lisp/language/tibetan.el

   1 ;;; tibetan.el --- support for Tibetan language -*- coding: iso-2022-7bit; -*-
   2
   3 ;; Copyright (C) 1997 Electrotechnical Laboratory, JAPAN.
   4 ;; Licensed to the Free Software Foundation.
   5
   6 ;; Keywords: multilingual, Tibetan
   7
   8 ;; This file is part of GNU Emacs.
   9
  10 ;; GNU Emacs is free software; you can redistribute it and/or modify
  11 ;; it under the terms of the GNU General Public License as published by
  12 ;; the Free Software Foundation; either version 2, or (at your option)
  13 ;; any later version.
  14
  15 ;; GNU Emacs is distributed in the hope that it will be useful,
  16 ;; but WITHOUT ANY WARRANTY; without even the implied warranty of
  17 ;; MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
  18 ;; GNU General Public License for more details.
  19
  20 ;; You should have received a copy of the GNU General Public License
  21 ;; along with GNU Emacs; see the file COPYING.  If not, write to the
  22 ;; Free Software Foundation, Inc., 59 Temple Place - Suite 330,
  23 ;; Boston, MA 02111-1307, USA.
  24
  25 ;; Author: Toru TOMABECHI, <Toru.Tomabechi@orient.unil.ch>
  26
  27 ;; Created: Feb. 17. 1997
  28
  29 ;; History:
  30 ;; 1997.03.13 Modification for special signs and punctuations.
  31
  32 ;;; Commentary:
  33
  34 ;;; Code:
  35
  36 ;;; Tibetan Character set.
  37 ;;; \x2130 -- \x234a is a subset of Unicode v.2 \x0f00 - \x0fb9
  38 ;;; with a slight modification. And there are some subjoined
  39 ;;; consonants which are not specified in Unicode.
  40 ;;; I hope I can add missing characters later.
  41 ;;;
  42 ;;;     00 01 02 03 04 05 06 07 08 09 0A 0B 0C 0D 0E 0F
  43 ;;;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)
  44 ;;;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,
  45 ;;;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
  46 ;;;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.
  47 ;;;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 ;
  48 ;;;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$(7!~\e(B // ;
  49 ;;;
  50 ;;;     00 01 02 03 04 05 06 07 08 09 0A 0B 0C 0D 0E 0F
  51 ;;;2220 // \e$(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$(7"/\e(B ; Base consonants
  52 ;;;2230 \e$(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$(7"?\e(B ; and
  53 ;;;2240 \e$(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$(7"O\e(B ; Vowel signs.
  54 ;;;2250 \e$(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$(7"_\e(B ; (\x2251 = vowel a)
  55 ;;;2260 \e$(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$(7"o\e(B ; Long vowels and
  56 ;;;2270 \e$(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
  57 ;;;                                                     ; atomically
  58 ;;;                                                     ; encoded.
  59 ;;;     00 01 02 03 04 05 06 07 08 09 0A 0B 0C 0D 0E 0F
  60 ;;;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
  61 ;;;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 ;
  62 ;;;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 ;
  63 ;;;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
  64 ;;;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
  65 ;;;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$(7#~\e(B // ; in Unicode.
  66 ;;;
  67 ;;;     00 01 02 03 04 05 06 07 08 09 0A 0B 0C 0D 0E 0F
  68 ;;;2420 // \e$(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
  69 ;;;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
  70 ;;;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.
  71 ;;;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
  72 ;;;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
  73 ;;;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
  74 ;;;                                                     ; when written on a
  75 ;;;     00 01 02 03 04 05 06 07 08 09 0A 0B 0C 0D 0E 0F ; file in Tibetan
  76 ;;;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.
  77 ;;;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 ;
  78 ;;;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 ;
  79 ;;;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 ;
  80 ;;;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 ;
  81 ;;;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 // ;
  82 ;;;
  83
  84
  85 (make-coding-system
  86  'tibetan-iso-8bit 2 ?Q
  87  "8-bit encoding for ASCII (MSB=0) and TIBETAN (MSB=1)."
