"C#2\e(B" . "\e$(7%'\e(B")
 538     ("\e$(7"C#4\e(B" . "\e$(7%(\e(B")
 539     ("\e$(7"C#7\e(B" . "\e$(7%)\e(B")
 540     ("\e$(7"C#9\e(B" . "\e$(7%*\e(B")
 541     ("\e$(7"D#!\e(B" . "\e$(7%@\e(B")
 542     ("\e$(7"D##\e(B" . "\e$(7%A\e(B")
 543     ("\e$(7"D#4\e(B" . "\e$(7!!\e(B") ; dummy 0x2121 added 2000/06/08 for transition l -> lng
 544     ("\e$(7"D#&\e(B" . "\e$(7%C\e(B")
 545     ("\e$(7"D#(\e(B" . "\e$(7%D\e(B")
 546     ("\e$(7"D#0\e(B" . "\e$(7%E\e(B")
 547     ("\e$(7"D#2\e(B" . "\e$(7%F\e(B")
 548     ("\e$(7"D#5\e(B" . "\e$(7%G\e(B")
 549     ("\e$(7"D#7\e(B" . "\e$(7%H\e(B")
 550     ("\e$(7"D#H\e(B" . "\e$(7%I\e(B")
 551     ("\e$(7"G#!\e(B" . "\e$(7%P\e(B")
 552     ("\e$(7"G##\e(B" . "\e$(7%Q\e(B")
 553     ("\e$(7"G#0\e(B" . "\e$(7%T\e(B")
 554     ("\e$(7"G#2\e(B" . "\e$(7%U\e(B")
 555     ("\e$(7"G#4\e(B" . "\e$(7%V\e(B")
 556     ("\e$(7"G#5\e(B" . "\e$(7%W\e(B")
 557     ("\e$(7"G#7\e(B" . "\e$(7%X\e(B")
 558     ("\e$(7"G#9\e(B" . "\e$(7%Y\e(B")))
 559
 560 (defconst tibetan-regexp
 561   (let ((l (list tibetan-precomposed-transcription-alist
 562                  tibetan-consonant-transcription-alist
 563                  tibetan-vowel-transcription-alist
 564                  tibetan-modifier-transcription-alist
 565                  tibetan-subjoined-transcription-alist))
 566         (separator "\\|")
 567         tail pattern)
 568     (while l
 569       (setq tail (car l) l (cdr l))
 570       (while tail
 571         (setq pattern (cons separator (cons (car (car tail)) pattern))
 572               tail (cdr tail))))
 573     (apply 'concat (nreverse (cdr pattern))))
 574   "Regexp matching a Tibetan transcription of a composable Tibetan sequence.
 575 The result of matching is to be used for indexing alists at conversion
 576 from a roman transcription to the corresponding Tibetan character.")
 577
 578 (defvar tibetan-precomposed-regexp
 579   (let ((l tibetan-precomposed-transcription-alist)
 580         temp)
 581     (setq temp "^\\(")
 582     (setq temp
 583           (concat temp (car (car l))))
 584     (setq l (cdr l))
 585     (while l
 586       (setq temp
 587             (concat temp "\\|" (car (car l))))
 588       (setq l (cdr l)))
 589     (concat temp "\\)"))
 590   "Regexp string to match a romanized Tibetan complex consonant.
 591 The result of matching is to be used for indexing alists when the input key
 592 from an input method is converted to the corresponding precomposed glyph.")
 593
 594 (defvar tibetan-precomposition-rule-regexp
 595   (let ((l tibetan-precomposition-rule-alist)
 596         temp)
 597     (setq temp "\\(")
 598     (setq temp (concat temp (car (car l))))
 599     (setq l (cdr l))
 600     (while l
 601       (setq temp (concat temp "\\|" (car (car l))))
 602       (setq l (cdr l)))
 603     (concat temp "\\)"))
 604   "Regexp string to match a sequence of Tibetan consonantic components, i.e.,
 605 one base consonant and one or more subjoined consonants.
 606 The result of matching is to be used for indexing alist when the component
 607 sequence is converted to the corresponding precomposed glyph.
 608 This also matches some punctuation characters which need conversion.")
 609
 610 (defvar tibetan-decomposed nil)
 611 (defvar tibetan-decomposed-temp nil)
 612
 613 (provide 'tibetan)
 614
 615 ;;; arch-tag: 8d37c8d7-f95d-450f-9ec2-819e61fc79a7
 616 ;;; tibetan.el ends here