lisp/language/tibetan.el

   1 ;;; tibetan.el --- support for Tibetan language -*- coding: iso-2022-7bit; -*-
   2
   3 ;; Copyright (C) 1997, 2001, 2002, 2003, 2004, 2005, 2006, 2007, 2008
   4 ;;   Free Software Foundation, Inc.
   5 ;; Copyright (C) 1997, 1998, 1999, 2000, 2001, 2002, 2003, 2004, 2005,
   6 ;;   2006, 2007, 2008
   7 ;;   National Institute of Advanced Industrial Science and Technology (AIST)
   8 ;;   Registration Number H14PRO021
   9 ;; Copyright (C) 2003
  10 ;;   National Institute of Advanced Industrial Science and Technology (AIST)
  11 ;;   Registration Number H13PRO009
  12
  13 ;; Keywords: multilingual, Tibetan, i18n
  14
  15 ;; This file is part of GNU Emacs.
  16
  17 ;; GNU Emacs is free software; you can redistribute it and/or modify
  18 ;; it under the terms of the GNU General Public License as published by
  19 ;; the Free Software Foundation; either version 3, or (at your option)
  20 ;; any later version.
  21
  22 ;; GNU Emacs is distributed in the hope that it will be useful,
  23 ;; but WITHOUT ANY WARRANTY; without even the implied warranty of
  24 ;; MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
  25 ;; GNU General Public License for more details.
  26
  27 ;; You should have received a copy of the GNU General Public License
  28 ;; along with GNU Emacs; see the file COPYING.  If not, write to the
  29 ;; Free Software Foundation, Inc., 51 Franklin Street, Fifth Floor,
  30 ;; Boston, MA 02110-1301, USA.
  31
  32 ;; Author: Toru TOMABECHI, <Toru.Tomabechi@orient.unil.ch>
  33
  34 ;; Created: Feb. 17. 1997
  35
  36 ;; History:
  37 ;; 1997.03.13 Modification for special signs and punctuations.
  38
  39 ;;; Commentary:
  40
  41 ;;; Code:
  42
  43 ;;; Tibetan Character set.
  44 ;;; \x2130 -- \x234a is a subset of Unicode v.2 \x0f00 - \x0fb9
  45 ;;; with a slight modification. And there are some subjoined
  46 ;;; consonants which are not specified in Unicode.
  47 ;;; I hope I can add missing characters later.
  48 ;;;
  49 ;;;     00 01 02 03 04 05 06 07 08 09 0A 0B 0C 0D 0E 0F
  50 ;;;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)
  51 ;;;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,
  52 ;;;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
  53 ;;;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.
  54 ;;;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 ;
  55 ;;;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$(7!~\e(B // ;
  56 ;;;
  57 ;;;     00 01 02 03 04 05 06 07 08 09 0A 0B 0C 0D 0E 0F
  58 ;;;2220 // \e$(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$(7"/\e(B ; Base consonants
  59 ;;;2230 \e$(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$(7"?\e(B ; and
  60 ;;;2240 \e$(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$(7"O\e(B ; Vowel signs.
  61 ;;;2250 \e$(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$(7"_\e(B ; (\x2251 = vowel a)
  62 ;;;2260 \e$(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$(7"o\e(B ; Long vowels and
  63 ;;;2270 \e$(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
  64 ;;;                                                     ; atomically
  65 ;;;                                                     ; encoded.
  66 ;;;     00 01 02 03 04 05 06 07 08 09 0A 0B 0C 0D 0E 0F
  67 ;;;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
  68 ;;;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 ;
  69 ;;;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 ;
  70 ;;;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
  71 ;;;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
  72 ;;;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$(7#~\e(B // ; in Unicode.
  73 ;;;
  74 ;;;     00 01 02 03 04 05 06 07 08 09 0A 0B 0C 0D 0E 0F
  75 ;;;2420 // \e$(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
  76 ;;;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
  77 ;;;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.
  78 ;;;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
  79 ;;;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
  80 ;;;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
  81 ;;;                                                     ; when written on a
  82 ;;;     00 01 02 03 04 05 06 07 08 09 0A 0B 0C 0D 0E 0F ; file in Tibetan
  83 ;;;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.