  88  '(ascii tibetan nil nil
  89    nil nil)
  90  '((safe-charsets ascii tibetan)
  91    (post-read-conversion . tibetan-post-read-conversion)
  92    (pre-write-conversion . tibetan-pre-write-conversion)))
  93
  94 (define-coding-system-alias 'tibetan 'tibetan-iso-8bit)
  95
  96 (set-language-info-alist
  97  "Tibetan" '((charset tibetan tibetan-1-column)
  98              (coding-system tibetan-iso-8bit)
  99              (coding-priority iso-2022-7bit tibetan-iso-8bit)
 100              (input-method . "tibetan-wylie")
 101              (features tibet-util)
 102              (documentation . t)
 103              (sample-text
 104               . (tibetan-compose-string
 105                  (copy-sequence
 106 "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")))))
 107
 108
 109 ;; `\e$(7"A\e(B' is included in the pattern for subjoined consonants because we
 110 ;; treat it specially in tibetan-add-components.
 111 ;; modified by Tomabechi 1999/12/10
 112 ;; modified by Tomabechi 2000/06/08
 113 ;;          To allow infinite addition of vowels/modifiers
 114 ;;          as specified in Unicode v.3
 115 ;; \e$(7"A\e(B is removed from the class of subjoined. Tomabechi 2000/06/08
 116 ;; (for Unicode support)
 117 (defconst tibetan-composable-pattern
 118   "[\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]*"
 119   "Regexp matching a composable sequence of Tibetan characters.")
 120
 121 ;; Register a function to compose Tibetan characters.
 122 (aset composition-function-table (make-char 'tibetan)
 123       (list (cons tibetan-composable-pattern 'tibetan-composition-function)))
 124
 125 ;;;
 126 ;;; Definitions of conversion data.
 127 ;;;
 128
 129
 130 ;;; alists for tibetan char <-> transcription conversion
 131 ;;; longer transcription should come first
 132 (defconst tibetan-consonant-transcription-alist
 133   '(("tsh" . "\e$(7";\e(B")
 134     ("dzh" . "\e$(7"=\e(B")
 135     ("kSH" . "\e$(7"J\e(B")
 136     ("kh" . "\e$(7""\e(B")
 137     ("gh" . "\e$(7"$\e(B")
 138     ("ng" . "\e$(7"%\e(B")
 139     ("ch" . "\e$(7"'\e(B")
 140     ("ny" . "\e$(7"*\e(B")
 141     ("TH" . "\e$(7",\e(B")
 142     ("DH" . "\e$(7".\e(B")
 143     ("th" . "\e$(7"1\e(B")
 144     ("dh" . "\e$(7"3\e(B")
 145     ("ph" . "\e$(7"6\e(B")
 146     ("bh" . "\e$(7"8\e(B")
 147     ("ts" . "\e$(7":\e(B")
 148     ("dz" . "\e$(7"<\e(B")
 149     ("zh" . "\e$(7"?\e(B")
 150     ("sh" . "\e$(7"E\e(B")
 151     ("SH" . "\e$(7"F\e(B")
 152     ("k" . "\e$(7"!\e(B")
 153     ("g" . "\e$(7"#\e(B")
 154     ("c" . "\e$(7"&\e(B")
 155     ("j" . "\e$(7"(\e(B")
 156     ("T" . "\e$(7"+\e(B")
 157     ("D" . "\e$(7"-\e(B")
 158     ("N" . "\e$(7"/\e(B")
 159     ("t" . "\e$(7"0\e(B")
 160     ("d" . "\e$(7"2\e(B")
 161     ("n" . "\e$(7"4\e(B")
 162     ("p" . "\e$(7"5\e(B")
 163     ("b" . "\e$(7"7\e(B")
 164     ("m" . "\e$(7"9\e(B")
 165     ("w" . "\e$(7">\e(B")
 166     ("z" . "\e$(7"@\e(B")
 167     ("'" . "\e$(7"A\e(B")
 168     ("y" . "\e$(7"B\e(B")
 169     ("r" . "\e$(7"C\e(B")
 170     ("l" . "\e$(7"D\e(B")
 171     ("s" . "\e$(7"G\e(B")
 172     ("h" . "\e$(7"H\e(B")
 173     ("H" . "\e$(7"H\e(B")
 174     ("A" . "\e$(7"I\e(B")
 175     ;; Added by Tomabechi 1999/12/10
 176     ("R" . "\e$(7"K\e(B") ;; fixed form RA
 177     ))
 178
 179
 180 (defconst tibetan-vowel-transcription-alist
 181   '(
 182     ;; Composite Vowels
 183     ;; Added by Tomabechi 2000/06/08
 184     ("frr" . "\e$(7"X\e(B")
 185     ("fll" . "\e$(7"Z\e(B")
 186     ("fa" . "\e$(7"R\e(B")
 187     ("fi" . "\e$(7"T\e(B")
 188     ("fu" . "\e$(7"V\e(B")
 189     ("fr" . "\e$(7"W\e(B")
 190     ("fl" . "\e$(7"Y\e(B")
 191     ("fI" . "\e$(7"b\e(B")
 192     ;; Normal Vowels
 193     ("ai" . "\e$(7"\\e(B")
 194     ("au" . "\e$(7"^\e(B")
 195     ("ee" . "\e$(7"\\e(B")
 196     ("oo" . "\e$(7"^\e(B")
 197     ("a" . "\e$(7"Q\e(B")                 ; invisible vowel sign (\x2251)
 198     ("i" . "\e$(7"S\e(B")
 199     ("u" . "\e$(7"U\e(B")
 200     ("e" . "\e$(7"[\e(B")
 201     ("o" . "\e$(7"]\e(B")
 202     ("E" . "\e$(7"\\e(B")
 203     ("O" . "\e$(7"^\e(B")
 204     ("I" . "\e$(7"a\e(B")
 205     ("," . "\e$(7"e\e(B")                 ; idem.
