771
|
1 ;;; thai-util.el --- utilities for Thai -*- coding: iso-2022-7bit; -*-
|
|
2
|
|
3 ;; Copyright (C) 1995 Electrotechnical Laboratory, JAPAN.
|
|
4 ;; Licensed to the Free Software Foundation.
|
|
5
|
|
6 ;; Keywords: mule, multilingual, thai
|
|
7
|
|
8 ;; This file is part of XEmacs.
|
|
9
|
|
10 ;; XEmacs is free software; you can redistribute it and/or modify it
|
|
11 ;; under the terms of the GNU General Public License as published by
|
|
12 ;; the Free Software Foundation; either version 2, or (at your option)
|
|
13 ;; any later version.
|
|
14
|
|
15 ;; XEmacs is distributed in the hope that it will be useful, but
|
|
16 ;; WITHOUT ANY WARRANTY; without even the implied warranty of
|
|
17 ;; MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
|
|
18 ;; General Public License for more details.
|
|
19
|
|
20 ;; You should have received a copy of the GNU General Public License
|
|
21 ;; along with XEmacs; see the file COPYING. If not, write to the Free
|
|
22 ;; Software Foundation, Inc., 59 Temple Place - Suite 330, Boston, MA
|
|
23 ;; 02111-1307, USA.
|
|
24
|
|
25 ;;; Synched up with: Emacs 21.0.103 (language/thai-util.el).
|
|
26
|
|
27 ;;; Code:
|
|
28
|
|
29 ;; Setting information of Thai characters.
|
|
30
|
|
31 (defconst thai-category-table (make-category-table))
|
|
32 (define-category ?c "Thai consonant" thai-category-table)
|
|
33 (define-category ?v "Thai upper/lower vowel" thai-category-table)
|
|
34 (define-category ?t "Thai tone" thai-category-table)
|
|
35
|
|
36 ;; The general composing rules are as follows:
|
|
37 ;;
|
|
38 ;; T
|
|
39 ;; V T V T
|
|
40 ;; CV -> C, CT -> C, CVT -> C, Cv -> C, CvT -> C
|
|
41 ;; v v
|
|
42 ;;
|
|
43 ;; where C: consonant, V: vowel upper, v: vowel lower, T: tone mark.
|
|
44
|
|
45 (defvar thai-composition-pattern "\\cc\\(\\ct\\|\\cv\\ct?\\)"
|
|
46 "Regular expression matching a Thai composite sequence.")
|
|
47
|
|
48 (let ((l '((?,T!(B consonant "LETTER KO KAI") ; 0xA1
|
|
49 (?,T"(B consonant "LETTER KHO KHAI") ; 0xA2
|
|
50 (?,T#(B consonant "LETTER KHO KHUAT") ; 0xA3
|
|
51 (?,T$(B consonant "LETTER KHO KHWAI") ; 0xA4
|
|
52 (?,T%(B consonant "LETTER KHO KHON") ; 0xA5
|
|
53 (?,T&(B consonant "LETTER KHO RAKHANG") ; 0xA6
|
|
54 (?,T'(B consonant "LETTER NGO NGU") ; 0xA7
|
|
55 (?,T((B consonant "LETTER CHO CHAN") ; 0xA8
|
|
56 (?,T)(B consonant "LETTER CHO CHING") ; 0xA9
|
|
57 (?,T*(B consonant "LETTER CHO CHANG") ; 0xAA
|
|
58 (?,T+(B consonant "LETTER SO SO") ; 0xAB
|
|
59 (?,T,(B consonant "LETTER CHO CHOE") ; 0xAC
|
|
60 (?,T-(B consonant "LETTER YO YING") ; 0xAD
|
|
61 (?,T.(B consonant "LETTER DO CHADA") ; 0xAE
|
|
62 (?,T/(B consonant "LETTER TO PATAK") ; 0xAF
|
|
63 (?,T0(B consonant "LETTER THO THAN") ; 0xB0
|
|
64 (?,T1(B consonant "LETTER THO NANGMONTHO") ; 0xB1
|
|
65 (?,T2(B consonant "LETTER THO PHUTHAO") ; 0xB2
|
|
66 (?,T3(B consonant "LETTER NO NEN") ; 0xB3
|
|
67 (?,T4(B consonant "LETTER DO DEK") ; 0xB4
|
|
68 (?,T5(B consonant "LETTER TO TAO") ; 0xB5
|
|
69 (?,T6(B consonant "LETTER THO THUNG") ; 0xB6
|
|
70 (?,T7(B consonant "LETTER THO THAHAN") ; 0xB7