  84 ;;;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 ;
  85 ;;;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 ;
  86 ;;;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 ;
  87 ;;;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 ;
  88 ;;;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 // ;
  89 ;;;
  90
  91
  92 (define-coding-system 'tibetan-iso-8bit
  93   "8-bit encoding for ASCII (MSB=0) and TIBETAN (MSB=1)."
  94   :coding-type 'iso-2022
  95   :mnemonic ?Q
  96   :designation [ascii tibetan nil nil]
  97   :charset-list '(ascii tibetan))
  98
  99 (define-coding-system-alias 'tibetan 'tibetan-iso-8bit)
 100
 101 (set-language-info-alist
 102  "Tibetan" '((charset tibetan tibetan-1-column)
 103              (coding-system tibetan-iso-8bit)
 104              (coding-priority iso-2022-7bit tibetan-iso-8bit)
 105              (input-method . "tibetan-wylie")
 106              (features tibet-util)
 107              (documentation . t)
 108              (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")))
 109
 110 ;; `\e$(7"A\e(B' is included in the pattern for subjoined consonants because we
 111 ;; treat it specially in tibetan-add-components.
 112 ;; modified by Tomabechi 1999/12/10
 113 ;; modified by Tomabechi 2000/06/08
 114 ;;          To allow infinite addition of vowels/modifiers
 115 ;;          as specified in Unicode v.3
 116 ;; \e$(7"A\e(B is removed from the class of subjoined. Tomabechi 2000/06/08
 117 ;; (for Unicode support)
 118 (defconst tibetan-composable-pattern
 119   "[\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]*"
 120   "Regexp matching a composable sequence of Tibetan characters.")
 121
 122 ;;;
 123 ;;; Definitions of conversion data.
 124 ;;;
 125
 126
 127 ;;; alists for tibetan char <-> transcription conversion
 128 ;;; longer transcription should come first
 129 (defconst tibetan-consonant-transcription-alist
 130   '(("tsh" . "\e$(7";\e(B")
 131     ("dzh" . "\e$(7"=\e(B")
 132     ("kSH" . "\e$(7"J\e(B")
 133     ("kh" . "\e$(7""\e(B")
 134     ("gh" . "\e$(7"$\e(B")
 135     ("ng" . "\e$(7"%\e(B")
 136     ("ch" . "\e$(7"'\e(B")
 137     ("ny" . "\e$(7"*\e(B")
 138     ("TH" . "\e$(7",\e(B")
 139     ("DH" . "\e$(7".\e(B")
 140     ("th" . "\e$(7"1\e(B")
 141     ("dh" . "\e$(7"3\e(B")
 142     ("ph" . "\e$(7"6\e(B")
 143     ("bh" . "\e$(7"8\e(B")
 144     ("ts" . "\e$(7":\e(B")
 145     ("dz" . "\e$(7"<\e(B")
 146     ("zh" . "\e$(7"?\e(B")
 147     ("sh" . "\e$(7"E\e(B")
 148     ("SH" . "\e$(7"F\e(B")
 149     ("k" . "\e$(7"!\e(B")
 150     ("g" . "\e$(7"#\e(B")
 151     ("c" . "\e$(7"&\e(B")
 152     ("j" . "\e$(7"(\e(B")
 153     ("T" . "\e$(7"+\e(B")
 154     ("D" . "\e$(7"-\e(B")
 155     ("N" . "\e$(7"/\e(B")
 156     ("t" . "\e$(7"0\e(B")
 157     ("d" . "\e$(7"2\e(B")
 158     ("n" . "\e$(7"4\e(B")
 159     ("p" . "\e$(7"5\e(B")
 160     ("b" . "\e$(7"7\e(B")
 161     ("m" . "\e$(7"9\e(B")