 206     ))
 207
 208 (defconst tibetan-modifier-transcription-alist
 209   '(("M" . "\e$(7"_\e(B")
 210     ("~" . "\e$(7"c\e(B")
 211     ("`" . "\e$(7"d\e(B")
 212     ("x" . "\e$(7"i\e(B")
 213     ("X" . "\e$(7"j\e(B")
 214     ("v" . "\e$(7"g\e(B")
 215     ("V" . "\e$(7"h\e(B")
 216     ("q" . "\e$(7"k\e(B")
 217     ("Q" . "\e$(7"l\e(B")
 218     ("_/" . "\e$(7!I\e(B")
 219     ("_o" . "\e$(7!g\e(B")
 220     ("_O" . "\e$(7!e\e(B")))
 221
 222 (defconst tibetan-precomposed-transcription-alist
 223   '(("phyw" . "\e$(7$G\e(B")
 224     ("tshw" . "\e$(7$)\e(B")
 225     ("rtsw" . "\e$(7%.\e(B")
 226     ("khw" . "\e$(7$"\e(B")
 227     ("nyw" . "\e$(7$%\e(B")
 228     ("tsw" . "\e$(7$(\e(B")
 229     ("zhw" . "\e$(7$*\e(B")
 230     ("shw" . "\e$(7$.\e(B")
 231     ("khy" . "\e$(7$A\e(B")
 232     ("phy" . "\e$(7$D\e(B")
 233     ("khr" . "\e$(7$Q\e(B")
 234     ("thr" . "\e$(7$T\e(B")
 235     ("phr" . "\e$(7$W\e(B")
 236     ("shr" . "\e$(7$Z\e(B")
 237     ("dzr" . "\e$(7$^\e(B")
 238     ("grw" . "\e$(7$_\e(B")
 239     ("rng" . "\e$(7%#\e(B")
 240     ("rny" . "\e$(7%%\e(B")
 241     ("rts" . "\e$(7%+\e(B")
 242     ("rdz" . "\e$(7%,\e(B")
 243     ("rgw" . "\e$(7%-\e(B")
 244     ("rky" . "\e$(7%0\e(B")
 245     ("rgy" . "\e$(7%1\e(B")
 246     ("rmy" . "\e$(7%2\e(B")
 247     ("lng" . "\e$(7%B\e(B")
 248     ("sng" . "\e$(7%R\e(B")
 249     ("sny" . "\e$(7%S\e(B")
 250     ("sts" . "\e$(7%Z\e(B")
 251     ("sky" . "\e$(7%`\e(B")
 252     ("sgy" . "\e$(7%a\e(B")
 253     ("spy" . "\e$(7%b\e(B")
 254     ("sby" . "\e$(7%c\e(B")
 255     ("smy" . "\e$(7%d\e(B")
 256     ("skr" . "\e$(7%p\e(B")
 257     ("sgr" . "\e$(7%q\e(B")
 258     ("snr" . "\e$(7%r\e(B")
 259     ("spr" . "\e$(7%s\e(B")
 260     ("sbr" . "\e$(7%t\e(B")
 261     ("smr" . "\e$(7%u\e(B")
 262     ("kw" . "\e$(7$!\e(B")
 263     ("gw" . "\e$(7$#\e(B")
 264     ("cw" . "\e$(7$$\e(B")
 265     ("tw" . "\e$(7$&\e(B")
 266     ("dw" . "\e$(7$'\e(B")
 267     ("zw" . "\e$(7$+\e(B")
 268     ("rw" . "\e$(7$,\e(B")
 269     ("lw" . "\e$(7$-\e(B")
 270     ("sw" . "\e$(7$/\e(B")
 271     ("hw" . "\e$(7$0\e(B")
 272     ("ky" . "\e$(7$@\e(B")
 273     ("gy" . "\e$(7$B\e(B")
 274     ("py" . "\e$(7$C\e(B")
 275     ("by" . "\e$(7$E\e(B")
 276     ("my" . "\e$(7$F\e(B")
 277     ("kr" . "\e$(7$P\e(B")
 278     ("gr" . "\e$(7$R\e(B")
 279     ("tr" . "\e$(7$S\e(B")
 280     ("dr" . "\e$(7$U\e(B")
 281     ("pr" . "\e$(7$V\e(B")
 282     ("brk" . "\e$(7"7%!\e(B")
 283     ("brg" . "\e$(7"7%"\e(B")