|
|
71 (?,T8(B consonant "LETTER THO THONG") ; 0xB8
|
|
72 (?,T9(B consonant "LETTER NO NU") ; 0xB9
|
|
73 (?,T:(B consonant "LETTER BO BAIMAI") ; 0xBA
|
|
74 (?,T;(B consonant "LETTER PO PLA") ; 0xBB
|
|
75 (?,T<(B consonant "LETTER PHO PHUNG") ; 0xBC
|
|
76 (?,T=(B consonant "LETTER FO FA") ; 0xBD
|
|
77 (?,T>(B consonant "LETTER PHO PHAN") ; 0xBE
|
|
78 (?,T?(B consonant "LETTER FO FAN") ; 0xBF
|
|
79 (?,T@(B consonant "LETTER PHO SAMPHAO") ; 0xC0
|
|
80 (?,TA(B consonant "LETTER MO MA") ; 0xC1
|
|
81 (?,TB(B consonant "LETTER YO YAK") ; 0xC2
|
|
82 (?,TC(B consonant "LETTER RO RUA") ; 0xC3
|
|
83 (?,TD(B vowel-base "LETTER RU (Pali vowel letter)") ; 0xC4
|
|
84 (?,TE(B consonant "LETTER LO LING") ; 0xC5
|
|
85 (?,TF(B vowel-base "LETTER LU (Pali vowel letter)") ; 0xC6
|
|
86 (?,TG(B consonant "LETTER WO WAEN") ; 0xC7
|
|
87 (?,TH(B consonant "LETTER SO SALA") ; 0xC8
|
|
88 (?,TI(B consonant "LETTER SO RUSI") ; 0xC9
|
|
89 (?,TJ(B consonant "LETTER SO SUA") ; 0xCA
|
|
90 (?,TK(B consonant "LETTER HO HIP") ; 0xCB
|
|
91 (?,TL(B consonant "LETTER LO CHULA") ; 0xCC
|
|
92 (?,TM(B consonant "LETTER O ANG") ; 0xCD
|
|
93 (?,TN(B consonant "LETTER HO NOK HUK") ; 0xCE
|
|
94 (?,TO(B special "PAI YAN NOI (abbreviation)") ; 0xCF
|
|
95 (?,TP(B vowel-base "VOWEL SIGN SARA A") ; 0xD0
|
|
96 (?,TQ(B vowel-upper "VOWEL SIGN MAI HAN-AKAT N/S-T") ; 0xD1
|
|
97 (?,TR(B vowel-base "VOWEL SIGN SARA AA") ; 0xD2
|
|
98 (?,TS(B vowel-base "VOWEL SIGN SARA AM") ; 0xD3
|
|
99 (?,TT(B vowel-upper "VOWEL SIGN SARA I N/S-T") ; 0xD4
|
|
100 (?,TU(B vowel-upper "VOWEL SIGN SARA II N/S-T") ; 0xD5
|
|
101 (?,TV(B vowel-upper "VOWEL SIGN SARA UE N/S-T") ; 0xD6
|
|
102 (?,TW(B vowel-upper "VOWEL SIGN SARA UEE N/S-T") ; 0xD7
|
|
103 (?,TX(B vowel-lower "VOWEL SIGN SARA U N/S-B") ; 0xD8
|
|
104 (?,TY(B vowel-lower "VOWEL SIGN SARA UU N/S-B") ; 0xD9
|
|
105 (?,TZ(B vowel-lower "VOWEL SIGN PHINTHU N/S-B (Pali virama)") ; 0xDA
|
|
106 (?,T[(B invalid nil) ; 0xDA
|
|
107 (?,T\(B invalid nil) ; 0xDC
|
|
108 (?,T](B invalid nil) ; 0xDC
|
|
109 (?,T^(B invalid nil) ; 0xDC
|
|
110 (?,T_(B special "BAHT SIGN (currency symbol)") ; 0xDF
|
|
111 (?,T`(B vowel-base "VOWEL SIGN SARA E") ; 0xE0
|
|
112 (?,Ta(B vowel-base "VOWEL SIGN SARA AE") ; 0xE1
|
|
113 (?,Tb(B vowel-base "VOWEL SIGN SARA O") ; 0xE2
|
|
114 (?,Tc(B vowel-base "VOWEL SIGN SARA MAI MUAN") ; 0xE3
|
|
115 (?,Td(B vowel-base "VOWEL SIGN SARA MAI MALAI") ; 0xE4
|
|
116 (?,Te(B vowel-base "LAK KHANG YAO") ; 0xE5
|
|
117 (?,Tf(B special "MAI YAMOK (repetion)") ; 0xE6
|
|
118 (?,Tg(B vowel-upper "VOWEL SIGN MAI TAI KHU N/S-T") ; 0xE7
|
|
119 (?,Th(B tone "TONE MAI EK N/S-T") ; 0xE8
|
|
120 (?,Ti(B tone "TONE MAI THO N/S-T") ; 0xE9
|
|
121 (?,Tj(B tone "TONE MAI TRI N/S-T") ; 0xEA
|
|
122 (?,Tk(B tone "TONE MAI CHATTAWA N/S-T") ; 0xEB
|
|
123 (?,Tl(B tone "THANTHAKHAT N/S-T (cancellation mark)") ; 0xEC
|
|
124 (?,Tm(B tone "NIKKHAHIT N/S-T (final nasal)") ; 0xED
|
|
125 (?,Tn(B vowel-upper "YAMAKKAN N/S-T") ; 0xEE
|
|
126 (?,To(B special "FONRMAN") ; 0xEF
|
|
127 (?,Tp(B special "DIGIT ZERO") ; 0xF0
|
|
128 (?,Tq(B special "DIGIT ONE") ; 0xF1