 162     ("w" . "\e$(7">\e(B")
 163     ("z" . "\e$(7"@\e(B")
 164     ("'" . "\e$(7"A\e(B")
 165     ("y" . "\e$(7"B\e(B")
 166     ("r" . "\e$(7"C\e(B")
 167     ("l" . "\e$(7"D\e(B")
 168     ("s" . "\e$(7"G\e(B")
 169     ("h" . "\e$(7"H\e(B")
 170     ("H" . "\e$(7"H\e(B")
 171     ("A" . "\e$(7"I\e(B")
 172     ;; Added by Tomabechi 1999/12/10
 173     ("R" . "\e$(7"K\e(B") ;; fixed form RA
 174     ))
 175
 176
 177 (defconst tibetan-vowel-transcription-alist
 178   '(
 179     ;; Composite Vowels
 180     ;; Added by Tomabechi 2000/06/08
 181     ("frr" . "\e$(7"X\e(B")
 182     ("fll" . "\e$(7"Z\e(B")
 183     ("fa" . "\e$(7"R\e(B")
 184     ("fi" . "\e$(7"T\e(B")
 185     ("fu" . "\e$(7"V\e(B")
 186     ("fr" . "\e$(7"W\e(B")
 187     ("fl" . "\e$(7"Y\e(B")
 188     ("fI" . "\e$(7"b\e(B")
 189     ;; Normal Vowels
 190     ("ai" . "\e$(7"\\e(B")
 191     ("au" . "\e$(7"^\e(B")
 192     ("ee" . "\e$(7"\\e(B")
 193     ("oo" . "\e$(7"^\e(B")
 194     ("a" . "\e$(7"Q\e(B")                 ; invisible vowel sign (\x2251)
 195     ("i" . "\e$(7"S\e(B")
 196     ("u" . "\e$(7"U\e(B")
 197     ("e" . "\e$(7"[\e(B")
 198     ("o" . "\e$(7"]\e(B")
 199     ("E" . "\e$(7"\\e(B")
 200     ("O" . "\e$(7"^\e(B")
 201     ("I" . "\e$(7"a\e(B")
 202     ("," . "\e$(7"e\e(B")                 ; idem.
 203     ))
 204
 205 (defconst tibetan-modifier-transcription-alist
 206   '(("M" . "\e$(7"_\e(B")
 207     ("~" . "\e$(7"c\e(B")
 208     ("`" . "\e$(7"d\e(B")
 209     ("x" . "\e$(7"i\e(B")
 210     ("X" . "\e$(7"j\e(B")
 211     ("v" . "\e$(7"g\e(B")
 212     ("V" . "\e$(7"h\e(B")
 213     ("q" . "\e$(7"k\e(B")
 214     ("Q" . "\e$(7"l\e(B")
 215     ("_/" . "\e$(7!I\e(B")
 216     ("_o" . "\e$(7!g\e(B")
 217     ("_O" . "\e$(7!e\e(B")))
 218
 219 (defconst tibetan-precomposed-transcription-alist
 220   '(("phyw" . "\e$(7$G\e(B")
 221     ("tshw" . "\e$(7$)\e(B")
 222     ("rtsw" . "\e$(7%.\e(B")
 223     ("khw" . "\e$(7$"\e(B")
 224     ("nyw" . "\e$(7$%\e(B")
 225     ("tsw" . "\e$(7$(\e(B")
 226     ("zhw" . "\e$(7$*\e(B")
 227     ("shw" . "\e$(7$.\e(B")
 228     ("khy" . "\e$(7$A\e(B")
 229     ("phy" . "\e$(7$D\e(B")
 230     ("khr" . "\e$(7$Q\e(B")
 231     ("thr" . "\e$(7$T\e(B")
 232     ("phr" . "\e$(7$W\e(B")
 233     ("shr" . "\e$(7$Z\e(B")
 234     ("dzr" . "\e$(7$^\e(B")
 235     ("grw" . "\e$(7$_\e(B")
 236     ("rng" . "\e$(7%#\e(B")
 237     ("rny" . "\e$(7%%\e(B")
 238     ("rts" . "\e$(7%+\e(B")
 239     ("rdz" . "\e$(7%,\e(B")
 240     ("rgw" . "\e$(7%-\e(B")
 241     ("rky" . "\e$(7%0\e(B")
 242     ("rgy" . "\e$(7%1\e(B")
 243     ("rmy" . "\e$(7%2\e(B")
 244     ("lng" . "\e$(7%B\e(B")
 245     ("sng" . "\e$(7%R\e(B")
 246     ("sny" . "\e$(7%S\e(B")
 247     ("sts" . "\e$(7%Z\e(B")
 248     ("sky" . "\e$(7%`\e(B")
 249     ("sgy" . "\e$(7%a\e(B")
 250     ("spy" . "\e$(7%b\e(B")
 251     ("sby" . "\e$(7%c\e(B")
 252     ("smy" . "\e$(7%d\e(B")