 284     ("brng" . "\e$(7"7%#\e(B")
 285     ("brj" . "\e$(7"7%$\e(B")
 286     ("brny" . "\e$(7"7%%\e(B")
 287     ("brt" .  "\e$(7"7%&\e(B")
 288     ("brd" . "\e$(7"7%'\e(B")
 289     ("brn" . "\e$(7"7%(\e(B")
 290     ("brts" . "\e$(7"7%+\e(B")
 291     ("brdz" . "\e$(7"7%,\e(B")
 292     ("brl" . "\e$(7"7$d\e(B")
 293     ("br" . "\e$(7$X\e(B")
 294     ("mr" . "\e$(7$Y\e(B")
 295     ("sr" . "\e$(7$[\e(B")
 296     ("hr" . "\e$(7$\\e(B")
 297     ("jr" . "\e$(7$]\e(B")
 298     ("kl" . "\e$(7$`\e(B")
 299     ("gl" . "\e$(7$a\e(B")
 300     ("blt" . "\e$(7"7%E\e(B")
 301     ("bld" . "\e$(7"7%F\e(B")
 302     ("bl" . "\e$(7$b\e(B")
 303     ("zl" . "\e$(7$c\e(B")
 304     ("rl" . "\e$(7$d\e(B")
 305     ("sl" . "\e$(7$e\e(B")
 306     ("rk" . "\e$(7%!\e(B")
 307     ("rg" . "\e$(7%"\e(B")
 308     ("rj" . "\e$(7%$\e(B")
 309     ("rt" . "\e$(7%&\e(B")
 310     ("rd" . "\e$(7%'\e(B")
 311     ("rn" . "\e$(7%(\e(B")
 312     ("rb" . "\e$(7%)\e(B")
 313     ("rm" . "\e$(7%*\e(B")
 314     ("lk" . "\e$(7%@\e(B")
 315     ("lg" . "\e$(7%A\e(B")
 316     ("lc" . "\e$(7%C\e(B")
 317     ("lj" . "\e$(7%D\e(B")
 318     ("lt" . "\e$(7%E\e(B")
 319     ("ld" . "\e$(7%F\e(B")
 320     ("ln" . "\e$(7!!\e(B")                        ; dummy \x2121
 321     ("lp" . "\e$(7%G\e(B")
 322     ("lb" . "\e$(7%H\e(B")
 323     ("lh" . "\e$(7%I\e(B")
 324     ("sk" . "\e$(7%P\e(B")
 325     ("sg" . "\e$(7%Q\e(B")
 326     ("st" . "\e$(7%T\e(B")
 327     ("sd" . "\e$(7%U\e(B")
 328     ("sn" . "\e$(7%V\e(B")
 329     ("sp" . "\e$(7%W\e(B")
 330     ("sb" . "\e$(7%X\e(B")
 331     ("sm" . "\e$(7%Y\e(B"))
 332   )
 333
 334
 335 (defconst tibetan-subjoined-transcription-alist
 336   (sort '(("+k"  . "\e$(7#!\e(B")
 337           ("+kh" . "\e$(7#"\e(B")
 338           ("+g"  . "\e$(7##\e(B")
 339           ("+gh" . "\e$(7#$\e(B")
 340           ("+ng" . "\e$(7#%\e(B")
 341           ("+c"  . "\e$(7#&\e(B")
 342           ("+ch" . "\e$(7#'\e(B")
 343           ("+j"  . "\e$(7#(\e(B")
 344           ("+ny"  . "\e$(7#*\e(B")
 345           ("+T"  . "\e$(7#+\e(B")
 346           ("+TH" . "\e$(7#,\e(B")
 347           ("+D"  . "\e$(7#-\e(B")
 348           ("+DH" . "\e$(7#.\e(B")
 349           ("+N"  . "\e$(7#/\e(B")
 350           ("+t"  . "\e$(7#0\e(B")
 351           ("+th" . "\e$(7#1\e(B")
 352           ("+d"  . "\e$(7#2\e(B")
 353           ("+dh" . "\e$(7#3\e(B")
 354           ("+n"  . "\e$(7#4\e(B")
 355           ("+p"  . "\e$(7#5\e(B")
 356           ("+ph" . "\e$(7#6\e(B")
