778
|
1 ;;; tibetan.el --- support for Tibetan language -*- coding: iso-2022-7bit; -*-
|
771
|
2
|
|
3 ;; Copyright (C) 1997 Electrotechnical Laboratory, JAPAN.
|
778
|
4 ;; Copyright (C) 2002 Ben Wing.
|
771
|
5 ;; Licensed to the Free Software Foundation.
|
|
6
|
|
7 ;; Keywords: multilingual, Tibetan
|
|
8
|
|
9 ;; This file is part of XEmacs.
|
|
10
|
|
11 ;; XEmacs is free software; you can redistribute it and/or modify it
|
|
12 ;; under the terms of the GNU General Public License as published by
|
|
13 ;; the Free Software Foundation; either version 2, or (at your option)
|
|
14 ;; any later version.
|
|
15
|
|
16 ;; XEmacs is distributed in the hope that it will be useful, but
|
|
17 ;; WITHOUT ANY WARRANTY; without even the implied warranty of
|
|
18 ;; MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
|
|
19 ;; General Public License for more details.
|
|
20
|
|
21 ;; You should have received a copy of the GNU General Public License
|
|
22 ;; along with XEmacs; see the file COPYING. If not, write to the Free
|
|
23 ;; Software Foundation, Inc., 59 Temple Place - Suite 330, Boston, MA
|
|
24 ;; 02111-1307, USA.
|
|
25
|
778
|
26 ;;; Synched up with: Emacs 21.1 (language/tibetan.el).
|
771
|
27
|
|
28 ;; Author: Toru TOMABECHI, <Toru.Tomabechi@orient.unil.ch>
|
|
29
|
|
30 ;; Created: Feb. 17. 1997
|
|
31
|
|
32 ;; History:
|
|
33 ;; 1997.03.13 Modification for special signs and punctuations.
|
|
34
|
778
|
35 ;;; Commentary:
|
|
36
|
771
|
37 ;;; Code:
|
|
38
|
|
39 ;;; Tibetan Character set.
|
|
40 ;;; \x2130 -- \x234a is a subset of Unicode v.2 \x0f00 - \x0fb9
|
|
41 ;;; with a slight modification. And there are some subjoined
|
|
42 ;;; consonants which are not specified in Unicode.
|
|
43 ;;; I hope I can add missing characters later.
|
|
44 ;;;
|
|
45 ;;; 00 01 02 03 04 05 06 07 08 09 0A 0B 0C 0D 0E 0F
|
|
46 ;;;2120 // $(7!!(B $(7!"(B $(7!#(B $(7!$(B $(7!%(B $(7!&(B $(7!'(B $(7!((B $(7!)(B $(7!*(B $(7!+(B $(7!,(B $(7!-(B $(7!.(B $(7!/(B ; obsolete glyphs (2123-5)
|
|
47 ;;;2130 $(7!0(B $(7!1(B $(7!2(B $(7!3(B $(7!4(B $(7!5(B $(7!6(B $(7!7(B $(7!8(B $(7!9(B $(7!:(B $(7!;(B $(7!<(B $(7!=(B $(7!>(B $(7!?(B ; Punctuations,
|
|
48 ;;;2140 $(7!@(B $(7!A(B $(7!B(B $(7!C(B $(7!D(B $(7!E(B $(7!F(B $(7!G(B $(7!H(B $(7!I(B $(7!J(B $(7!K(B $(7!L(B $(7!M(B $(7!N(B $(7!O(B ; Digits and
|
|
49 ;;;2150 $(7!P(B $(7!Q(B $(7!R(B $(7!S(B $(7!T(B $(7!U(B $(7!V(B $(7!W(B $(7!X(B $(7!Y(B $(7!Z(B $(7![(B $(7!\(B $(7!](B $(7!^(B $(7!_(B ; Special signs.
|
|
50 ;;;2160 $(7!`(B $(7!a(B $(7!b(B $(7!c(B $(7!d(B $(7!e(B $(7!f(B $(7!g(B $(7!h(B $(7!i(B $(7!j(B $(7!k(B $(7!l(B $(7!m(B $(7!n(B $(7!o(B ;
|
|
51 ;;;2170 $(7!p(B $(7!q(B $(7!r(B $(7!s(B $(7!t(B $(7!u(B $(7!v(B $(7!w(B $(7!x(B $(7!y(B $(7!z(B $(7!{(B $(7!|(B $(7!}(B $(7!~(B // ;
|
|
52 ;;;
|
|
53 ;;; 00 01 02 03 04 05 06 07 08 09 0A 0B 0C 0D 0E 0F
|
|
54 ;;;2220 // $(7"!(B $(7""(B $(7"#(B $(7"$(B $(7"%(B $(7"&(B $(7"'(B $(7"((B $(7")(B $(7"*(B $(7"+(B $(7",(B $(7"-(B $(7".(B $(7"/(B ; Base consonants
|
|
55 ;;;2230 $(7"0(B $(7"1(B $(7"2(B $(7"3(B $(7"4(B $(7"5(B $(7"6(B $(7"7(B $(7"8(B $(7"9(B $(7":(B $(7";(B $(7"<(B $(7"=(B $(7">(B $(7"?(B ; and
|
|
56 ;;;2240 $(7"@(B $(7"A(B $(7"B(B $(7"C(B $(7"D(B $(7"E(B $(7"F(B $(7"G(B $(7"H(B $(7"I(B $(7"J(B $(7"K(B $(7"L(B $(7"M(B $(7"N(B $(7"O(B ; Vowel signs.