 253     ("skr" . "\e$(7%p\e(B")
 254     ("sgr" . "\e$(7%q\e(B")
 255     ("snr" . "\e$(7%r\e(B")
 256     ("spr" . "\e$(7%s\e(B")
 257     ("sbr" . "\e$(7%t\e(B")
 258     ("smr" . "\e$(7%u\e(B")
 259     ("kw" . "\e$(7$!\e(B")
 260     ("gw" . "\e$(7$#\e(B")
 261     ("cw" . "\e$(7$$\e(B")
 262     ("tw" . "\e$(7$&\e(B")
 263     ("dw" . "\e$(7$'\e(B")
 264     ("zw" . "\e$(7$+\e(B")
 265     ("rw" . "\e$(7$,\e(B")
 266     ("lw" . "\e$(7$-\e(B")
 267     ("sw" . "\e$(7$/\e(B")
 268     ("hw" . "\e$(7$0\e(B")
 269     ("ky" . "\e$(7$@\e(B")
 270     ("gy" . "\e$(7$B\e(B")
 271     ("py" . "\e$(7$C\e(B")
 272     ("by" . "\e$(7$E\e(B")
 273     ("my" . "\e$(7$F\e(B")
 274     ("kr" . "\e$(7$P\e(B")
 275     ("gr" . "\e$(7$R\e(B")
 276     ("tr" . "\e$(7$S\e(B")
 277     ("dr" . "\e$(7$U\e(B")
 278     ("pr" . "\e$(7$V\e(B")
 279     ("brk" . "\e$(7"7%!\e(B")
 280     ("brg" . "\e$(7"7%"\e(B")
 281     ("brng" . "\e$(7"7%#\e(B")
 282     ("brj" . "\e$(7"7%$\e(B")
 283     ("brny" . "\e$(7"7%%\e(B")
 284     ("brt" .  "\e$(7"7%&\e(B")
 285     ("brd" . "\e$(7"7%'\e(B")
 286     ("brn" . "\e$(7"7%(\e(B")
 287     ("brts" . "\e$(7"7%+\e(B")
 288     ("brdz" . "\e$(7"7%,\e(B")
 289     ("brl" . "\e$(7"7$d\e(B")
 290     ("br" . "\e$(7$X\e(B")
 291     ("mr" . "\e$(7$Y\e(B")
 292     ("sr" . "\e$(7$[\e(B")
 293     ("hr" . "\e$(7$\\e(B")
 294     ("jr" . "\e$(7$]\e(B")
 295     ("kl" . "\e$(7$`\e(B")
 296     ("gl" . "\e$(7$a\e(B")
 297     ("blt" . "\e$(7"7%E\e(B")
 298     ("bld" . "\e$(7"7%F\e(B")
 299     ("bl" . "\e$(7$b\e(B")
 300     ("zl" . "\e$(7$c\e(B")
 301     ("rl" . "\e$(7$d\e(B")
 302     ("sl" . "\e$(7$e\e(B")
 303     ("rk" . "\e$(7%!\e(B")
 304     ("rg" . "\e$(7%"\e(B")
 305     ("rj" . "\e$(7%$\e(B")
 306     ("rt" . "\e$(7%&\e(B")
 307     ("rd" . "\e$(7%'\e(B")
 308     ("rn" . "\e$(7%(\e(B")
 309     ("rb" . "\e$(7%)\e(B")
 310     ("rm" . "\e$(7%*\e(B")
 311     ("lk" . "\e$(7%@\e(B")
 312     ("lg" . "\e$(7%A\e(B")
 313     ("lc" . "\e$(7%C\e(B")
 314     ("lj" . "\e$(7%D\e(B")
 315     ("lt" . "\e$(7%E\e(B")
 316     ("ld" . "\e$(7%F\e(B")
 317     ("ln" . "\e$(7!!\e(B")                        ; dummy \x2121
 318     ("lp" . "\e$(7%G\e(B")
 319     ("lb" . "\e$(7%H\e(B")
 320     ("lh" . "\e$(7%I\e(B")
 321     ("sk" . "\e$(7%P\e(B")
 322     ("sg" . "\e$(7%Q\e(B")
 323     ("st" . "\e$(7%T\e(B")
 324     ("sd" . "\e$(7%U\e(B")
 325     ("sn" . "\e$(7%V\e(B")
 326     ("sp" . "\e$(7%W\e(B")
 327     ("sb" . "\e$(7%X\e(B")
 328     ("sm" . "\e$(7%Y\e(B"))
 329   )
 330
 331
 332 (defconst tibetan-subjoined-transcription-alist
 333   (sort '(("+k"  . "\e$(7#!\e(B")
 334           ("+kh" . "\e$(7#"\e(B")
 335           ("+g"  . "\e$(7##\e(B")
 336           ("+gh" . "\e$(7#$\e(B")
 337           ("+ng" . "\e$(7#%\e(B")
 338           ("+c"  . "\e$(7#&\e(B")
 339           ("+ch" . "\e$(7#'\e(B")
 340           ("+j"  . "\e$(7#(\e(B")
 341           ("+ny"  . "\e$(7#*\e(B")