 357           ("+b"  . "\e$(7#7\e(B")
 358           ("+bh" . "\e$(7#8\e(B")
 359           ("+m"  . "\e$(7#9\e(B")
 360           ("+ts" . "\e$(7#:\e(B")
 361           ("+tsh" . "\e$(7#;\e(B")
 362           ("+dz" . "\e$(7#<\e(B")
 363           ("+dzh" . "\e$(7#=\e(B")
 364           ("+w"  . "\e$(7#>\e(B")
 365           ("+zh" . "\e$(7#?\e(B")
 366           ("+z"  . "\e$(7#@\e(B")
 367           ("+'"  . "\e$(7#A\e(B")
 368           ("+y"  . "\e$(7#B\e(B")
 369           ("+r"  . "\e$(7#C\e(B")
 370           ("+l"  . "\e$(7#D\e(B")
 371           ("+sh" . "\e$(7#E\e(B")
 372           ("+SH" . "\e$(7#F\e(B")
 373           ("+s"  . "\e$(7#G\e(B")
 374           ("+h"  . "\e$(7#H\e(B")
 375           ("+A"  . "\e$(7#I\e(B")
 376           ("+kSH" . "\e$(7#J\e(B")
 377           ;; Added by Tomabechi 1999/12/10
 378           ("+W" . "\e$(7#K\e(B") ;; fixed form subscribed WA
 379           ("+Y" . "\e$(7#L\e(B") ;; fixed form subscribed YA
 380           ("+R" . "\e$(7#M\e(B") ;; fixed form subscribed RA
 381           )
 382         (lambda (x y) (> (length (car x)) (length (car y))))))
 383
 384 ;;;
 385 ;;; alist for Tibetan base consonant <-> subjoined consonant conversion.
 386 ;;;
 387 (defconst tibetan-base-to-subjoined-alist
 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$(7"*\e(B" . "\e$(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$(7"0\e(B" . "\e$(7#0\e(B")
 403     ("\e$(7"1\e(B" . "\e$(7#1\e(B")
 404     ("\e$(7"2\e(B" . "\e$(7#2\e(B")
 405     ("\e$(7"3\e(B" . "\e$(7#3\e(B")
 406     ("\e$(7"4\e(B" . "\e$(7#4\e(B")
 407     ("\e$(7"5\e(B" . "\e$(7#5\e(B")
 408     ("\e$(7"6\e(B" . "\e$(7#6\e(B")
 409     ("\e$(7"7\e(B" . "\e$(7#7\e(B")
 410     ("\e$(7"8\e(B" . "\e$(7#8\e(B")
 411     ("\e$(7"9\e(B" . "\e$(7#9\e(B")
 412     ("\e$(7":\e(B" . "\e$(7#:\e(B")
 413     ("\e$(7";\e(B" . "\e$(7#;\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$(7"A\e(B" . "\e$(7#A\e(B")
 420     ("\e$(7"B\e(B" . "\e$(7#B\e(B")
 421     ("\e$(7"C\e(B" . "\e$(7#C\e(B")
 422     ("\e$(7"D\e(B" . "\e$(7#D\e(B")
 423     ("\e$(7"E\e(B" . "\e$(7#E\e(B")
 424     ("\e$(7"F\e(B" . "\e$(7#F\e(B")
 425     ("\e$(7"G\e(B" . "\e$(7#G\e(B")
 426     ("\e$(7"H\e(B" . "\e$(7#H\e(B")
 427     ("\e$(7"I\e(B" . "\e$(7#I\e(B")
 428     ("\e$(7"J\e(B" . "\e$(7#J\e(B")
 429     ;; Added by Tomabechi 1999/12/10
 430     ("\e$(7"K\e(B" . "\e$(7#M\e(B") ;; Fixed form RA (224B->234D)
 431     ))
 432
 433 ;;; alist for Tibetan composite vowels (long i, vocalic r, etc.)