|
|
57 ;;;2250 $(7"P(B $(7"Q(B $(7"R(B $(7"S(B $(7"T(B $(7"U(B $(7"V(B $(7"W(B $(7"X(B $(7"Y(B $(7"Z(B $(7"[(B $(7"\(B $(7"](B $(7"^(B $(7"_(B ; (\x2251 = vowel a)
|
|
58 ;;;2260 $(7"`(B $(7"a(B $(7"b(B $(7"c(B $(7"d(B $(7"e(B $(7"f(B $(7"g(B $(7"h(B $(7"i(B $(7"j(B $(7"k(B $(7"l(B $(7"m(B $(7"n(B $(7"o(B ; Long vowels and
|
|
59 ;;;2270 $(7"p(B $(7"q(B $(7"r(B $(7"s(B $(7"t(B $(7"u(B $(7"v(B $(7"w(B $(7"x(B $(7"y(B $(7"z(B $(7"{(B $(7"|(B $(7"}(B $(7"~(B // ; vocalic r, l ARE
|
|
60 ;;; ; atomically
|
|
61 ;;; ; encoded.
|
|
62 ;;; 00 01 02 03 04 05 06 07 08 09 0A 0B 0C 0D 0E 0F
|
|
63 ;;;2320 // $(7#!(B $(7#"(B $(7##(B $(7#$(B $(7#%(B $(7#&(B $(7#'(B $(7#((B $(7#)(B $(7#*(B $(7#+(B $(7#,(B $(7#-(B $(7#.(B $(7#/(B ; Subjoined consonants
|
|
64 ;;;2330 $(7#0(B $(7#1(B $(7#2(B $(7#3(B $(7#4(B $(7#5(B $(7#6(B $(7#7(B $(7#8(B $(7#9(B $(7#:(B $(7#;(B $(7#<(B $(7#=(B $(7#>(B $(7#?(B ;
|
|
65 ;;;2340 $(7#@(B $(7#A(B $(7#B(B $(7#C(B $(7#D(B $(7#E(B $(7#F(B $(7#G(B $(7#H(B $(7#I(B $(7#J(B $(7#K(B $(7#L(B $(7#M(B $(7#N(B $(7#O(B ;
|
|
66 ;;;2350 $(7#P(B $(7#Q(B $(7#R(B $(7#S(B $(7#T(B $(7#U(B $(7#V(B $(7#W(B $(7#X(B $(7#Y(B $(7#Z(B $(7#[(B $(7#\(B $(7#](B $(7#^(B $(7#_(B ; Hereafter, the chars
|
|
67 ;;;2360 $(7#`(B $(7#a(B $(7#b(B $(7#c(B $(7#d(B $(7#e(B $(7#f(B $(7#g(B $(7#h(B $(7#i(B $(7#j(B $(7#k(B $(7#l(B $(7#m(B $(7#n(B $(7#o(B ; are not specified
|
|
68 ;;;2370 $(7#p(B $(7#q(B $(7#r(B $(7#s(B $(7#t(B $(7#u(B $(7#v(B $(7#w(B $(7#x(B $(7#y(B $(7#z(B $(7#{(B $(7#|(B $(7#}(B $(7#~(B // ; in Unicode.
|
|
69 ;;;
|
|
70 ;;; 00 01 02 03 04 05 06 07 08 09 0A 0B 0C 0D 0E 0F
|
|
71 ;;;2420 // $(7$!(B $(7$"(B $(7$#(B $(7$$(B $(7$%(B $(7$&(B $(7$'(B $(7$((B $(7$)(B $(7$*(B $(7$+(B $(7$,(B $(7$-(B $(7$.(B $(7$/(B ; Precomposed
|
|
72 ;;;2430 $(7$0(B $(7$1(B $(7$2(B $(7$3(B $(7$4(B $(7$5(B $(7$6(B $(7$7(B $(7$8(B $(7$9(B $(7$:(B $(7$;(B $(7$<(B $(7$=(B $(7$>(B $(7$?(B ; consonants for
|
|
73 ;;;2440 $(7$@(B $(7$A(B $(7$B(B $(7$C(B $(7$D(B $(7$E(B $(7$F(B $(7$G(B $(7$H(B $(7$I(B $(7$J(B $(7$K(B $(7$L(B $(7$M(B $(7$N(B $(7$O(B ; ordinary Tibetan.
|
|
74 ;;;2450 $(7$P(B $(7$Q(B $(7$R(B $(7$S(B $(7$T(B $(7$U(B $(7$V(B $(7$W(B $(7$X(B $(7$Y(B $(7$Z(B $(7$[(B $(7$\(B $(7$](B $(7$^(B $(7$_(B ; They are decomposed
|
|
75 ;;;2460 $(7$`(B $(7$a(B $(7$b(B $(7$c(B $(7$d(B $(7$e(B $(7$f(B $(7$g(B $(7$h(B $(7$i(B $(7$j(B $(7$k(B $(7$l(B $(7$m(B $(7$n(B $(7$o(B ; into base and
|
|
76 ;;;2470 $(7$p(B $(7$q(B $(7$r(B $(7$s(B $(7$t(B $(7$u(B $(7$v(B $(7$w(B $(7$x(B $(7$y(B $(7$z(B $(7${(B $(7$|(B $(7$}(B $(7$~(B // ; subjoined consonants
|
|
77 ;;; ; when written on a
|
|
78 ;;; 00 01 02 03 04 05 06 07 08 09 0A 0B 0C 0D 0E 0F ; file in Tibetan
|
|
79 ;;;2520 // $(7%!(B $(7%"(B $(7%#(B $(7%$(B $(7%%(B $(7%&(B $(7%'(B $(7%((B $(7%)(B $(7%*(B $(7%+(B $(7%,(B $(7%-(B $(7%.(B $(7%/(B ; coding system.
