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
   4 ;;   Free Software Foundation, Inc.
   5 ;; Copyright (C) 1997, 1998, 1999, 2000, 2001, 2002, 2003, 2004, 2005, 2006, 2007
   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 ;; Keywords: multilingual, Tibetan, i18n
  13
  14 ;; This file is part of GNU Emacs.
  15
  16 ;; GNU Emacs is free software; you can redistribute it and/or modify
  17 ;; it under the terms of the GNU General Public License as published by
  18 ;; the Free Software Foundation; either version 3, or (at your option)
  19 ;; any later version.
  20
  21 ;; GNU Emacs is distributed in the hope that it will be useful,
  22 ;; but WITHOUT ANY WARRANTY; without even the implied warranty of
  23 ;; MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
  24 ;; GNU General Public License for more details.
  25
  26 ;; You should have received a copy of the GNU General Public License
  27 ;; along with GNU Emacs; see the file COPYING.  If not, write to the
  28 ;; Free Software Foundation, Inc., 51 Franklin Street, Fifth Floor,
  29 ;; Boston, MA 02110-1301, USA.
  30
  31 ;; Author: Toru TOMABECHI, <Toru.Tomabechi@orient.unil.ch>
  32
  33 ;; Created: Feb. 17. 1997
  34
  35 ;; History:
  36 ;; 1997.03.13 Modification for special signs and punctuations.
  37
  38 ;;; Commentary:
  39
  40 ;;; Code:
  41
  42 ;;; Tibetan Character set.
  43 ;;; \x2130 -- \x234a is a subset of Unicode v.2 \x0f00 - \x0fb9
  44 ;;; with a slight modification. And there are some subjoined
  45 ;;; consonants which are not specified in Unicode.
  46 ;;; I hope I can add missing characters later.
  47 ;;;
  48 ;;;     00 01 02 03 04 05 06 07 08 09 0A 0B 0C 0D 0E 0F
  49 ;;;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)
  50 ;;;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,
  51 ;;;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
  52 ;;;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.
  53 ;;;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 ;
  54 ;;;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$(7!~\e(B // ;
  55 ;;;
  56 ;;;     00 01 02 03 04 05 06 07 08 09 0A 0B 0C 0D 0E 0F
  57 ;;;2220 // \e$(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$(7"/\e(B ; Base consonants
  58 ;;;2230 \e$(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$(7"?\e(B ; and
  59 ;;;2240 \e$(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$(7"O\e(B ; Vowel signs.
  60 ;;;2250 \e$(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$(7"_\e(B ; (\x2251 = vowel a)
  61 ;;;2260 \e$(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$(7"o\e(B ; Long vowels and
  62 ;;;2270 \e$(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
  63 ;;;                                                     ; atomically
  64 ;;;                                                     ; encoded.
  65 ;;;     00 01 02 03 04 05 06 07 08 09 0A 0B 0C 0D 0E 0F
  66 ;;;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
  67 ;;;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 ;
  68 ;;;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 ;
  69 ;;;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
  70 ;;;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
  71 ;;;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$(7#~\e(B // ; in Unicode.
  72 ;;;
  73 ;;;     00 01 02 03 04 05 06 07 08 09 0A 0B 0C 0D 0E 0F
  74 ;;;2420 // \e$(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
  75 ;;;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
  76 ;;;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.
  77 ;;;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
  78 ;;;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
  79 ;;;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
  80 ;;;                                                     ; when written on a
  81 ;;;     00 01 02 03 04 05 06 07 08 09 0A 0B 0C 0D 0E 0F ; file in Tibetan
  82 ;;;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.
  83 ;;;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 ;
  84 ;;;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 ;
  85 ;;;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 ;
  86 ;;;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 ;
  87 ;;;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 // ;
  88 ;;;
  89
  90
  91 (define-coding-system 'tibetan-iso-8bit
  92   "8-bit encoding for ASCII (MSB=0) and TIBETAN (MSB=1)."