 434 ;;; New varialble. created by Tomabechi 2000/06/08
 435 (defconst tibetan-composite-vowel-alist
 436   '(;; LONG A
 437     ;; ("\e$(7"R\e(B" . ((bc . tc) ?\e$(7"R\e(B))
 438     ;; LONG I
 439     ("\e$(7"T\e(B" . (?\e$(7"R\e(B (tc . bc) ?\e$(7"S\e(B))
 440     ;; LONG U
 441     ("\e$(7"V\e(B" . (?\e$(7"R\e(B (bc . tc) ?\e$(7"U\e(B))
 442     ;; VOCALIC R
 443     ("\e$(7"W\e(B" . (?\e$(7#C\e(B (tc . bc) ?\e$(7"a\e(B))
 444     ;; LONG VOCALIC R
 445     ("\e$(7"X\e(B" . (?\e$(7#C\e(B (bc . tc) ?\e$(7"R\e(B (tc . bc) ?\e$(7"a\e(B))
 446     ;; VOCALIC L
 447     ("\e$(7"Y\e(B" . (?\e$(7#D\e(B (tc . bc) ?\e$(7"a\e(B))
 448     ;;\e$(7!;\e(BLONG VOCALIC L
 449     ("\e$(7"Z\e(B" . (?\e$(7#D\e(B (bc . tc) ?\e$(7"R\e(B (tc . bc) ?\e$(7"a\e(B))
 450     ;; LONG REVERSE I
 451     ("\e$(7"b\e(B" . (?\e$(7"R\e(B (tc . bc) ?\e$(7"a\e(B))
 452     ))
 453
 454
 455
 456 ;;;
 457 ;;; alist for Tibetan consonantic components <-> precomposed glyph conversion.
 458 ;;; (includes some punctuation conversion rules)
 459 ;;;
 460 (defconst tibetan-precomposition-rule-alist
 461   `(("\e$(7"6#B#>\e(B" . "\e$(7$G\e(B")
 462     ("\e$(7"##C#>\e(B" . "\e$(7$_\e(B")
 463     ("\e$(7";#>\e(B" . "\e$(7$)\e(B")
 464     ("\e$(7"C#:#>\e(B" . "\e$(7%.\e(B")
 465     ("\e$(7"C###>\e(B" . "\e$(7%-\e(B")
 466     ("\e$(7"C#!#B\e(B" . "\e$(7%0\e(B")
 467     ("\e$(7"C###B\e(B" . "\e$(7%1\e(B")
 468     ("\e$(7"C#9#B\e(B" . "\e$(7%2\e(B")
 469     ("\e$(7"G#!#B\e(B" . "\e$(7%`\e(B")
 470     ("\e$(7"G###B\e(B" . "\e$(7%a\e(B")
 471     ("\e$(7"G#5#B\e(B" . "\e$(7%b\e(B")
 472     ("\e$(7"G#7#B\e(B" . "\e$(7%c\e(B")
 473     ("\e$(7"G#9#B\e(B" . "\e$(7%d\e(B")
 474     ("\e$(7"G#!#C\e(B" . "\e$(7%p\e(B")
 475     ("\e$(7"G###C\e(B" . "\e$(7%q\e(B")
 476     ("\e$(7"G#4#C\e(B" . "\e$(7%r\e(B")
 477     ("\e$(7"G#5#C\e(B" . "\e$(7%s\e(B")
 478     ("\e$(7"G#7#C\e(B" . "\e$(7%t\e(B")
 479     ("\e$(7"G#9#C\e(B" . "\e$(7%u\e(B")
 480     ("\e$(7""#>\e(B" . "\e$(7$"\e(B")
 481     ("\e$(7"*#>\e(B" . "\e$(7$%\e(B")
 482     ("\e$(7":#>\e(B" . "\e$(7$(\e(B")
 483     ("\e$(7"?#>\e(B" . "\e$(7$*\e(B")
 484     ("\e$(7"E#>\e(B" . "\e$(7$.\e(B")
 485     ("\e$(7""#B\e(B" . "\e$(7$A\e(B")
 486     ("\e$(7"6#B\e(B" . "\e$(7$D\e(B")
 487     ("\e$(7""#C\e(B" . "\e$(7$Q\e(B")
 488     ("\e$(7"1#C\e(B" . "\e$(7$T\e(B")
 489     ("\e$(7"6#C\e(B" . "\e$(7$W\e(B")
 490     ("\e$(7"E#C\e(B" . "\e$(7$Z\e(B")
 491     ("\e$(7"<#C\e(B" . "\e$(7$^\e(B")
 492     ("\e$(7"C#%\e(B" . "\e$(7%#\e(B")
 493     ("\e$(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"D#%\e(B" . "\e$(7%B\e(B")
