Trailing whitepace deleted.
[emacs.git] / lisp / language / thai-util.el
blob58588f974d774369800753dcd2f285b7701d7948
1 ;;; thai-util.el --- utilities for Thai -*- coding: iso-2022-7bit; -*-
3 ;; Copyright (C) 1995 Electrotechnical Laboratory, JAPAN.
4 ;; Licensed to the Free Software Foundation.
6 ;; Keywords: mule, multilingual, thai
8 ;; This file is part of GNU Emacs.
10 ;; GNU Emacs is free software; you can redistribute it and/or modify
11 ;; it 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.
15 ;; GNU Emacs is distributed in the hope that it will be useful,
16 ;; but WITHOUT ANY WARRANTY; without even the implied warranty of
17 ;; MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
18 ;; GNU General Public License for more details.
20 ;; You should have received a copy of the GNU General Public License
21 ;; along with GNU Emacs; see the file COPYING. If not, write to the
22 ;; Free Software Foundation, Inc., 59 Temple Place - Suite 330,
23 ;; Boston, MA 02111-1307, USA.
25 ;;; Commentary:
27 ;;; Code:
29 ;; Setting information of Thai characters.
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)
36 ;; The general composing rules are as follows:
38 ;; T
39 ;; V T V T
40 ;; CV -> C, CT -> C, CVT -> C, Cv -> C, CvT -> C
41 ;; v v
43 ;; where C: consonant, V: vowel upper, v: vowel lower, T: tone mark.
45 (defvar thai-composition-pattern "\\cc\\(\\ct\\|\\cv\\ct?\\)"
46 "Regular expression matching a Thai composite sequence.")
48 (let ((l '((?\e,T!\e(B consonant "LETTER KO KAI") ; 0xA1
49 (?\e,T"\e(B consonant "LETTER KHO KHAI") ; 0xA2
50 (?\e,T#\e(B consonant "LETTER KHO KHUAT") ; 0xA3
51 (?\e,T$\e(B consonant "LETTER KHO KHWAI") ; 0xA4
52 (?\e,T%\e(B consonant "LETTER KHO KHON") ; 0xA5
53 (?\e,T&\e(B consonant "LETTER KHO RAKHANG") ; 0xA6
54 (?\e,T'\e(B consonant "LETTER NGO NGU") ; 0xA7
55 (?\e,T(\e(B consonant "LETTER CHO CHAN") ; 0xA8
56 (?\e,T)\e(B consonant "LETTER CHO CHING") ; 0xA9
57 (?\e,T*\e(B consonant "LETTER CHO CHANG") ; 0xAA
58 (?\e,T+\e(B consonant "LETTER SO SO") ; 0xAB
59 (?\e,T,\e(B consonant "LETTER CHO CHOE") ; 0xAC
60 (?\e,T-\e(B consonant "LETTER YO YING") ; 0xAD
61 (?\e,T.\e(B consonant "LETTER DO CHADA") ; 0xAE
62 (?\e,T/\e(B consonant "LETTER TO PATAK") ; 0xAF
63 (?\e,T0\e(B consonant "LETTER THO THAN") ; 0xB0
64 (?\e,T1\e(B consonant "LETTER THO NANGMONTHO") ; 0xB1
65 (?\e,T2\e(B consonant "LETTER THO PHUTHAO") ; 0xB2
66 (?\e,T3\e(B consonant "LETTER NO NEN") ; 0xB3
67 (?\e,T4\e(B consonant "LETTER DO DEK") ; 0xB4
68 (?\e,T5\e(B consonant "LETTER TO TAO") ; 0xB5
69 (?\e,T6\e(B consonant "LETTER THO THUNG") ; 0xB6
70 (?\e,T7\e(B consonant "LETTER THO THAHAN") ; 0xB7
71 (?\e,T8\e(B consonant "LETTER THO THONG") ; 0xB8
72 (?\e,T9\e(B consonant "LETTER NO NU") ; 0xB9
73 (?\e,T:\e(B consonant "LETTER BO BAIMAI") ; 0xBA