  93   :coding-type 'iso-2022
  94   :mnemonic ?Q
  95   :designation [ascii tibetan nil nil]
  96   :charset-list '(ascii tibetan))
  97
  98 (define-coding-system-alias 'tibetan 'tibetan-iso-8bit)
  99
 100 (set-language-info-alist
 101  "Tibetan" '((charset tibetan tibetan-1-column)
 102              (coding-system tibetan-iso-8bit)
 103              (coding-priority iso-2022-7bit tibetan-iso-8bit)
 104              (input-method . "tibetan-wylie")
 105              (features tibet-util)
 106              (documentation . t)
 107              (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")))
 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$,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]*"
 119   "Regexp matching a composable sequence of Tibetan characters.")
 120
 121 ;;;
 122 ;;; Definitions of conversion data.
 123 ;;;
 124
 125
 126 ;;; alists for tibetan char <-> transcription conversion
 127 ;;; longer transcription should come first
 128 (defconst tibetan-consonant-transcription-alist
 129   '(("tsh" . "\e$(7";\e(B")
 130     ("dzh" . "\e$(7"=\e(B")
 131     ("kSH" . "\e$(7"J\e(B")
 132     ("kh" . "\e$(7""\e(B")
 133     ("gh" . "\e$(7"$\e(B")
 134     ("ng" . "\e$(7"%\e(B")
 135     ("ch" . "\e$(7"'\e(B")
 136     ("ny" . "\e$(7"*\e(B")
 137     ("TH" . "\e$(7",\e(B")
 138     ("DH" . "\e$(7".\e(B")
 139     ("th" . "\e$(7"1\e(B")
 140     ("dh" . "\e$(7"3\e(B")
 141     ("ph" . "\e$(7"6\e(B")
 142     ("bh" . "\e$(7"8\e(B")
 143     ("ts" . "\e$(7":\e(B")
 144     ("dz" . "\e$(7"<\e(B")
 145     ("zh" . "\e$(7"?\e(B")
 146     ("sh" . "\e$(7"E\e(B")
 147     ("SH" . "\e$(7"F\e(B")
 148     ("k" . "\e$(7"!\e(B")
 149     ("g" . "\e$(7"#\e(B")
 150     ("c" . "\e$(7"&\e(B")
 151     ("j" . "\e$(7"(\e(B")
 152     ("T" . "\e$(7"+\e(B")
 153     ("D" . "\e$(7"-\e(B")
 154     ("N" . "\e$(7"/\e(B")
 155     ("t" . "\e$(7"0\e(B")
 156     ("d" . "\e$(7"2\e(B")
 157     ("n" . "\e$(7"4\e(B")
 158     ("p" . "\e$(7"5\e(B")
 159     ("b" . "\e$(7"7\e(B")
 160     ("m" . "\e$(7"9\e(B")
 161     ("w" . "\e$(7">\e(B")
 162     ("z" . "\e$(7"@\e(B")
 163     ("'" . "\e$(7"A\e(B")
 164     ("y" . "\e$(7"B\e(B")
 165     ("r" . "\e$(7"C\e(B")
 166     ("l" . "\e$(7"D\e(B")
 167     ("s" . "\e$(7"G\e(B")
 168     ("h" . "\e$(7"H\e(B")
 169     ("H" . "\e$(7"H\e(B")
 170     ("A" . "\e$(7"I\e(B")
 171     ;; Added by Tomabechi 1999/12/10
 172     ("R" . "\e$(7"K\e(B") ;; fixed form RA
 173     ))
 174
 175
 176 (defconst tibetan-vowel-transcription-alist
 177   '(
 178     ;; Composite Vowels
 179     ;; Added by Tomabechi 2000/06/08
 180     ("frr" . "\e$(7"X\e(B")
 181     ("fll" . "\e$(7"Z\e(B")
 182     ("fa" . "\e$(7"R\e(B")
 183     ("fi" . "\e$(7"T\e(B")
 184     ("fu" . "\e$(7"V\e(B")
 185     ("fr" . "\e$(7"W\e(B")
 186     ("fl" . "\e$(7"Y\e(B")
 187     ("fI" . "\e$(7"b\e(B")
 188     ;; Normal Vowels
 189     ("ai" . "\e$(7"\\e(B")
 190     ("au" . "\e$(7"^\e(B")
 191     ("ee" . "\e$(7"\\e(B")
 192     ("oo" . "\e$(7"^\e(B")
 193     ("a" . "\e$(7"Q\e(B")                 ; invisible vowel sign (\x2251)
 194     ("i" . "\e$(7"S\e(B")
 195     ("u" . "\e$(7"U\e(B")
 196     ("e" . "\e$(7"[\e(B")
 197     ("o" . "\e$(7"]\e(B")
 198     ("E" . "\e$(7"\\e(B")
 199     ("O" . "\e$(7"^\e(B")
 200     ("I" . "\e$(7"a\e(B")
 201     ("," . "\e$(7"e\e(B")                 ; idem.