 497     ("\e$(7"G#%\e(B" . "\e$(7%R\e(B")
 498     ("\e$(7"G#*\e(B" . "\e$(7%S\e(B")
 499     ("\e$(7"G#:\e(B" . "\e$(7%Z\e(B")
 500     ("\e$(7"!#>\e(B" . "\e$(7$!\e(B")
 501     ("\e$(7"##>\e(B" . "\e$(7$#\e(B")
 502     ("\e$(7"&#>\e(B" . "\e$(7$$\e(B")
 503     ("\e$(7"0#>\e(B" . "\e$(7$&\e(B")
 504     ("\e$(7"2#>\e(B" . "\e$(7$'\e(B")
 505     ("\e$(7"@#>\e(B" . "\e$(7$+\e(B")
 506     ("\e$(7"C#>\e(B" . "\e$(7$,\e(B")
 507     ("\e$(7"D#>\e(B" . "\e$(7$-\e(B")
 508     ("\e$(7"G#>\e(B" . "\e$(7$/\e(B")
 509     ("\e$(7"H#>\e(B" . "\e$(7$0\e(B")
 510     ("\e$(7"!#B\e(B" . "\e$(7$@\e(B")
 511     ("\e$(7"##B\e(B" . "\e$(7$B\e(B")
 512     ("\e$(7"5#B\e(B" . "\e$(7$C\e(B")
 513     ("\e$(7"7#B\e(B" . "\e$(7$E\e(B")
 514     ("\e$(7"9#B\e(B" . "\e$(7$F\e(B")
 515     ("\e$(7"!#C\e(B" . "\e$(7$P\e(B")
 516     ("\e$(7"##C\e(B" . "\e$(7$R\e(B")
 517     ("\e$(7"0#C\e(B" . "\e$(7$S\e(B")
 518     ("\e$(7"2#C\e(B" . "\e$(7$U\e(B")
 519     ("\e$(7"5#C\e(B" . "\e$(7$V\e(B")
 520     ("\e$(7"7#C\e(B" . "\e$(7$X\e(B")
 521     ("\e$(7"9#C\e(B" . "\e$(7$Y\e(B")
 522     ("\e$(7"G#C\e(B" . "\e$(7$[\e(B")
 523     ("\e$(7"H#C\e(B" . "\e$(7$\\e(B")
 524     ("\e$(7"(#C\e(B" . "\e$(7$]\e(B")
 525     ("\e$(7"!#D\e(B" . "\e$(7$`\e(B")
 526     ("\e$(7"##D\e(B" . "\e$(7$a\e(B")
 527     ("\e$(7"7#D\e(B" . "\e$(7$b\e(B")
 528     ("\e$(7"@#D\e(B" . "\e$(7$c\e(B")
 529     ("\e$(7"C#D\e(B" . "\e$(7$d\e(B")
 530     ("\e$(7"G#D\e(B" . "\e$(7$e\e(B")
 531     ("\e$(7"C#!\e(B" . "\e$(7%!\e(B")
 532     ("\e$(7"C##\e(B" . "\e$(7%"\e(B")
 533     ("\e$(7"C#(\e(B" . "\e$(7%$\e(B")
 534     ("\e$(7"C#0\e(B" . "\e$(7%&\e(B")
 535     ("\e$(7"C#2\e(B" . "\e$(7%'\e(B")
 536     ("\e$(7"C#4\e(B" . "\e$(7%(\e(B")
 537     ("\e$(7"C#7\e(B" . "\e$(7%)\e(B")
 538     ("\e$(7"C#9\e(B" . "\e$(7%*\e(B")
 539     ("\e$(7"D#!\e(B" . "\e$(7%@\e(B")
 540     ("\e$(7"D##\e(B" . "\e$(7%A\e(B")
 541     ("\e$(7"D#4\e(B" . "\e$(7!!\e(B") ; dummy 0x2121 added 2000/06/08 for transition l -> lng
 542     ("\e$(7"D#&\e(B" . "\e$(7%C\e(B")
 543     ("\e$(7"D#(\e(B" . "\e$(7%D\e(B")
 544     ("\e$(7"D#0\e(B" . "\e$(7%E\e(B")
 545     ("\e$(7"D#2\e(B" . "\e$(7%F\e(B")
 546     ("\e$(7"D#5\e(B" . "\e$(7%G\e(B")
 547     ("\e$(7"D#7\e(B" . "\e$(7%H\e(B")
 548     ("\e$(7"D#H\e(B" . "\e$(7%I\e(B")
 549     ("\e$(7"G#!\e(B" . "\e$(7%P\e(B")
 550     ("\e$(7"G##\e(B" . "\e$(7%Q\e(B")
 551     ("\e$(7"G#0\e(B" . "\e$(7%T\e(B")
 552     ("\e$(7"G#2\e(B" . "\e$(7%U\e(B")
 553     ("\e$(7"G#4\e(B" . "\e$(7%V\e(B")