 342           ("+T"  . "\e$(7#+\e(B")
 343           ("+TH" . "\e$(7#,\e(B")
 344           ("+D"  . "\e$(7#-\e(B")
 345           ("+DH" . "\e$(7#.\e(B")
 346           ("+N"  . "\e$(7#/\e(B")
 347           ("+t"  . "\e$(7#0\e(B")
 348           ("+th" . "\e$(7#1\e(B")
 349           ("+d"  . "\e$(7#2\e(B")
 350           ("+dh" . "\e$(7#3\e(B")
 351           ("+n"  . "\e$(7#4\e(B")
 352           ("+p"  . "\e$(7#5\e(B")
 353           ("+ph" . "\e$(7#6\e(B")
 354           ("+b"  . "\e$(7#7\e(B")
 355           ("+bh" . "\e$(7#8\e(B")
 356           ("+m"  . "\e$(7#9\e(B")
 357           ("+ts" . "\e$(7#:\e(B")
 358           ("+tsh" . "\e$(7#;\e(B")
 359           ("+dz" . "\e$(7#<\e(B")
 360           ("+dzh" . "\e$(7#=\e(B")
 361           ("+w"  . "\e$(7#>\e(B")
 362           ("+zh" . "\e$(7#?\e(B")
 363           ("+z"  . "\e$(7#@\e(B")
 364           ("+'"  . "\e$(7#A\e(B")
 365           ("+y"  . "\e$(7#B\e(B")
 366           ("+r"  . "\e$(7#C\e(B")
 367           ("+l"  . "\e$(7#D\e(B")
 368           ("+sh" . "\e$(7#E\e(B")
 369           ("+SH" . "\e$(7#F\e(B")
 370           ("+s"  . "\e$(7#G\e(B")
 371           ("+h"  . "\e$(7#H\e(B")
 372           ("+A"  . "\e$(7#I\e(B")
 373           ("+kSH" . "\e$(7#J\e(B")
 374           ;; Added by Tomabechi 1999/12/10
 375           ("+W" . "\e$(7#K\e(B") ;; fixed form subscribed WA
 376           ("+Y" . "\e$(7#L\e(B") ;; fixed form subscribed YA
 377           ("+R" . "\e$(7#M\e(B") ;; fixed form subscribed RA
 378           )
 379         (lambda (x y) (> (length (car x)) (length (car y))))))
 380
 381 ;;;
 382 ;;; alist for Tibetan base consonant <-> subjoined consonant conversion.
 383 ;;;
 384 (defconst tibetan-base-to-subjoined-alist
 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$(7"/\e(B" . "\e$(7#/\e(B")
 399     ("\e$(7"0\e(B" . "\e$(7#0\e(B")
 400     ("\e$(7"1\e(B" . "\e$(7#1\e(B")
 401     ("\e$(7"2\e(B" . "\e$(7#2\e(B")
 402     ("\e$(7"3\e(B" . "\e$(7#3\e(B")
 403     ("\e$(7"4\e(B" . "\e$(7#4\e(B")
 404     ("\e$(7"5\e(B" . "\e$(7#5\e(B")
 405     ("\e$(7"6\e(B" . "\e$(7#6\e(B")
 406     ("\e$(7"7\e(B" . "\e$(7#7\e(B")
 407     ("\e$(7"8\e(B" . "\e$(7#8\e(B")
 408     ("\e$(7"9\e(B" . "\e$(7#9\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$(7"@\e(B" . "\e$(7#@\e(B")
 416     ("\e$(7"A\e(B" . "\e$(7#A\e(B")
 417     ("\e$(7"B\e(B" . "\e$(7#B\e(B")
 418     ("\e$(7"C\e(B" . "\e$(7#C\e(B")
 419     ("\e$(7"D\e(B" . "\e$(7#D\e(B")
 420     ("\e$(7"E\e(B" . "\e$(7#E\e(B")
 421     ("\e$(7"F\e(B" . "\e$(7#F\e(B")
 422     ("\e$(7"G\e(B" . "\e$(7#G\e(B")
 423     ("\e$(7"H\e(B" . "\e$(7#H\e(B")
 424     ("\e$(7"I\e(B" . "\e$(7#I\e(B")
 425     ("\e$(7"J\e(B" . "\e$(7#J\e(B")
 426     ;; Added by Tomabechi 1999/12/10
 427     ("\e$(7"K\e(B" . "\e$(7#M\e(B") ;; Fixed form RA (224B->234D)
 428     ))
 429
 430 ;;; alist for Tibetan composite vowels (long i, vocalic r, etc.)