 202     ))
 203
 204 (defconst tibetan-modifier-transcription-alist
 205   '(("M" . "\e$(7"_\e(B")
 206     ("~" . "\e$(7"c\e(B")
 207     ("`" . "\e$(7"d\e(B")
 208     ("x" . "\e$(7"i\e(B")
 209     ("X" . "\e$(7"j\e(B")
 210     ("v" . "\e$(7"g\e(B")
 211     ("V" . "\e$(7"h\e(B")
 212     ("q" . "\e$(7"k\e(B")
 213     ("Q" . "\e$(7"l\e(B")
 214     ("_/" . "\e$(7!I\e(B")
 215     ("_o" . "\e$(7!g\e(B")
 216     ("_O" . "\e$(7!e\e(B")))
 217
 218 (defconst tibetan-precomposed-transcription-alist
 219   '(("phyw" . "\e$(7$G\e(B")
 220     ("tshw" . "\e$(7$)\e(B")
 221     ("rtsw" . "\e$(7%.\e(B")
 222     ("khw" . "\e$(7$"\e(B")
 223     ("nyw" . "\e$(7$%\e(B")
 224     ("tsw" . "\e$(7$(\e(B")
 225     ("zhw" . "\e$(7$*\e(B")
 226     ("shw" . "\e$(7$.\e(B")
 227     ("khy" . "\e$(7$A\e(B")
 228     ("phy" . "\e$(7$D\e(B")
 229     ("khr" . "\e$(7$Q\e(B")
 230     ("thr" . "\e$(7$T\e(B")
 231     ("phr" . "\e$(7$W\e(B")
 232     ("shr" . "\e$(7$Z\e(B")
 233     ("dzr" . "\e$(7$^\e(B")
 234     ("grw" . "\e$(7$_\e(B")
 235     ("rng" . "\e$(7%#\e(B")
 236     ("rny" . "\e$(7%%\e(B")
 237     ("rts" . "\e$(7%+\e(B")
 238     ("rdz" . "\e$(7%,\e(B")
 239     ("rgw" . "\e$(7%-\e(B")
 240     ("rky" . "\e$(7%0\e(B")
 241     ("rgy" . "\e$(7%1\e(B")
 242     ("rmy" . "\e$(7%2\e(B")
 243     ("lng" . "\e$(7%B\e(B")
 244     ("sng" . "\e$(7%R\e(B")
 245     ("sny" . "\e$(7%S\e(B")
 246     ("sts" . "\e$(7%Z\e(B")
 247     ("sky" . "\e$(7%`\e(B")
 248     ("sgy" . "\e$(7%a\e(B")
 249     ("spy" . "\e$(7%b\e(B")
 250     ("sby" . "\e$(7%c\e(B")
 251     ("smy" . "\e$(7%d\e(B")
 252     ("skr" . "\e$(7%p\e(B")
 253     ("sgr" . "\e$(7%q\e(B")
 254     ("snr" . "\e$(7%r\e(B")
 255     ("spr" . "\e$(7%s\e(B")
 256     ("sbr" . "\e$(7%t\e(B")
 257     ("smr" . "\e$(7%u\e(B")
 258     ("kw" . "\e$(7$!\e(B")
 259     ("gw" . "\e$(7$#\e(B")
 260     ("cw" . "\e$(7$$\e(B")
 261     ("tw" . "\e$(7$&\e(B")
 262     ("dw" . "\e$(7$'\e(B")
 263     ("zw" . "\e$(7$+\e(B")
 264     ("rw" . "\e$(7$,\e(B")
 265     ("lw" . "\e$(7$-\e(B")
 266     ("sw" . "\e$(7$/\e(B")
 267     ("hw" . "\e$(7$0\e(B")
 268     ("ky" . "\e$(7$@\e(B")
 269     ("gy" . "\e$(7$B\e(B")
 270     ("py" . "\e$(7$C\e(B")
 271     ("by" . "\e$(7$E\e(B")
 272     ("my" . "\e$(7$F\e(B")
 273     ("kr" . "\e$(7$P\e(B")
 274     ("gr" . "\e$(7$R\e(B")
 275     ("tr" . "\e$(7$S\e(B")
 276     ("dr" . "\e$(7$U\e(B")
 277     ("pr" . "\e$(7$V\e(B")
 278     ("brk" . "\e$(7"7%!\e(B")
 279     ("brg" . "\e$(7"7%"\e(B")