|
|
80 ;;;2530 $(7%0(B $(7%1(B $(7%2(B $(7%3(B $(7%4(B $(7%5(B $(7%6(B $(7%7(B $(7%8(B $(7%9(B $(7%:(B $(7%;(B $(7%<(B $(7%=(B $(7%>(B $(7%?(B ;
|
|
81 ;;;2540 $(7%@(B $(7%A(B $(7%B(B $(7%C(B $(7%D(B $(7%E(B $(7%F(B $(7%G(B $(7%H(B $(7%I(B $(7%J(B $(7%K(B $(7%L(B $(7%M(B $(7%N(B $(7%O(B ;
|
|
82 ;;;2550 $(7%P(B $(7%Q(B $(7%R(B $(7%S(B $(7%T(B $(7%U(B $(7%V(B $(7%W(B $(7%X(B $(7%Y(B $(7%Z(B $(7%[(B $(7%\(B $(7%](B $(7%^(B $(7%_(B ;
|
|
83 ;;;2560 $(7%`(B $(7%a(B $(7%b(B $(7%c(B $(7%d(B $(7%e(B $(7%f(B $(7%g(B $(7%h(B $(7%i(B $(7%j(B $(7%k(B $(7%l(B $(7%m(B $(7%n(B $(7%o(B ;
|
|
84 ;;;2570 $(7%p(B $(7%q(B $(7%r(B $(7%s(B $(7%t(B $(7%u(B $(7%v(B $(7%w(B $(7%x(B $(7%y(B $(7%z(B $(7%{(B $(7%|(B $(7%}(B $(7%~(B // ;
|
|
85 ;;;
|
|
86
|
778
|
87 (make-charset 'tibetan-1-column "Tibetan 1 column glyph"
|
|
88 '(dimension
|
|
89 2
|
3659
|
90 registries ["MuleTibetan-1"]
|
778
|
91 chars 94
|
|
92 columns 1
|
|
93 direction l2r
|
|
94 final ?8
|
|
95 graphic 0
|
|
96 short-name "Tibetan 1-col"
|
|
97 long-name "Tibetan 1 column"
|
|
98 ))
|
|
99
|
|
100 ;; Tibetan script.
|
|
101 (make-charset 'tibetan "Tibetan characters"
|
|
102 '(dimension
|
|
103 2
|
3659
|
104 registries ["MuleTibetan-2"]
|
778
|
105 chars 94
|
|
106 columns 2
|
|
107 direction l2r
|
|
108 final ?7
|
|
109 graphic 0
|
|
110 short-name "Tibetan 2-col"
|
|
111 long-name "Tibetan 2 column"
|
|
112 ))
|
771
|
113
|
|
114 ; (make-coding-system
|
|
115 ; 'tibetan-iso-8bit 2 ?Q
|
|
116 ; "8-bit encoding for ASCII (MSB=0) and TIBETAN (MSB=1)"
|
|
117 ; '(ascii tibetan nil nil
|
|
118 ; nil nil)
|
|
119 ; '((safe-charsets ascii tibetan)
|
|
120 ; (post-read-conversion . tibetan-post-read-conversion)
|
|
121 ; (pre-write-conversion . tibetan-pre-write-conversion)))
|
|
122
|
|
123 (make-coding-system
|
|
124 'tibetan-iso-8bit 'iso2022 "Tibetan/8-bit"
|
|
125 '(charset-g0 ascii
|
|
126 charset-g1 tibetan
|
|
127 mnemonic "Tibetan"
|
|
128 safe-charsets (ascii tibetan)
|
|
129 post-read-conversion tibetan-post-read-conversion
|
|
130 pre-write-conversion tibetan-pre-write-conversion
|
|
131 documentation "8-bit encoding for ASCII (MSB=0) and TIBETAN (MSB=1)"))
|
|
132
|
|
133 (define-coding-system-alias 'tibetan 'tibetan-iso-8bit)
|
|
134
|
|
135 (set-language-info-alist
|
|
136 "Tibetan" '((charset tibetan tibetan-1-column)
|
|
137 (coding-system tibetan-iso-8bit)
|
|
138 (coding-priority iso-2022-7bit tibetan-iso-8bit)
|
3970
|
139 ;; Not available in packages.
|
|
140 ;; (input-method . "tibetan-wylie")
|
771
|
141 (features tibet-util)
|
|
142 (documentation . t)
|
|
143 (sample-text
|
|
144 . (tibetan-compose-string
|
|
145 (copy-sequence
|
|
146 "Tibetan (4$(7"7r'"]0"7"]14"20"21!;4%P0"G#!"Q14"20"21!;(B) $(7!4!5!5!>4"70"714$P0"!#C"Q1!;4"Er'"S0"E"S14"G0"G1!;4"70"714"2r'"[0"2"[1!;4"Dr'"[0"D"[14"#0"#14"G0"G1!>4"Ir'"]r'"_0"I"]"_1!;4"90"9"Q1!;4"/r'"S0"/"S1!;4"50"5"Q14#2x!#9r'"[0"2#9"[1!;4"Hx!"Rx!"Ur'"c0"H"A"U"c1!>(B")))))
|
|
147
|
|
148
|
|
149 ;; `$(7"A(B' is included in the pattern for subjoined consonants because we
|
|
150 ;; treat it specially in tibetan-add-components.
|
|
151 ;; modified by Tomabechi 1999/12/10
|
|
152 ;; modified by Tomabechi 2000/06/08
|
|
153 ;; To allow infinite addition of vowels/modifiers
|
|
154 ;; as specified in Unicode v.3
|
|
155 ;; $(7"A(B is removed from the class of subjoined. Tomabechi 2000/06/08
|
|
156 ;; (for Unicode support)
|
|
157 (defconst tibetan-composable-pattern
|
|
158 "[$(7"!(B-$(7"J"K(B][$(7#!(B-$(7#J#K#L#M(B]*[$(7"Q"R"S(B-$(7"^"a"b"e(B]*[$(7"_"c"d"g(B-$(7"l!I!e!g(B]*"
|
|
159 "Regexp matching a composable sequence of Tibetan characters.")
|
|
160
|
|
161 ;; Register a function to compose Tibetan characters.
|
778
|
162 (put-char-table 'tibetan
|
|
163 (list (cons tibetan-composable-pattern
|
|
164 'tibetan-composition-function))
|
|
165 composition-function-table)
|
771
|
166
|
|
167 ;;;
|
|
168 ;;; Definitions of conversion data.
