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