 280     ("brng" . "\e$(7"7%#\e(B")
 281     ("brj" . "\e$(7"7%$\e(B")
 282     ("brny" . "\e$(7"7%%\e(B")
 283     ("brt" .  "\e$(7"7%&\e(B")
 284     ("brd" . "\e$(7"7%'\e(B")
 285     ("brn" . "\e$(7"7%(\e(B")
 286     ("brts" . "\e$(7"7%+\e(B")
 287     ("brdz" . "\e$(7"7%,\e(B")
 288     ("brl" . "\e$(7"7$d\e(B")
 289     ("br" . "\e$(7$X\e(B")
 290     ("mr" . "\e$(7$Y\e(B")
 291     ("sr" . "\e$(7$[\e(B")
 292     ("hr" . "\e$(7$\\e(B")
 293     ("jr" . "\e$(7$]\e(B")
 294     ("kl" . "\e$(7$`\e(B")
 295     ("gl" . "\e$(7$a\e(B")
 296     ("blt" . "\e$(7"7%E\e(B")
 297     ("bld" . "\e$(7"7%F\e(B")
 298     ("bl" . "\e$(7$b\e(B")
 299     ("zl" . "\e$(7$c\e(B")
 300     ("rl" . "\e$(7$d\e(B")
 301     ("sl" . "\e$(7$e\e(B")
 302     ("rk" . "\e$(7%!\e(B")
 303     ("rg" . "\e$(7%"\e(B")
 304     ("rj" . "\e$(7%$\e(B")
 305     ("rt" . "\e$(7%&\e(B")
 306     ("rd" . "\e$(7%'\e(B")
 307     ("rn" . "\e$(7%(\e(B")
 308     ("rb" . "\e$(7%)\e(B")
 309     ("rm" . "\e$(7%*\e(B")
 310     ("lk" . "\e$(7%@\e(B")
 311     ("lg" . "\e$(7%A\e(B")
 312     ("lc" . "\e$(7%C\e(B")
 313     ("lj" . "\e$(7%D\e(B")
 314     ("lt" . "\e$(7%E\e(B")
 315     ("ld" . "\e$(7%F\e(B")
 316     ("ln" . "\e$(7!!\e(B")                        ; dummy \x2121
 317     ("lp" . "\e$(7%G\e(B")
 318     ("lb" . "\e$(7%H\e(B")
 319     ("lh" . "\e$(7%I\e(B")
 320     ("sk" . "\e$(7%P\e(B")
 321     ("sg" . "\e$(7%Q\e(B")
 322     ("st" . "\e$(7%T\e(B")
 323     ("sd" . "\e$(7%U\e(B")
 324     ("sn" . "\e$(7%V\e(B")
 325     ("sp" . "\e$(7%W\e(B")
 326     ("sb" . "\e$(7%X\e(B")
 327     ("sm" . "\e$(7%Y\e(B"))
 328   )
 329
 330
 331 (defconst tibetan-subjoined-transcription-alist
 332   (sort '(("+k"  . "\e$(7#!\e(B")
 333           ("+kh" . "\e$(7#"\e(B")
 334           ("+g"  . "\e$(7##\e(B")
 335           ("+gh" . "\e$(7#$\e(B")
 336           ("+ng" . "\e$(7#%\e(B")
 337           ("+c"  . "\e$(7#&\e(B")
 338           ("+ch" . "\e$(7#'\e(B")
 339           ("+j"  . "\e$(7#(\e(B")
 340           ("+ny"  . "\e$(7#*\e(B")
 341           ("+T"  . "\e$(7#+\e(B")
 342           ("+TH" . "\e$(7#,\e(B")
 343           ("+D"  . "\e$(7#-\e(B")
 344           ("+DH" . "\e$(7#.\e(B")
 345           ("+N"  . "\e$(7#/\e(B")
 346           ("+t"  . "\e$(7#0\e(B")
 347           ("+th" . "\e$(7#1\e(B")
 348           ("+d"  . "\e$(7#2\e(B")
 349           ("+dh" . "\e$(7#3\e(B")
 350           ("+n"  . "\e$(7#4\e(B")
 351           ("+p"  . "\e$(7#5\e(B")
 352           ("+ph" . "\e$(7#6\e(B")