|
|
169 ;;;
|
|
170
|
|
171
|
|
172 ;;; alists for tibetan char <-> transcription conversion
|
|
173 ;;; longer transcription should come first
|
|
174 (defconst tibetan-consonant-transcription-alist
|
|
175 '(("tsh" . "$(7";(B")
|
|
176 ("dzh" . "$(7"=(B")
|
|
177 ("kSH" . "$(7"J(B")
|
|
178 ("kh" . "$(7""(B")
|
|
179 ("gh" . "$(7"$(B")
|
|
180 ("ng" . "$(7"%(B")
|
|
181 ("ch" . "$(7"'(B")
|
|
182 ("ny" . "$(7"*(B")
|
|
183 ("TH" . "$(7",(B")
|
|
184 ("DH" . "$(7".(B")
|
|
185 ("th" . "$(7"1(B")
|
|
186 ("dh" . "$(7"3(B")
|
|
187 ("ph" . "$(7"6(B")
|
|
188 ("bh" . "$(7"8(B")
|
|
189 ("ts" . "$(7":(B")
|
|
190 ("dz" . "$(7"<(B")
|
|
191 ("zh" . "$(7"?(B")
|
|
192 ("sh" . "$(7"E(B")
|
|
193 ("SH" . "$(7"F(B")
|
|
194 ("k" . "$(7"!(B")
|
|
195 ("g" . "$(7"#(B")
|
|
196 ("c" . "$(7"&(B")
|
|
197 ("j" . "$(7"((B")
|
|
198 ("T" . "$(7"+(B")
|
|
199 ("D" . "$(7"-(B")
|
|
200 ("N" . "$(7"/(B")
|
|
201 ("t" . "$(7"0(B")
|
|
202 ("d" . "$(7"2(B")
|
|
203 ("n" . "$(7"4(B")
|
|
204 ("p" . "$(7"5(B")
|
|
205 ("b" . "$(7"7(B")
|
|
206 ("m" . "$(7"9(B")
|
|
207 ("w" . "$(7">(B")
|
|
208 ("z" . "$(7"@(B")
|
|
209 ("'" . "$(7"A(B")
|
|
210 ("y" . "$(7"B(B")
|
|
211 ("r" . "$(7"C(B")
|
|
212 ("l" . "$(7"D(B")
|
|
213 ("s" . "$(7"G(B")
|
|
214 ("h" . "$(7"H(B")
|
|
215 ("H" . "$(7"H(B")
|
|
216 ("A" . "$(7"I(B")
|
|
217 ;; Added by Tomabechi 1999/12/10
|
|
218 ("R" . "$(7"K(B") ;; fixed form RA
|
|
219 ))
|
|
220
|
|
221
|
|
222 (defconst tibetan-vowel-transcription-alist
|
|
223 '(
|
|
224 ;; Composite Vowels
|
|
225 ;; Added by Tomabechi 2000/06/08
|
|
226 ("frr" . "$(7"X(B")
|
|
227 ("fll" . "$(7"Z(B")
|
|
228 ("fa" . "$(7"R(B")
|
|
229 ("fi" . "$(7"T(B")
|
|
230 ("fu" . "$(7"V(B")
|
|
231 ("fr" . "$(7"W(B")
|
|
232 ("fl" . "$(7"Y(B")
|
|
233 ("fI" . "$(7"b(B")
|
|
234 ;; Normal Vowels
|
|
235 ("ai" . "$(7"\(B")
|
|
236 ("au" . "$(7"^(B")
|
|
237 ("ee" . "$(7"\(B")
|
|
238 ("oo" . "$(7"^(B")
|
|
239 ("a" . "$(7"Q(B") ; invisible vowel sign (\x2251)
|
|
240 ("i" . "$(7"S(B")
|
|
241 ("u" . "$(7"U(B")
|
|
242 ("e" . "$(7"[(B")
|
|
243 ("o" . "$(7"](B")
|
|
244 ("E" . "$(7"\(B")
|
|
245 ("O" . "$(7"^(B")
|
|
246 ("I" . "$(7"a(B")
|
|
247 ("," . "$(7"e(B") ; idem.
|
|
248 ))
|
|
249
|
|
250 (defconst tibetan-modifier-transcription-alist
|
|
251 '(("M" . "$(7"_(B")
|
|
252 ("~" . "$(7"c(B")
|
|
253 ("`" . "$(7"d(B")
|
|
254 ("x" . "$(7"i(B")
|
|
255 ("X" . "$(7"j(B")
|
|
256 ("v" . "$(7"g(B")
|
|
257 ("V" . "$(7"h(B")
|
|
258 ("q" . "$(7"k(B")
|
|
259 ("Q" . "$(7"l(B")
|
|
260 ("_/" . "$(7!I(B")
|
|
261 ("_o" . "$(7!g(B")
|
|
262 ("_O" . "$(7!e(B")))
|
|
263
|
|
264 (defconst tibetan-precomposed-transcription-alist
|
|
265 '(("phyw" . "$(7$G(B")
|
|
266 ("tshw" . "$(7$)(B")
|
|
267 ("rtsw" . "$(7%.(B")
|
|
268 ("khw" . "$(7$"(B")
|
|
269 ("nyw" . "$(7$%(B")
|
|
270 ("tsw" . "$(7$((B")
|
|
271 ("zhw" . "$(7$*(B")
|
|
272 ("shw" . "$(7$.(B")
|
|
273 ("khy" . "$(7$A(B")
|
|
274 ("phy" . "$(7$D(B")
|
|
275 ("khr" . "$(7$Q(B")
|
|
276 ("thr" . "$(7$T(B")
|
|
277 ("phr" . "$(7$W(B")
|
|
278 ("shr" . "$(7$Z(B")
|
|
279 ("dzr" . "$(7$^(B")
|
|
280 ("grw" . "$(7$_(B")
|
|
281 ("rng" . "$(7%#(B")
|
|
282 ("rny" . "$(7%%(B")
|
|
283 ("rts" . "$(7%+(B")
|
|
284 ("rdz" . "$(7%,(B")
|
|
285 ("rgw" . "$(7%-(B")
|
|
286 ("rky" . "$(7%0(B")
|
|
287 ("rgy" . "$(7%1(B")
|
|
288 ("rmy" . "$(7%2(B")
|
|
289 ("lng" . "$(7%B(B")
|
|
290 ("sng" . "$(7%R(B")
|
|
291 ("sny" . "$(7%S(B")
|
|
292 ("sts" . "$(7%Z(B")
|
|
293 ("sky" . "$(7%`(B")
|
|
294 ("sgy" . "$(7%a(B")
|
|
295 ("spy" . "$(7%b(B")
|
|
296 ("sby" . "$(7%c(B")
|
|
297 ("smy" . "$(7%d(B")
|
|
298 ("skr" . "$(7%p(B")
|
|
299 ("sgr" . "$(7%q(B")
|
|