74 (?\e,T;\e(B consonant "LETTER PO PLA") ; 0xBB
75 (?\e,T<\e(B consonant "LETTER PHO PHUNG") ; 0xBC
76 (?\e,T=\e(B consonant "LETTER FO FA") ; 0xBD
77 (?\e,T>\e(B consonant "LETTER PHO PHAN") ; 0xBE
78 (?\e,T?\e(B consonant "LETTER FO FAN") ; 0xBF
79 (?\e,T@\e(B consonant "LETTER PHO SAMPHAO") ; 0xC0
80 (?\e,TA\e(B consonant "LETTER MO MA") ; 0xC1
81 (?\e,TB\e(B consonant "LETTER YO YAK") ; 0xC2
82 (?\e,TC\e(B consonant "LETTER RO RUA") ; 0xC3
83 (?\e,TD\e(B vowel-base "LETTER RU (Pali vowel letter)") ; 0xC4
84 (?\e,TE\e(B consonant "LETTER LO LING") ; 0xC5
85 (?\e,TF\e(B vowel-base "LETTER LU (Pali vowel letter)") ; 0xC6
86 (?\e,TG\e(B consonant "LETTER WO WAEN") ; 0xC7
87 (?\e,TH\e(B consonant "LETTER SO SALA") ; 0xC8
88 (?\e,TI\e(B consonant "LETTER SO RUSI") ; 0xC9
89 (?\e,TJ\e(B consonant "LETTER SO SUA") ; 0xCA
90 (?\e,TK\e(B consonant "LETTER HO HIP") ; 0xCB
91 (?\e,TL\e(B consonant "LETTER LO CHULA") ; 0xCC
92 (?\e,TM\e(B consonant "LETTER O ANG") ; 0xCD
93 (?\e,TN\e(B consonant "LETTER HO NOK HUK") ; 0xCE
94 (?\e,TO\e(B special "PAI YAN NOI (abbreviation)") ; 0xCF
95 (?\e,TP\e(B vowel-base "VOWEL SIGN SARA A") ; 0xD0
96 (?\e,TQ\e(B vowel-upper "VOWEL SIGN MAI HAN-AKAT N/S-T") ; 0xD1
97 (?\e,TR\e(B vowel-base "VOWEL SIGN SARA AA") ; 0xD2
98 (?\e,TS\e(B vowel-base "VOWEL SIGN SARA AM") ; 0xD3
99 (?\e,TT\e(B vowel-upper "VOWEL SIGN SARA I N/S-T") ; 0xD4
100 (?\e,TU\e(B vowel-upper "VOWEL SIGN SARA II N/S-T") ; 0xD5
101 (?\e,TV\e(B vowel-upper "VOWEL SIGN SARA UE N/S-T") ; 0xD6
102 (?\e,TW\e(B vowel-upper "VOWEL SIGN SARA UEE N/S-T") ; 0xD7
103 (?\e,TX\e(B vowel-lower "VOWEL SIGN SARA U N/S-B") ; 0xD8
104 (?\e,TY\e(B vowel-lower "VOWEL SIGN SARA UU N/S-B") ; 0xD9
105 (?\e,TZ\e(B vowel-lower "VOWEL SIGN PHINTHU N/S-B (Pali virama)") ; 0xDA
106 (?\e,T[\e(B invalid nil) ; 0xDA
107 (?\e,T\\e(B invalid nil) ; 0xDC
108 (?\e,T]\e(B invalid nil) ; 0xDC
109 (?\e,T^\e(B invalid nil) ; 0xDC
110 (?\e,T_\e(B special "BAHT SIGN (currency symbol)") ; 0xDF
111 (?\e,T`\e(B vowel-base "VOWEL SIGN SARA E") ; 0xE0
112 (?\e,Ta\e(B vowel-base "VOWEL SIGN SARA AE") ; 0xE1
113 (?\e,Tb\e(B vowel-base "VOWEL SIGN SARA O") ; 0xE2
114 (?\e,Tc\e(B vowel-base "VOWEL SIGN SARA MAI MUAN") ; 0xE3
115 (?\e,Td\e(B vowel-base "VOWEL SIGN SARA MAI MALAI") ; 0xE4
116 (?\e,Te\e(B vowel-base "LAK KHANG YAO") ; 0xE5
117 (?\e,Tf\e(B special "MAI YAMOK (repetion)") ; 0xE6
118 (?\e,Tg\e(B vowel-upper "VOWEL SIGN MAI TAI KHU N/S-T") ; 0xE7
119 (?\e,Th\e(B tone "TONE MAI EK N/S-T") ; 0xE8
120 (?\e,Ti\e(B tone "TONE MAI THO N/S-T") ; 0xE9
121 (?\e,Tj\e(B tone "TONE MAI TRI N/S-T") ; 0xEA
122 (?\e,Tk\e(B tone "TONE MAI CHATTAWA N/S-T") ; 0xEB
123 (?\e,Tl\e(B tone "THANTHAKHAT N/S-T (cancellation mark)") ; 0xEC
124 (?\e,Tm\e(B tone "NIKKHAHIT N/S-T (final nasal)") ; 0xED