 353           ("+b"  . "\e$(7#7\e(B")
 354           ("+bh" . "\e$(7#8\e(B")
 355           ("+m"  . "\e$(7#9\e(B")
 356           ("+ts" . "\e$(7#:\e(B")
 357           ("+tsh" . "\e$(7#;\e(B")
 358           ("+dz" . "\e$(7#<\e(B")
 359           ("+dzh" . "\e$(7#=\e(B")
 360           ("+w"  . "\e$(7#>\e(B")
 361           ("+zh" . "\e$(7#?\e(B")
 362           ("+z"  . "\e$(7#@\e(B")
 363           ("+'"  . "\e$(7#A\e(B")
 364           ("+y"  . "\e$(7#B\e(B")
 365           ("+r"  . "\e$(7#C\e(B")
 366           ("+l"  . "\e$(7#D\e(B")
 367           ("+sh" . "\e$(7#E\e(B")
 368           ("+SH" . "\e$(7#F\e(B")
 369           ("+s"  . "\e$(7#G\e(B")
 370           ("+h"  . "\e$(7#H\e(B")
 371           ("+A"  . "\e$(7#I\e(B")
 372           ("+kSH" . "\e$(7#J\e(B")
 373           ;; Added by Tomabechi 1999/12/10
 374           ("+W" . "\e$(7#K\e(B") ;; fixed form subscribed WA
 375           ("+Y" . "\e$(7#L\e(B") ;; fixed form subscribed YA
 376           ("+R" . "\e$(7#M\e(B") ;; fixed form subscribed RA
 377           )
 378         (lambda (x y) (> (length (car x)) (length (car y))))))
 379
 380 ;;;
 381 ;;; alist for Tibetan base consonant <-> subjoined consonant conversion.
 382 ;;;
 383 (defconst tibetan-base-to-subjoined-alist
 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$(7".\e(B" . "\e$(7#.\e(B")
 397     ("\e$(7"/\e(B" . "\e$(7#/\e(B")
 398     ("\e$(7"0\e(B" . "\e$(7#0\e(B")
 399     ("\e$(7"1\e(B" . "\e$(7#1\e(B")
 400     ("\e$(7"2\e(B" . "\e$(7#2\e(B")
 401     ("\e$(7"3\e(B" . "\e$(7#3\e(B")
 402     ("\e$(7"4\e(B" . "\e$(7#4\e(B")
 403     ("\e$(7"5\e(B" . "\e$(7#5\e(B")
 404     ("\e$(7"6\e(B" . "\e$(7#6\e(B")
 405     ("\e$(7"7\e(B" . "\e$(7#7\e(B")
 406     ("\e$(7"8\e(B" . "\e$(7#8\e(B")
 407     ("\e$(7"9\e(B" . "\e$(7#9\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$(7"?\e(B" . "\e$(7#?\e(B")
 414     ("\e$(7"@\e(B" . "\e$(7#@\e(B")
 415     ("\e$(7"A\e(B" . "\e$(7#A\e(B")
 416     ("\e$(7"B\e(B" . "\e$(7#B\e(B")
 417     ("\e$(7"C\e(B" . "\e$(7#C\e(B")
 418     ("\e$(7"D\e(B" . "\e$(7#D\e(B")
 419     ("\e$(7"E\e(B" . "\e$(7#E\e(B")
 420     ("\e$(7"F\e(B" . "\e$(7#F\e(B")
 421     ("\e$(7"G\e(B" . "\e$(7#G\e(B")
 422     ("\e$(7"H\e(B" . "\e$(7#H\e(B")
 423     ("\e$(7"I\e(B" . "\e$(7#I\e(B")
 424     ("\e$(7"J\e(B" . "\e$(7#J\e(B")
 425     ;; Added by Tomabechi 1999/12/10
 426     ("\e$(7"K\e(B" . "\e$(7#M\e(B") ;; Fixed form RA (224B->234D)
 427     ))
 428
 429 ;;; alist for Tibetan composite vowels (long i, vocalic r, etc.)