|
|
129 (?,Tr(B special "DIGIT TWO") ; 0xF2
|
|
130 (?,Ts(B special "DIGIT THREE") ; 0xF3
|
|
131 (?,Tt(B special "DIGIT FOUR") ; 0xF4
|
|
132 (?,Tu(B special "DIGIT FIVE") ; 0xF5
|
|
133 (?,Tv(B special "DIGIT SIX") ; 0xF6
|
|
134 (?,Tw(B special "DIGIT SEVEN") ; 0xF7
|
|
135 (?,Tx(B special "DIGIT EIGHT") ; 0xF8
|
|
136 (?,Ty(B special "DIGIT NINE") ; 0xF9
|
|
137 (?,Tz(B special "ANGKHANKHU (ellipsis)") ; 0xFA
|
|
138 (?,T{(B special "KHOMUT (beginning of religious texts)") ; 0xFB
|
|
139 (?,T|(B invalid nil) ; 0xFC
|
|
140 (?,T}(B invalid nil) ; 0xFD
|
|
141 (?,T~(B invalid nil) ; 0xFE
|
|
142 ))
|
|
143 elm)
|
|
144 (while l
|
|
145 (setq elm (car l) l (cdr l))
|
|
146 (let ((char (car elm))
|
|
147 (ptype (nth 1 elm)))
|
|
148 (put-char-code-property char 'phonetic-type ptype)
|
|
149 (cond ((eq ptype 'consonant)
|
|
150 (modify-category-entry char ?c thai-category-table))
|
|
151 ((memq ptype '(vowel-upper vowel-lower))
|
|
152 (modify-category-entry char ?v thai-category-table))
|
|
153 ((eq ptype 'tone)
|
|
154 (modify-category-entry char ?t thai-category-table)))
|
|
155 (put-char-code-property char 'name (nth 2 elm)))))
|
|
156
|
|
157 ;;;###autoload
|
|
158 (defun thai-compose-region (beg end)
|
|
159 "Compose Thai characters in the region.
|
|
160 When called from a program, expects two arguments,
|
|
161 positions (integers or markers) specifying the region."
|
|
162 (interactive "r")
|
|
163 (save-restriction
|
|
164 (narrow-to-region beg end)
|
|
165 (goto-char (point-min))
|
|
166 (with-category-table thai-category-table
|
|
167 (while (re-search-forward thai-composition-pattern nil t)
|
|
168 (compose-region (match-beginning 0) (match-end 0))))))
|
|
169
|
|
170 ;;;###autoload
|
|
171 (defun thai-compose-string (string)
|
|
172 "Compose Thai characters in STRING and return the resulting string."
|
|
173 (with-category-table thai-category-table
|
|
174 (let ((idx 0))
|
|
175 (while (setq idx (string-match thai-composition-pattern string idx))
|
|
176 (compose-string string idx (match-end 0))
|
|
177 (setq idx (match-end 0)))))
|
|
178 string)
|
|
179
|
|
180 ;;;###autoload
|
|
181 (defun thai-compose-buffer ()
|
|
182 "Compose Thai characters in the current buffer."
|
|
183 (interactive)
|
|
184 (thai-compose-region (point-min) (point-max)))
|
|
185
|
|
186 ;;;###autoload
|
|
187 (defun thai-post-read-conversion (len)
|
|
188 (thai-compose-region (point) (+ (point) len))
|
|
189 len)
|
|
190
|
|
191 ;;;###autoload
|
|
192 (defun thai-composition-function (from to pattern &optional string)
|
|
193 "Compose Thai text in the region FROM and TO.
|
|
194 The text matches the regular expression PATTERN.
|
|
195 Optional 4th argument STRING, if non-nil, is a string containing text
|
|
196 to compose.
|
|
197
|
|
198 The return value is number of composed characters."
|
|
199 (if (< (1+ from) to)
|
|
200 (prog1 (- to from)
|
|
201 (if string
|
|
202 (compose-string string from to)
|
|
203 (compose-region from to))
|
|
204 (- to from))))
|
|
205
|
|
206 ;;
|
|
207 (provide 'thai-util)
|
|
208
|
|
209 ;;; thai-util.el ends here
|