 431 ;;; New varialble. created by Tomabechi 2000/06/08
 432 (defconst tibetan-composite-vowel-alist
 433   '(;; LONG A
 434     ;; ("\e$(7"R\e(B" . ((bc . tc) ?\e$(7"R\e(B))
 435     ;; LONG I
 436     ("\e$(7"T\e(B" . (?\e$(7"R\e(B (tc . bc) ?\e$(7"S\e(B))
 437     ;; LONG U
 438     ("\e$(7"V\e(B" . (?\e$(7"R\e(B (bc . tc) ?\e$(7"U\e(B))
 439     ;; VOCALIC R
 440     ("\e$(7"W\e(B" . (?\e$(7#C\e(B (tc . bc) ?\e$(7"a\e(B))
 441     ;; LONG VOCALIC R
 442     ("\e$(7"X\e(B" . (?\e$(7#C\e(B (bc . tc) ?\e$(7"R\e(B (tc . bc) ?\e$(7"a\e(B))
 443     ;; VOCALIC L
 444     ("\e$(7"Y\e(B" . (?\e$(7#D\e(B (tc . bc) ?\e$(7"a\e(B))
 445     ;;\e$(7!;\e(BLONG VOCALIC L
 446     ("\e$(7"Z\e(B" . (?\e$(7#D\e(B (bc . tc) ?\e$(7"R\e(B (tc . bc) ?\e$(7"a\e(B))
 447     ;; LONG REVERSE I
 448     ("\e$(7"b\e(B" . (?\e$(7"R\e(B (tc . bc) ?\e$(7"a\e(B))
 449     ))
 450
 451
 452
 453 ;;;
 454 ;;; alist for Tibetan consonantic components <-> precomposed glyph conversion.
 455 ;;; (includes some punctuation conversion rules)
 456 ;;;
 457 (defconst tibetan-precomposition-rule-alist
 458   `(("\e$(7"6#B#>\e(B" . "\e$(7$G\e(B")
 459     ("\e$(7"##C#>\e(B" . "\e$(7$_\e(B")
 460     ("\e$(7";#>\e(B" . "\e$(7$)\e(B")
 461     ("\e$(7"C#:#>\e(B" . "\e$(7%.\e(B")
 462     ("\e$(7"C###>\e(B" . "\e$(7%-\e(B")
 463     ("\e$(7"C#!#B\e(B" . "\e$(7%0\e(B")
 464     ("\e$(7"C###B\e(B" . "\e$(7%1\e(B")
 465     ("\e$(7"C#9#B\e(B" . "\e$(7%2\e(B")
 466     ("\e$(7"G#!#B\e(B" . "\e$(7%`\e(B")
 467     ("\e$(7"G###B\e(B" . "\e$(7%a\e(B")
 468     ("\e$(7"G#5#B\e(B" . "\e$(7%b\e(B")
 469     ("\e$(7"G#7#B\e(B" . "\e$(7%c\e(B")
 470     ("\e$(7"G#9#B\e(B" . "\e$(7%d\e(B")
 471     ("\e$(7"G#!#C\e(B" . "\e$(7%p\e(B")
 472     ("\e$(7"G###C\e(B" . "\e$(7%q\e(B")
 473     ("\e$(7"G#4#C\e(B" . "\e$(7%r\e(B")
 474     ("\e$(7"G#5#C\e(B" . "\e$(7%s\e(B")
 475     ("\e$(7"G#7#C\e(B" . "\e$(7%t\e(B")
 476     ("\e$(7"G#9#C\e(B" . "\e$(7%u\e(B")
 477     ("\e$(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$(7"?#>\e(B" . "\e$(7$*\e(B")
 481     ("\e$(7"E#>\e(B" . "\e$(7$.\e(B")
 482     ("\e$(7""#B\e(B" . "\e$(7$A\e(B")
 483     ("\e$(7"6#B\e(B" . "\e$(7$D\e(B")
 484     ("\e$(7""#C\e(B" . "\e$(7$Q\e(B")
 485     ("\e$(7"1#C\e(B" . "\e$(7$T\e(B")
 486     ("\e$(7"6#C\e(B" . "\e$(7$W\e(B")
 487     ("\e$(7"E#C\e(B" . "\e$(7$Z\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"C#<\e(B" . "\e$(7%,\e(B")
 493     ("\e$(7"D#%\e(B" . "\e$(7%B\e(B")
 494     ("\e$(7"G#%\e(B" . "\e$(7%R\e(B")
 495     ("\e$(7"G#*\e(B" . "\e$(7%S\e(B")
 496     ("\e$(7"G#:\e(B" . "\e$(7%Z\e(B")