 430 ;;; New varialble. created by Tomabechi 2000/06/08
 431 (defconst tibetan-composite-vowel-alist
 432   '(;; LONG A
 433     ;; ("\e$(7"R\e(B" . ((bc . tc) ?\e$(7"R\e(B))
 434     ;; LONG I
 435     ("\e$(7"T\e(B" . (?\e$(7"R\e(B (tc . bc) ?\e$(7"S\e(B))
 436     ;; LONG U
 437     ("\e$(7"V\e(B" . (?\e$(7"R\e(B (bc . tc) ?\e$(7"U\e(B))
 438     ;; VOCALIC R
 439     ("\e$(7"W\e(B" . (?\e$(7#C\e(B (tc . bc) ?\e$(7"a\e(B))
 440     ;; LONG VOCALIC R
 441     ("\e$(7"X\e(B" . (?\e$(7#C\e(B (bc . tc) ?\e$(7"R\e(B (tc . bc) ?\e$(7"a\e(B))
 442     ;; VOCALIC L
 443     ("\e$(7"Y\e(B" . (?\e$(7#D\e(B (tc . bc) ?\e$(7"a\e(B))
 444     ;;\e$(7!;\e(BLONG VOCALIC L
 445     ("\e$(7"Z\e(B" . (?\e$(7#D\e(B (bc . tc) ?\e$(7"R\e(B (tc . bc) ?\e$(7"a\e(B))
 446     ;; LONG REVERSE I
 447     ("\e$(7"b\e(B" . (?\e$(7"R\e(B (tc . bc) ?\e$(7"a\e(B))
 448     ))
 449
 450
 451
 452 ;;;
 453 ;;; alist for Tibetan consonantic components <-> precomposed glyph conversion.
 454 ;;; (includes some punctuation conversion rules)
 455 ;;;
 456 (defconst tibetan-precomposition-rule-alist
 457   `(("\e$(7"6#B#>\e(B" . "\e$(7$G\e(B")
 458     ("\e$(7"##C#>\e(B" . "\e$(7$_\e(B")
 459     ("\e$(7";#>\e(B" . "\e$(7$)\e(B")
 460     ("\e$(7"C#:#>\e(B" . "\e$(7%.\e(B")
 461     ("\e$(7"C###>\e(B" . "\e$(7%-\e(B")
 462     ("\e$(7"C#!#B\e(B" . "\e$(7%0\e(B")
 463     ("\e$(7"C###B\e(B" . "\e$(7%1\e(B")
 464     ("\e$(7"C#9#B\e(B" . "\e$(7%2\e(B")
 465     ("\e$(7"G#!#B\e(B" . "\e$(7%`\e(B")
 466     ("\e$(7"G###B\e(B" . "\e$(7%a\e(B")
 467     ("\e$(7"G#5#B\e(B" . "\e$(7%b\e(B")
 468     ("\e$(7"G#7#B\e(B" . "\e$(7%c\e(B")
 469     ("\e$(7"G#9#B\e(B" . "\e$(7%d\e(B")
 470     ("\e$(7"G#!#C\e(B" . "\e$(7%p\e(B")
 471     ("\e$(7"G###C\e(B" . "\e$(7%q\e(B")
 472     ("\e$(7"G#4#C\e(B" . "\e$(7%r\e(B")
 473     ("\e$(7"G#5#C\e(B" . "\e$(7%s\e(B")
 474     ("\e$(7"G#7#C\e(B" . "\e$(7%t\e(B")
 475     ("\e$(7"G#9#C\e(B" . "\e$(7%u\e(B")
 476     ("\e$(7""#>\e(B" . "\e$(7$"\e(B")
 477     ("\e$(7"*#>\e(B" . "\e$(7$%\e(B")
 478     ("\e$(7":#>\e(B" . "\e$(7$(\e(B")
 479     ("\e$(7"?#>\e(B" . "\e$(7$*\e(B")
 480     ("\e$(7"E#>\e(B" . "\e$(7$.\e(B")
 481     ("\e$(7""#B\e(B" . "\e$(7$A\e(B")
 482     ("\e$(7"6#B\e(B" . "\e$(7$D\e(B")
 483     ("\e$(7""#C\e(B" . "\e$(7$Q\e(B")
 484     ("\e$(7"1#C\e(B" . "\e$(7$T\e(B")
 485     ("\e$(7"6#C\e(B" . "\e$(7$W\e(B")