300 ("snr" . "$(7%r(B")
|
|
301 ("spr" . "$(7%s(B")
|
|
302 ("sbr" . "$(7%t(B")
|
|
303 ("smr" . "$(7%u(B")
|
|
304 ("kw" . "$(7$!(B")
|
|
305 ("gw" . "$(7$#(B")
|
|
306 ("cw" . "$(7$$(B")
|
|
307 ("tw" . "$(7$&(B")
|
|
308 ("dw" . "$(7$'(B")
|
|
309 ("zw" . "$(7$+(B")
|
|
310 ("rw" . "$(7$,(B")
|
|
311 ("lw" . "$(7$-(B")
|
|
312 ("sw" . "$(7$/(B")
|
|
313 ("hw" . "$(7$0(B")
|
|
314 ("ky" . "$(7$@(B")
|
|
315 ("gy" . "$(7$B(B")
|
|
316 ("py" . "$(7$C(B")
|
|
317 ("by" . "$(7$E(B")
|
|
318 ("my" . "$(7$F(B")
|
|
319 ("kr" . "$(7$P(B")
|
|
320 ("gr" . "$(7$R(B")
|
|
321 ("tr" . "$(7$S(B")
|
|
322 ("dr" . "$(7$U(B")
|
|
323 ("pr" . "$(7$V(B")
|
|
324 ("brk" . "$(7"7%!(B")
|
|
325 ("brg" . "$(7"7%"(B")
|
|
326 ("brng" . "$(7"7%#(B")
|
|
327 ("brj" . "$(7"7%$(B")
|
|
328 ("brny" . "$(7"7%%(B")
|
|
329 ("brt" . "$(7"7%&(B")
|
|
330 ("brd" . "$(7"7%'(B")
|
|
331 ("brn" . "$(7"7%((B")
|
|
332 ("brts" . "$(7"7%+(B")
|
|
333 ("brdz" . "$(7"7%,(B")
|
|
334 ("brl" . "$(7"7$d(B")
|
|
335 ("br" . "$(7$X(B")
|
|
336 ("mr" . "$(7$Y(B")
|
|
337 ("sr" . "$(7$[(B")
|
|
338 ("hr" . "$(7$\(B")
|
|
339 ("jr" . "$(7$](B")
|
|
340 ("kl" . "$(7$`(B")
|
|
341 ("gl" . "$(7$a(B")
|
|
342 ("blt" . "$(7"7%E(B")
|
|
343 ("bld" . "$(7"7%F(B")
|
|
344 ("bl" . "$(7$b(B")
|
|
345 ("zl" . "$(7$c(B")
|
|
346 ("rl" . "$(7$d(B")
|
|
347 ("sl" . "$(7$e(B")
|
|
348 ("rk" . "$(7%!(B")
|
|
349 ("rg" . "$(7%"(B")
|
|
350 ("rj" . "$(7%$(B")
|
|
351 ("rt" . "$(7%&(B")
|
|
352 ("rd" . "$(7%'(B")
|
|
353 ("rn" . "$(7%((B")
|
|
354 ("rb" . "$(7%)(B")
|
|
355 ("rm" . "$(7%*(B")
|
|
356 ("lk" . "$(7%@(B")
|
|
357 ("lg" . "$(7%A(B")
|
|
358 ("lc" . "$(7%C(B")
|
|
359 ("lj" . "$(7%D(B")
|
|
360 ("lt" . "$(7%E(B")
|
|
361 ("ld" . "$(7%F(B")
|
|
362 ("ln" . "$(7!!(B") ; dummy \x2121
|
|
363 ("lp" . "$(7%G(B")
|
|
364 ("lb" . "$(7%H(B")
|
|
365 ("lh" . "$(7%I(B")
|
|
366 ("sk" . "$(7%P(B")
|
|
367 ("sg" . "$(7%Q(B")
|
|
368 ("st" . "$(7%T(B")
|
|
369 ("sd" . "$(7%U(B")
|
|
370 ("sn" . "$(7%V(B")
|
|
371 ("sp" . "$(7%W(B")
|
|
372 ("sb" . "$(7%X(B")
|
|
373 ("sm" . "$(7%Y(B"))
|
|
374 )
|
|
375
|
|
376
|
|
377 (defconst tibetan-subjoined-transcription-alist
|
|
378 (sort '(("+k" . "$(7#!(B")
|
|
379 ("+kh" . "$(7#"(B")
|
|
380 ("+g" . "$(7##(B")
|
|
381 ("+gh" . "$(7#$(B")
|
|
382 ("+ng" . "$(7#%(B")
|
|
383 ("+c" . "$(7#&(B")
|
|
384 ("+ch" . "$(7#'(B")
|
|
385 ("+j" . "$(7#((B")
|
|
386 ("+ny" . "$(7#*(B")
|
|
387 ("+T" . "$(7#+(B")
|
|
388 ("+TH" . "$(7#,(B")
|
|
389 ("+D" . "$(7#-(B")
|
|
390 ("+DH" . "$(7#.(B")
|
|
391 ("+N" . "$(7#/(B")
|
|
392 ("+t" . "$(7#0(B")
|
|
393 ("+th" . "$(7#1(B")
|
|
394 ("+d" . "$(7#2(B")
|
|
395 ("+dh" . "$(7#3(B")
|
|
396 ("+n" . "$(7#4(B")
|
|
397 ("+p" . "$(7#5(B")
|
|
398 ("+ph" . "$(7#6(B")
|
|
399 ("+b" . "$(7#7(B")
|
|
400 ("+bh" . "$(7#8(B")
|
|
401 ("+m" . "$(7#9(B")
|
|
402 ("+ts" . "$(7#:(B")
|
|
403 ("+tsh" . "$(7#;(B")
|
|
404 ("+dz" . "$(7#<(B")
|
|
405 ("+dzh" . "$(7#=(B")
|
|
406 ("+w" . "$(7#>(B")
|
|
407 ("+zh" . "$(7#?(B")
|
|
408 ("+z" . "$(7#@(B")
|
|
409 ("+'" . "$(7#A(B")
|
|
410 ("+y" . "$(7#B(B")
|
|
411 ("+r" . "$(7#C(B")
|
|
412 ("+l" . "$(7#D(B")
|
|
413 ("+sh" . "$(7#E(B")
|
|
414 ("+SH" . "$(7#F(B")
|
|
415 ("+s" . "$(7#G(B")
|
|
416 ("+h" . "$(7#H(B")
|
|
417 ("+A" . "$(7#I(B")
|
|
418 ("+kSH" . "$(7#J(B")
|
|
419 ;; Added by Tomabechi 1999/12/10
|
|
420 ("+W" . "$(7#K(B") ;; fixed form subscribed WA
|
|
421 ("+Y" . "$(7#L(B") ;; fixed form subscribed YA
|
|
422 ("+R" . "$(7#M(B") ;; fixed form subscribed RA
|
|
423 )
|
|
424 (lambda (x y) (> (length (car x)) (length (car y))))))
|
|
425
|
|
426 ;;;
|
|
427 ;;; alist for Tibetan base consonant <-> subjoined consonant conversion.