 497     ("\e$(7"!#>\e(B" . "\e$(7$!\e(B")
 498     ("\e$(7"##>\e(B" . "\e$(7$#\e(B")
 499     ("\e$(7"&#>\e(B" . "\e$(7$$\e(B")
 500     ("\e$(7"0#>\e(B" . "\e$(7$&\e(B")
 501     ("\e$(7"2#>\e(B" . "\e$(7$'\e(B")
 502     ("\e$(7"@#>\e(B" . "\e$(7$+\e(B")
 503     ("\e$(7"C#>\e(B" . "\e$(7$,\e(B")
 504     ("\e$(7"D#>\e(B" . "\e$(7$-\e(B")
 505     ("\e$(7"G#>\e(B" . "\e$(7$/\e(B")
 506     ("\e$(7"H#>\e(B" . "\e$(7$0\e(B")
 507     ("\e$(7"!#B\e(B" . "\e$(7$@\e(B")
 508     ("\e$(7"##B\e(B" . "\e$(7$B\e(B")
 509     ("\e$(7"5#B\e(B" . "\e$(7$C\e(B")
 510     ("\e$(7"7#B\e(B" . "\e$(7$E\e(B")
 511     ("\e$(7"9#B\e(B" . "\e$(7$F\e(B")
 512     ("\e$(7"!#C\e(B" . "\e$(7$P\e(B")
 513     ("\e$(7"##C\e(B" . "\e$(7$R\e(B")
 514     ("\e$(7"0#C\e(B" . "\e$(7$S\e(B")
 515     ("\e$(7"2#C\e(B" . "\e$(7$U\e(B")
 516     ("\e$(7"5#C\e(B" . "\e$(7$V\e(B")
 517     ("\e$(7"7#C\e(B" . "\e$(7$X\e(B")
 518     ("\e$(7"9#C\e(B" . "\e$(7$Y\e(B")
 519     ("\e$(7"G#C\e(B" . "\e$(7$[\e(B")
 520     ("\e$(7"H#C\e(B" . "\e$(7$\\e(B")
 521     ("\e$(7"(#C\e(B" . "\e$(7$]\e(B")
 522     ("\e$(7"!#D\e(B" . "\e$(7$`\e(B")
 523     ("\e$(7"##D\e(B" . "\e$(7$a\e(B")
 524     ("\e$(7"7#D\e(B" . "\e$(7$b\e(B")
 525     ("\e$(7"@#D\e(B" . "\e$(7$c\e(B")
 526     ("\e$(7"C#D\e(B" . "\e$(7$d\e(B")
 527     ("\e$(7"G#D\e(B" . "\e$(7$e\e(B")
 528     ("\e$(7"C#!\e(B" . "\e$(7%!\e(B")
 529     ("\e$(7"C##\e(B" . "\e$(7%"\e(B")
 530     ("\e$(7"C#(\e(B" . "\e$(7%$\e(B")
 531     ("\e$(7"C#0\e(B" . "\e$(7%&\e(B")
 532     ("\e$(7"C#2\e(B" . "\e$(7%'\e(B")
 533     ("\e$(7"C#4\e(B" . "\e$(7%(\e(B")
 534     ("\e$(7"C#7\e(B" . "\e$(7%)\e(B")
 535     ("\e$(7"C#9\e(B" . "\e$(7%*\e(B")
 536     ("\e$(7"D#!\e(B" . "\e$(7%@\e(B")
 537     ("\e$(7"D##\e(B" . "\e$(7%A\e(B")
 538     ("\e$(7"D#4\e(B" . "\e$(7!!\e(B") ; dummy 0x2121 added 2000/06/08 for transition l -> lng
 539     ("\e$(7"D#&\e(B" . "\e$(7%C\e(B")
 540     ("\e$(7"D#(\e(B" . "\e$(7%D\e(B")
 541     ("\e$(7"D#0\e(B" . "\e$(7%E\e(B")
 542     ("\e$(7"D#2\e(B" . "\e$(7%F\e(B")
 543     ("\e$(7"D#5\e(B" . "\e$(7%G\e(B")
 544     ("\e$(7"D#7\e(B" . "\e$(7%H\e(B")
 545     ("\e$(7"D#H\e(B" . "\e$(7%I\e(B")
 546     ("\e$(7"G#!\e(B" . "\e$(7%P\e(B")
 547     ("\e$(7"G##\e(B" . "\e$(7%Q\e(B")
 548     ("\e$(7"G#0\e(B" . "\e$(7%T\e(B")
 549     ("\e$(7"G#2\e(B" . "\e$(7%U\e(B")
 550     ("\e$(7"G#4\e(B" . "\e$(7%V\e(B")
 551     ("\e$(7"G#5\e(B" . "\e$(7%W\e(B")
 552     ("\e$(7"G#7\e(B" . "\e$(7%X\e(B")
 553     ("\e$(7"G#9\e(B" . "\e$(7%Y\e(B")))
 554
 555 (defconst tibetan-regexp