125 (?\e,Tn\e(B vowel-upper "YAMAKKAN N/S-T") ; 0xEE
126 (?\e,To\e(B special "FONRMAN") ; 0xEF
127 (?\e,Tp\e(B special "DIGIT ZERO") ; 0xF0
128 (?\e,Tq\e(B special "DIGIT ONE") ; 0xF1
129 (?\e,Tr\e(B special "DIGIT TWO") ; 0xF2
130 (?\e,Ts\e(B special "DIGIT THREE") ; 0xF3
131 (?\e,Tt\e(B special "DIGIT FOUR") ; 0xF4
132 (?\e,Tu\e(B special "DIGIT FIVE") ; 0xF5
133 (?\e,Tv\e(B special "DIGIT SIX") ; 0xF6
134 (?\e,Tw\e(B special "DIGIT SEVEN") ; 0xF7
135 (?\e,Tx\e(B special "DIGIT EIGHT") ; 0xF8
136 (?\e,Ty\e(B special "DIGIT NINE") ; 0xF9
137 (?\e,Tz\e(B special "ANGKHANKHU (ellipsis)") ; 0xFA
138 (?\e,T{\e(B special "KHOMUT (beginning of religious texts)") ; 0xFB
139 (?\e,T|\e(B invalid nil) ; 0xFC
140 (?\e,T}\e(B invalid nil) ; 0xFD
141 (?\e,T~\e(B invalid nil) ; 0xFE
143 ;; Unicode equivalents
144 (?\e$,1Ba\e(B consonant "LETTER KO KAI")
145 (?\e$,1Bb\e(B consonant "LETTER KHO KHAI")
146 (?\e$,1Bc\e(B consonant "LETTER KHO KHUAT")
147 (?\e$,1Bd\e(B consonant "LETTER KHO KHWAI")
148 (?\e$,1Be\e(B consonant "LETTER KHO KHON")
149 (?\e$,1Bf\e(B consonant "LETTER KHO RAKHANG")
150 (?\e$,1Bg\e(B consonant "LETTER NGO NGU")
151 (?\e$,1Bh\e(B consonant "LETTER CHO CHAN")
152 (?\e$,1Bi\e(B consonant "LETTER CHO CHING")
153 (?\e$,1Bj\e(B consonant "LETTER CHO CHANG")
154 (?\e$,1Bk\e(B consonant "LETTER SO SO")
155 (?\e$,1Bl\e(B consonant "LETTER CHO CHOE")
156 (?\e$,1Bm\e(B consonant "LETTER YO YING")
157 (?\e$,1Bn\e(B consonant "LETTER DO CHADA")
158 (?\e$,1Bo\e(B consonant "LETTER TO PATAK")
159 (?\e$,1Bp\e(B consonant "LETTER THO THAN")
160 (?\e$,1Bq\e(B consonant "LETTER THO NANGMONTHO")
161 (?\e$,1Br\e(B consonant "LETTER THO PHUTHAO")
162 (?\e$,1Bs\e(B consonant "LETTER NO NEN")
163 (?\e$,1Bt\e(B consonant "LETTER DO DEK")
164 (?\e$,1Bu\e(B consonant "LETTER TO TAO")
165 (?\e$,1Bv\e(B consonant "LETTER THO THUNG")
166 (?\e$,1Bw\e(B consonant "LETTER THO THAHAN")
167 (?\e$,1Bx\e(B consonant "LETTER THO THONG")
168 (?\e$,1By\e(B consonant "LETTER NO NU")
169 (?\e$,1Bz\e(B consonant "LETTER BO BAIMAI")
170 (?\e$,1B{\e(B consonant "LETTER PO PLA")
171 (?\e$,1B|\e(B consonant "LETTER PHO PHUNG")
172 (?\e$,1B}\e(B consonant "LETTER FO FA")
173 (?\e$,1B~\e(B consonant "LETTER PHO PHAN")
174 (?\e$,1B\x7f\e(B consonant "LETTER FO FAN")
175 (?\e$,1C \e(B consonant "LETTER PHO SAMPHAO")
176 (?\e$,1C!\e(B consonant "LETTER MO MA")
177 (?\e$,1C"\e(B consonant "LETTER YO YAK")
178 (?\e$,1C#\e(B consonant "LETTER RO RUA")
179 (?\e$,1C$\e(B vowel-base "LETTER RU (Pali vowel letter)")
180 (?\e$,1C%\e(B consonant "LETTER LO LING")
181 (?\e$,1C&\e(B vowel-base "LETTER LU (Pali vowel letter)")
182 (?\e$,1C'\e(B consonant "LETTER WO WAEN")
183 (?\e$,1C(\e(B consonant "LETTER SO SALA")
184 (?\e$,1C)\e(B consonant "LETTER SO RUSI")