|
|
428 ;;;
|
|
429 (defconst tibetan-base-to-subjoined-alist
|
|
430 '(("$(7"!(B" . "$(7#!(B")
|
|
431 ("$(7""(B" . "$(7#"(B")
|
|
432 ("$(7"#(B" . "$(7##(B")
|
|
433 ("$(7"$(B" . "$(7#$(B")
|
|
434 ("$(7"%(B" . "$(7#%(B")
|
|
435 ("$(7"&(B" . "$(7#&(B")
|
|
436 ("$(7"'(B" . "$(7#'(B")
|
|
437 ("$(7"((B" . "$(7#((B")
|
|
438 ("$(7"*(B" . "$(7#*(B")
|
|
439 ("$(7"+(B" . "$(7#+(B")
|
|
440 ("$(7",(B" . "$(7#,(B")
|
|
441 ("$(7"-(B" . "$(7#-(B")
|
|
442 ("$(7".(B" . "$(7#.(B")
|
|
443 ("$(7"/(B" . "$(7#/(B")
|
|
444 ("$(7"0(B" . "$(7#0(B")
|
|
445 ("$(7"1(B" . "$(7#1(B")
|
|
446 ("$(7"2(B" . "$(7#2(B")
|
|
447 ("$(7"3(B" . "$(7#3(B")
|
|
448 ("$(7"4(B" . "$(7#4(B")
|
|
449 ("$(7"5(B" . "$(7#5(B")
|
|
450 ("$(7"6(B" . "$(7#6(B")
|
|
451 ("$(7"7(B" . "$(7#7(B")
|
|
452 ("$(7"8(B" . "$(7#8(B")
|
|
453 ("$(7"9(B" . "$(7#9(B")
|
|
454 ("$(7":(B" . "$(7#:(B")
|
|
455 ("$(7";(B" . "$(7#;(B")
|
|
456 ("$(7"<(B" . "$(7#<(B")
|
|
457 ("$(7"=(B" . "$(7#=(B")
|
|
458 ("$(7">(B" . "$(7#>(B")
|
|
459 ("$(7"?(B" . "$(7#?(B")
|
|
460 ("$(7"@(B" . "$(7#@(B")
|
|
461 ("$(7"A(B" . "$(7#A(B")
|
|
462 ("$(7"B(B" . "$(7#B(B")
|
|
463 ("$(7"C(B" . "$(7#C(B")
|
|
464 ("$(7"D(B" . "$(7#D(B")
|
|
465 ("$(7"E(B" . "$(7#E(B")
|
|
466 ("$(7"F(B" . "$(7#F(B")
|
|
467 ("$(7"G(B" . "$(7#G(B")
|
|
468 ("$(7"H(B" . "$(7#H(B")
|
|
469 ("$(7"I(B" . "$(7#I(B")
|
|
470 ("$(7"J(B" . "$(7#J(B")
|
|
471 ;; Added by Tomabechi 1999/12/10
|
|
472 ("$(7"K(B" . "$(7#M(B") ;; Fixed form RA (224B->234D)
|
|
473 ))
|
|
474
|
|
475 ;;; alist for Tibetan composite vowels (long i, vocalic r, etc.)
|
|
476 ;;; New varialble. created by Tomabechi 2000/06/08
|
|
477 (defconst tibetan-composite-vowel-alist
|
|
478 '(;; LONG A
|
|
479 ;; ("$(7"R(B" . ((bc . tc) ?$(7"R(B))
|
|
480 ;; LONG I
|
|
481 ("$(7"T(B" . (?$(7"R(B (tc . bc) ?$(7"S(B))
|
|
482 ;; LONG U
|
|
483 ("$(7"V(B" . (?$(7"R(B (bc . tc) ?$(7"U(B))
|
|
484 ;; VOCALIC R
|
|
485 ("$(7"W(B" . (?$(7#C(B (tc . bc) ?$(7"a(B))
|
|
486 ;; LONG VOCALIC R
|
|
487 ("$(7"X(B" . (?$(7#C(B (bc . tc) ?$(7"R(B (tc . bc) ?$(7"a(B))
|
|
488 ;; VOCALIC L
|
|
489 ("$(7"Y(B" . (?$(7#D(B (tc . bc) ?$(7"a(B))
|
|
490 ;;$(7!;(BLONG VOCALIC L
|
|
491 ("$(7"Z(B" . (?$(7#D(B (bc . tc) ?$(7"R(B (tc . bc) ?$(7"a(B))
|
|
492 ;; LONG REVERSE I
|
|
493 ("$(7"b(B" . (?$(7"R(B (tc . bc) ?$(7"a(B))
|
|
494 ))
|
|
495
|
|
496
|
|
497
|
|
498 ;;;
|
|
499 ;;; alist for Tibetan consonantic components <-> precomposed glyph conversion.