 556   (let ((l (list tibetan-precomposed-transcription-alist
 557                  tibetan-consonant-transcription-alist
 558                  tibetan-vowel-transcription-alist
 559                  tibetan-modifier-transcription-alist
 560                  tibetan-subjoined-transcription-alist))
 561         (separator "\\|")
 562         tail pattern)
 563     (while l
 564       (setq tail (car l) l (cdr l))
 565       (while tail
 566         (setq pattern (cons separator (cons (car (car tail)) pattern))
 567               tail (cdr tail))))
 568     (apply 'concat (nreverse (cdr pattern))))
 569   "Regexp matching a Tibetan transcription of a composable Tibetan sequence.
 570 The result of matching is to be used for indexing alists at conversion
 571 from a roman transcription to the corresponding Tibetan character.")
 572
 573 (defvar tibetan-precomposed-regexp
 574   (let ((l tibetan-precomposed-transcription-alist)
 575         temp)
 576     (setq temp "^\\(")
 577     (setq temp
 578           (concat temp (car (car l))))
 579     (setq l (cdr l))
 580     (while l
 581       (setq temp
 582             (concat temp "\\|" (car (car l))))
 583       (setq l (cdr l)))
 584     (concat temp "\\)"))
 585   "Regexp string to match a romanized Tibetan complex consonant.
 586 The result of matching is to be used for indexing alists when the input key
 587 from an input method is converted to the corresponding precomposed glyph.")
 588
 589 (defvar tibetan-precomposition-rule-regexp
 590   (let ((l tibetan-precomposition-rule-alist)
 591         temp)
 592     (setq temp "\\(")
 593     (setq temp (concat temp (car (car l))))
 594     (setq l (cdr l))
 595     (while l
 596       (setq temp (concat temp "\\|" (car (car l))))
 597       (setq l (cdr l)))
 598     (concat temp "\\)"))
 599   "Regexp string to match a sequence of Tibetan consonantic components, i.e.,
 600 one base consonant and one or more subjoined consonants.
 601 The result of matching is to be used for indexing alist when the component
 602 sequence is converted to the corresponding precomposed glyph.
 603 This also matches some punctuation characters which need conversion.")
 604
 605 (defvar tibetan-decomposed nil)
 606 (defvar tibetan-decomposed-temp nil)
 607
 608 ;; For automatic composition.
 609 (set-char-table-range composition-function-table '(#xF00 . #xFD1)
 610                       '(("[\xF00-\xFD1]+" . tibetan-composition-function)))
 611
 612 (provide 'tibetan)
 613
 614 ;; arch-tag: 8d37c8d7-f95d-450f-9ec2-819e61fc79a7
 615 ;;; tibetan.el ends here