 486     ("\e$(7"E#C\e(B" . "\e$(7$Z\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"C#:\e(B" . "\e$(7%+\e(B")
 491     ("\e$(7"C#<\e(B" . "\e$(7%,\e(B")
 492     ("\e$(7"D#%\e(B" . "\e$(7%B\e(B")
 493     ("\e$(7"G#%\e(B" . "\e$(7%R\e(B")
 494     ("\e$(7"G#*\e(B" . "\e$(7%S\e(B")
 495     ("\e$(7"G#:\e(B" . "\e$(7%Z\e(B")
 496     ("\e$(7"!#>\e(B" . "\e$(7$!\e(B")
 497     ("\e$(7"##>\e(B" . "\e$(7$#\e(B")
 498     ("\e$(7"&#>\e(B" . "\e$(7$$\e(B")
 499     ("\e$(7"0#>\e(B" . "\e$(7$&\e(B")
 500     ("\e$(7"2#>\e(B" . "\e$(7$'\e(B")
 501     ("\e$(7"@#>\e(B" . "\e$(7$+\e(B")
 502     ("\e$(7"C#>\e(B" . "\e$(7$,\e(B")
 503     ("\e$(7"D#>\e(B" . "\e$(7$-\e(B")
 504     ("\e$(7"G#>\e(B" . "\e$(7$/\e(B")
 505     ("\e$(7"H#>\e(B" . "\e$(7$0\e(B")
 506     ("\e$(7"!#B\e(B" . "\e$(7$@\e(B")
 507     ("\e$(7"##B\e(B" . "\e$(7$B\e(B")
 508     ("\e$(7"5#B\e(B" . "\e$(7$C\e(B")
 509     ("\e$(7"7#B\e(B" . "\e$(7$E\e(B")
 510     ("\e$(7"9#B\e(B" . "\e$(7$F\e(B")
 511     ("\e$(7"!#C\e(B" . "\e$(7$P\e(B")
 512     ("\e$(7"##C\e(B" . "\e$(7$R\e(B")
 513     ("\e$(7"0#C\e(B" . "\e$(7$S\e(B")
 514     ("\e$(7"2#C\e(B" . "\e$(7$U\e(B")
 515     ("\e$(7"5#C\e(B" . "\e$(7$V\e(B")
 516     ("\e$(7"7#C\e(B" . "\e$(7$X\e(B")
 517     ("\e$(7"9#C\e(B" . "\e$(7$Y\e(B")
 518     ("\e$(7"G#C\e(B" . "\e$(7$[\e(B")
 519     ("\e$(7"H#C\e(B" . "\e$(7$\\e(B")
 520     ("\e$(7"(#C\e(B" . "\e$(7$]\e(B")
 521     ("\e$(7"!#D\e(B" . "\e$(7$`\e(B")
 522     ("\e$(7"##D\e(B" . "\e$(7$a\e(B")
 523     ("\e$(7"7#D\e(B" . "\e$(7$b\e(B")
 524     ("\e$(7"@#D\e(B" . "\e$(7$c\e(B")
 525     ("\e$(7"C#D\e(B" . "\e$(7$d\e(B")
 526     ("\e$(7"G#D\e(B" . "\e$(7$e\e(B")
 527     ("\e$(7"C#!\e(B" . "\e$(7%!\e(B")
 528     ("\e$(7"C##\e(B" . "\e$(7%"\e(B")
 529     ("\e$(7"C#(\e(B" . "\e$(7%$\e(B")
 530     ("\e$(7"C#0\e(B" . "\e$(7%&\e(B")
 531     ("\e$(7"C#2\e(B" . "\e$(7%'\e(B")
 532     ("\e$(7"C#4\e(B" . "\e$(7%(\e(B")
 533     ("\e$(7"C#7\e(B" . "\e$(7%)\e(B")
 534     ("\e$(7"C#9\e(B" . "\e$(7%*\e(B")
 535     ("\e$(7"D#!\e(B" . "\e$(7%@\e(B")
 536     ("\e$(7"D##\e(B" . "\e$(7%A\e(B")
 537     ("\e$(7"D#4\e(B" . "\e$(7!!\e(B") ; dummy 0x2121 added 2000/06/08 for transition l -> lng
 538     ("\e$(7"D#&\e(B" . "\e$(7%C\e(B")
 539     ("\e$(7"D#(\e(B" . "\e$(7%D\e(B")
 540     ("\e$(7"D#0\e(B" . "\e$(7%E\e(B")