|
|
500 ;;; (includes some punctuation conversion rules)
|
|
501 ;;;
|
|
502 (defconst tibetan-precomposition-rule-alist
|
|
503 `(("$(7"6#B#>(B" . "$(7$G(B")
|
|
504 ("$(7"##C#>(B" . "$(7$_(B")
|
|
505 ("$(7";#>(B" . "$(7$)(B")
|
|
506 ("$(7"C#:#>(B" . "$(7%.(B")
|
|
507 ("$(7"C###>(B" . "$(7%-(B")
|
|
508 ("$(7"C#!#B(B" . "$(7%0(B")
|
|
509 ("$(7"C###B(B" . "$(7%1(B")
|
|
510 ("$(7"C#9#B(B" . "$(7%2(B")
|
|
511 ("$(7"G#!#B(B" . "$(7%`(B")
|
|
512 ("$(7"G###B(B" . "$(7%a(B")
|
|
513 ("$(7"G#5#B(B" . "$(7%b(B")
|
|
514 ("$(7"G#7#B(B" . "$(7%c(B")
|
|
515 ("$(7"G#9#B(B" . "$(7%d(B")
|
|
516 ("$(7"G#!#C(B" . "$(7%p(B")
|
|
517 ("$(7"G###C(B" . "$(7%q(B")
|
|
518 ("$(7"G#4#C(B" . "$(7%r(B")
|
|
519 ("$(7"G#5#C(B" . "$(7%s(B")
|
|
520 ("$(7"G#7#C(B" . "$(7%t(B")
|
|
521 ("$(7"G#9#C(B" . "$(7%u(B")
|
|
522 ("$(7""#>(B" . "$(7$"(B")
|
|
523 ("$(7"*#>(B" . "$(7$%(B")
|
|
524 ("$(7":#>(B" . "$(7$((B")
|
|
525 ("$(7"?#>(B" . "$(7$*(B")
|
|
526 ("$(7"E#>(B" . "$(7$.(B")
|
|
527 ("$(7""#B(B" . "$(7$A(B")
|
|
528 ("$(7"6#B(B" . "$(7$D(B")
|
|
529 ("$(7""#C(B" . "$(7$Q(B")
|
|
530 ("$(7"1#C(B" . "$(7$T(B")
|
|
531 ("$(7"6#C(B" . "$(7$W(B")
|
|
532 ("$(7"E#C(B" . "$(7$Z(B")
|
|
533 ("$(7"<#C(B" . "$(7$^(B")
|
|
534 ("$(7"C#%(B" . "$(7%#(B")
|
|
535 ("$(7"C#*(B" . "$(7%%(B")
|
|
536 ("$(7"C#:(B" . "$(7%+(B")
|
|
537 ("$(7"C#<(B" . "$(7%,(B")
|
|
538 ("$(7"D#%(B" . "$(7%B(B")
|
|
539 ("$(7"G#%(B" . "$(7%R(B")
|
|
540 ("$(7"G#*(B" . "$(7%S(B")
|
|
541 ("$(7"G#:(B" . "$(7%Z(B")
|
|
542 ("$(7"!#>(B" . "$(7$!(B")
|
|
543 ("$(7"##>(B" . "$(7$#(B")
|
|
544 ("$(7"&#>(B" . "$(7$$(B")
|
|
545 ("$(7"0#>(B" . "$(7$&(B")
|
|
546 ("$(7"2#>(B" . "$(7$'(B")
|
|
547 ("$(7"@#>(B" . "$(7$+(B")
|
|
548 ("$(7"C#>(B" . "$(7$,(B")
|
|
549 ("$(7"D#>(B" . "$(7$-(B")
|
|
550 ("$(7"G#>(B" . "$(7$/(B")
|
|
551 ("$(7"H#>(B" . "$(7$0(B")
|
|
552 ("$(7"!#B(B" . "$(7$@(B")
|
|
553 ("$(7"##B(B" . "$(7$B(B")
|
|
554 ("$(7"5#B(B" . "$(7$C(B")
|
|
555 ("$(7"7#B(B" . "$(7$E(B")
|
|
556 ("$(7"9#B(B" . "$(7$F(B")
|
|
557 ("$(7"!#C(B" . "$(7$P(B")
|
|
558 ("$(7"##C(B" . "$(7$R(B")
|
|
559 ("$(7"0#C(B" . "$(7$S(B")
|
|
560 ("$(7"2#C(B" . "$(7$U(B")
|
|
561 ("$(7"5#C(B" . "$(7$V(B")
|
|
562 ("$(7"7#C(B" . "$(7$X(B")
|
|
563 ("$(7"9#C(B" . "$(7$Y(B")
|
|
564 ("$(7"G#C(B" . "$(7$[(B")
|
|
565 ("$(7"H#C(B" . "$(7$\(B")
|
|
566 ("$(7"(#C(B" . "$(7$](B")
|
|
567 ("$(7"!#D(B" . "$(7$`(B")
|
|
568 ("$(7"##D(B" . "$(7$a(B")
|
|
569 ("$(7"7#D(B" . "$(7$b(B")
|
|
570 ("$(7"@#D(B" . "$(7$c(B")
|
|
571 ("$(7"C#D(B" . "$(7$d(B")
|
|
572 ("$(7"G#D(B" . "$(7$e(B")
|
|
573 ("$(7"C#!(B" . "$(7%!(B")
|
|
574 ("$(7"C##(B" . "$(7%"(B")
|
|
575 ("$(7"C#((B" . "$(7%$(B")
|
|
576 ("$(7"C#0(B" . "$(7%&(B")
|
|
577 ("$(7"C#2(B" . "$(7%'(B")
|
|
578 ("$(7"C#4(B" . "$(7%((B")
|
|
579 ("$(7"C#7(B" . "$(7%)(B")
|
|
580 ("$(7"C#9(B" . "$(7%*(B")
|
|
581 ("$(7"D#!(B" . "$(7%@(B")
|
|
582 ("$(7"D##(B" . "$(7%A(B")
|
|
583 ("$(7"D#4(B" . "$(7!!(B") ; dummy 0x2121 added 2000/06/08 for transition l -> lng
|
|
584 ("$(7"D#&(B" . "$(7%C(B")
|
|
585 ("$(7"D#((B" . "$(7%D(B")
|
|
586 ("$(7"D#0(B" . "$(7%E(B")
|
|
587 ("$(7"D#2(B" . "$(7%F(B")
|
|
588 ("$(7"D#5(B" . "$(7%G(B")
|
|
589 ("$(7"D#7(B" . "$(7%H(B")
|
|
590 ("$(7"D#H(B" . "$(7%I(B")
|
|
591 ("$(7"G#!(B" . "$(7%P(B")
|
|
592 ("$(7"G##(B" . "$(7%Q(B")
|
|
593 ("$(7"G#0(B" . "$(7%T(B")
|
|
594 ("$(7"G#2(B" . "$(7%U(B")