 554     ("\e$(7"G#5\e(B" . "\e$(7%W\e(B")
 555     ("\e$(7"G#7\e(B" . "\e$(7%X\e(B")
 556     ("\e$(7"G#9\e(B" . "\e$(7%Y\e(B")))
 557
 558 (defconst tibetan-regexp
 559   (let ((l (list tibetan-precomposed-transcription-alist
 560                  tibetan-consonant-transcription-alist
 561                  tibetan-vowel-transcription-alist
 562                  tibetan-modifier-transcription-alist
 563                  tibetan-subjoined-transcription-alist))
 564         (separator "\\|")
 565         tail pattern)
 566     (while l
 567       (setq tail (car l) l (cdr l))
 568       (while tail
 569         (setq pattern (cons separator (cons (car (car tail)) pattern))
 570               tail (cdr tail))))
 571     (apply 'concat (nreverse (cdr pattern))))
 572   "Regexp matching a Tibetan transcription of a composable Tibetan sequence.
 573 The result of matching is to be used for indexing alists at conversion
 574 from a roman transcription to the corresponding Tibetan character.")
 575
 576 (defvar tibetan-precomposed-regexp
 577   (let ((l tibetan-precomposed-transcription-alist)
 578         temp)
 579     (setq temp "^\\(")
 580     (setq temp
 581           (concat temp (car (car l))))
 582     (setq l (cdr l))
 583     (while l
 584       (setq temp
 585             (concat temp "\\|" (car (car l))))
 586       (setq l (cdr l)))
 587     (concat temp "\\)"))
 588   "Regexp string to match a romanized Tibetan complex consonant.
 589 The result of matching is to be used for indexing alists when the input key
 590 from an input method is converted to the corresponding precomposed glyph.")
 591
 592 (defvar tibetan-precomposition-rule-regexp
 593   (let ((l tibetan-precomposition-rule-alist)
 594         temp)
 595     (setq temp "\\(")
 596     (setq temp (concat temp (car (car l))))
 597     (setq l (cdr l))
 598     (while l
 599       (setq temp (concat temp "\\|" (car (car l))))
 600       (setq l (cdr l)))
 601     (concat temp "\\)"))
 602   "Regexp string to match a sequence of Tibetan consonantic components, i.e.,
 603 one base consonant and one or more subjoined consonants.
 604 The result of matching is to be used for indexing alist when the component
 605 sequence is converted to the corresponding precomposed glyph.
 606 This also matches some punctuation characters which need conversion.")
 607
 608 (defvar tibetan-decomposed nil)
 609 (defvar tibetan-decomposed-temp nil)
 610
 611 (provide 'tibetan)
 612
 613 ;;; arch-tag: 8d37c8d7-f95d-450f-9ec2-819e61fc79a7
 614 ;;; tibetan.el ends here