 541     ("\e$(7"D#2\e(B" . "\e$(7%F\e(B")
 542     ("\e$(7"D#5\e(B" . "\e$(7%G\e(B")
 543     ("\e$(7"D#7\e(B" . "\e$(7%H\e(B")
 544     ("\e$(7"D#H\e(B" . "\e$(7%I\e(B")
 545     ("\e$(7"G#!\e(B" . "\e$(7%P\e(B")
 546     ("\e$(7"G##\e(B" . "\e$(7%Q\e(B")
 547     ("\e$(7"G#0\e(B" . "\e$(7%T\e(B")
 548     ("\e$(7"G#2\e(B" . "\e$(7%U\e(B")
 549     ("\e$(7"G#4\e(B" . "\e$(7%V\e(B")
 550     ("\e$(7"G#5\e(B" . "\e$(7%W\e(B")
 551     ("\e$(7"G#7\e(B" . "\e$(7%X\e(B")
 552     ("\e$(7"G#9\e(B" . "\e$(7%Y\e(B")))
 553
 554 (defconst tibetan-regexp
 555   (let ((l (list tibetan-precomposed-transcription-alist
 556                  tibetan-consonant-transcription-alist
 557                  tibetan-vowel-transcription-alist
 558                  tibetan-modifier-transcription-alist
 559                  tibetan-subjoined-transcription-alist))
 560         (separator "\\|")
 561         tail pattern)
 562     (while l
 563       (setq tail (car l) l (cdr l))
 564       (while tail
 565         (setq pattern (cons separator (cons (car (car tail)) pattern))
 566               tail (cdr tail))))
 567     (apply 'concat (nreverse (cdr pattern))))
 568   "Regexp matching a Tibetan transcription of a composable Tibetan sequence.
 569 The result of matching is to be used for indexing alists at conversion
 570 from a roman transcription to the corresponding Tibetan character.")
 571
 572 (defvar tibetan-precomposed-regexp
 573   (let ((l tibetan-precomposed-transcription-alist)
 574         temp)
 575     (setq temp "^\\(")
 576     (setq temp
 577           (concat temp (car (car l))))
 578     (setq l (cdr l))
 579     (while l
 580       (setq temp
 581             (concat temp "\\|" (car (car l))))
 582       (setq l (cdr l)))
 583     (concat temp "\\)"))
 584   "Regexp string to match a romanized Tibetan complex consonant.
 585 The result of matching is to be used for indexing alists when the input key
 586 from an input method is converted to the corresponding precomposed glyph.")
 587
 588 (defvar tibetan-precomposition-rule-regexp
 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 composition-function-table '(#xF00 . #xFD1)
 609                       '(("[\xF00-\xFD1]+" . font-shape-text)))
 610
 611 (provide 'tibetan)
 612
 613 ;;; arch-tag: 8d37c8d7-f95d-450f-9ec2-819e61fc79a7
 614 ;;; tibetan.el ends here