|
|
595 ("$(7"G#4(B" . "$(7%V(B")
|
|
596 ("$(7"G#5(B" . "$(7%W(B")
|
|
597 ("$(7"G#7(B" . "$(7%X(B")
|
|
598 ("$(7"G#9(B" . "$(7%Y(B")))
|
|
599
|
|
600 (defconst tibetan-obsolete-glyphs
|
|
601 `(("$(7!=(B" . "$(8!=(B") ; 2 col <-> 1 col
|
|
602 ("$(7!?(B" . "$(8!?(B")
|
|
603 ("$(7!@(B" . "$(8!@(B")
|
|
604 ("$(7!A(B" . "$(8!A(B")
|
|
605 ("$(7"`(B" . "$(8"`(B")
|
|
606 ("$(7!;(B" . "$(8!;(B")
|
|
607 ("$(7!D(B" . "$(8!D(B")
|
|
608 ;; Yes these are dirty. But ...
|
|
609 ("$(7!>(B $(7!>(B" . ,(compose-string "$(7!>(B $(7!>(B" 0 3 [?$(7!>(B (Br . Bl) ? (Br . Bl) ?$(7!>(B]))
|
|
610 ("$(7!4!5!5(B" . ,(compose-string
|
|
611 "$(7#R#S#S#S(B" 0 4
|
|
612 [?$(7#R(B (Br . Bl) ?$(7#S(B (Br . Bl) ?$(7#S(B (Br . Bl) ?$(7#S(B]))
|
|
613 ("$(7!4!5(B" . ,(compose-string "$(7#R#S#S(B" 0 3 [?$(7#R(B (Br . Bl) ?$(7#S(B (Br . Bl) ?$(7#S(B]))
|
|
614 ("$(7!6(B" . ,(compose-string "$(7#R#S!I(B" 0 3 [?$(7#R(B (Br . Bl) ?$(7#S(B (br . tr) ?$(7!I(B]))
|
|
615 ("$(7!4(B" . ,(compose-string "$(7#R#S(B" 0 2 [?$(7#R(B (Br . Bl) ?$(7#S(B]))))
|
|
616
|
|
617 (defconst tibetan-regexp
|
|
618 (let ((l (list tibetan-precomposed-transcription-alist
|
|
619 tibetan-consonant-transcription-alist
|
|
620 tibetan-vowel-transcription-alist
|
|
621 tibetan-modifier-transcription-alist
|
|
622 tibetan-subjoined-transcription-alist))
|
|
623 (separator "\\|")
|
|
624 tail pattern)
|
|
625 (while l
|
|
626 (setq tail (car l) l (cdr l))
|
|
627 (while tail
|
|
628 (setq pattern (cons separator (cons (car (car tail)) pattern))
|
|
629 tail (cdr tail))))
|
|
630 (apply 'concat (nreverse (cdr pattern))))
|
|
631 "Regexp matching a Tibetan transcription of a composable Tibetan sequence.
|
|
632 The result of matching is to be used for indexing alists at conversion
|
|
633 from a roman transcription to the corresponding Tibetan character.")
|
|
634
|
|
635 (defvar tibetan-precomposed-regexp
|
|
636 (let ((l tibetan-precomposed-transcription-alist)
|
|
637 temp)
|
|
638 (setq temp "^\\(")
|
|
639 (setq temp
|
|
640 (concat temp (car (car l))))
|
|
641 (setq l (cdr l))
|
|
642 (while l
|
|
643 (setq temp
|
|
644 (concat temp "\\|" (car (car l))))
|
|
645 (setq l (cdr l)))
|
|
646 (concat temp "\\)"))
|
|
647 "Regexp string to match a romanized Tibetan complex consonant.
|
|
648 The result of matching is to be used for indexing alists when the input key
|
|
649 from an input method is converted to the corresponding precomposed glyph.")
|
|
650
|
|
651 (defvar tibetan-precomposition-rule-regexp
|
|
652 (let ((l tibetan-precomposition-rule-alist)
|
|
653 temp)
|
|
654 (setq temp "\\(")
|
|
655 (setq temp (concat temp (car (car l))))
|
|
656 (setq l (cdr l))
|
|
657 (while l
|
|
658 (setq temp (concat temp "\\|" (car (car l))))
|
|
659 (setq l (cdr l)))
|
|
660 (concat temp "\\)"))
|
|
661 "Regexp string to match a sequence of Tibetan consonantic components, i.e.,
|
|
662 one base consonant and one or more subjoined consonants.
|
|
663 The result of matching is to be used for indexing alist when the component
|
|
664 sequence is converted to the corresponding precomposed glyph.
|
|
665 This also matches some punctuation characters which need conversion.")
|
|
666
|
|
667 (defvar tibetan-decomposed nil)
|
|
668 (defvar tibetan-decomposed-temp nil)
|
|
669
|
|
670 (provide 'tibetan)
|
|
671
|
778
|
672 ;;; tibetan.el ends here
|