185 (?\e$,1C*\e(B consonant "LETTER SO SUA")
186 (?\e$,1C+\e(B consonant "LETTER HO HIP")
187 (?\e$,1C,\e(B consonant "LETTER LO CHULA")
188 (?\e$,1C-\e(B consonant "LETTER O ANG")
189 (?\e$,1C.\e(B consonant "LETTER HO NOK HUK")
190 (?\e$,1C/\e(B special "PAI YAN NOI (abbreviation)")
191 (?\e$,1C0\e(B vowel-base "VOWEL SIGN SARA A")
192 (?\e$,1C1\e(B vowel-upper "VOWEL SIGN MAI HAN-AKAT N/S-T")
193 (?\e$,1C2\e(B vowel-base "VOWEL SIGN SARA AA")
194 (?\e$,1C3\e(B vowel-base "VOWEL SIGN SARA AM")
195 (?\e$,1C4\e(B vowel-upper "VOWEL SIGN SARA I N/S-T")
196 (?\e$,1C5\e(B vowel-upper "VOWEL SIGN SARA II N/S-T")
197 (?\e$,1C6\e(B vowel-upper "VOWEL SIGN SARA UE N/S-T")
198 (?\e$,1C7\e(B vowel-upper "VOWEL SIGN SARA UEE N/S-T")
199 (?\e$,1C8\e(B vowel-lower "VOWEL SIGN SARA U N/S-B")
200 (?\e$,1C9\e(B vowel-lower "VOWEL SIGN SARA UU N/S-B")
201 (?\e$,1C:\e(B vowel-lower "VOWEL SIGN PHINTHU N/S-B (Pali virama)")
202 (?\e$,1C?\e(B special "BAHT SIGN (currency symbol)")
203 (?\e$,1C@\e(B vowel-base "VOWEL SIGN SARA E")
204 (?\e$,1CA\e(B vowel-base "VOWEL SIGN SARA AE")
205 (?\e$,1CB\e(B vowel-base "VOWEL SIGN SARA O")
206 (?\e$,1CC\e(B vowel-base "VOWEL SIGN SARA MAI MUAN")
207 (?\e$,1CD\e(B vowel-base "VOWEL SIGN SARA MAI MALAI")
208 (?\e$,1CE\e(B vowel-base "LAK KHANG YAO")
209 (?\e$,1CF\e(B special "MAI YAMOK (repetion)")
210 (?\e$,1CG\e(B vowel-upper "VOWEL SIGN MAI TAI KHU N/S-T")
211 (?\e$,1CH\e(B tone "TONE MAI EK N/S-T")
212 (?\e$,1CI\e(B tone "TONE MAI THO N/S-T")
213 (?\e$,1CJ\e(B tone "TONE MAI TRI N/S-T")
214 (?\e$,1CK\e(B tone "TONE MAI CHATTAWA N/S-T")
215 (?\e$,1CL\e(B tone "THANTHAKHAT N/S-T (cancellation mark)")
216 (?\e$,1CM\e(B tone "NIKKHAHIT N/S-T (final nasal)")
217 (?\e$,1CN\e(B vowel-upper "YAMAKKAN N/S-T")
218 (?\e$,1CO\e(B special "FONRMAN")
219 (?\e$,1CP\e(B special "DIGIT ZERO")
220 (?\e$,1CQ\e(B special "DIGIT ONE")
221 (?\e$,1CR\e(B special "DIGIT TWO")
222 (?\e$,1CS\e(B special "DIGIT THREE")
223 (?\e$,1CT\e(B special "DIGIT FOUR")
224 (?\e$,1CU\e(B special "DIGIT FIVE")
225 (?\e$,1CV\e(B special "DIGIT SIX")
226 (?\e$,1CW\e(B special "DIGIT SEVEN")
227 (?\e$,1CX\e(B special "DIGIT EIGHT")
228 (?\e$,1CY\e(B special "DIGIT NINE")
229 (?\e$,1CZ\e(B special "ANGKHANKHU (ellipsis)")
230 (?\e$,1C[\e(B special "KHOMUT (beginning of religious texts)")
232 elm)
233 (while l
234 (setq elm (car l) l (cdr l))
235 (let ((char (car elm))
236 (ptype (nth 1 elm)))
237 (put-char-code-property char 'phonetic-type ptype)
238 (cond ((eq ptype 'consonant)
239 (modify-category-entry char ?c thai-category-table))
240 ((memq ptype '(vowel-upper vowel-lower))
241 (modify-category-entry char ?v thai-category-table))
242 ((eq ptype 'tone)
243 (modify-category-entry char ?t thai-category-table)))
244 (put-char-code-property char 'name (nth 2 elm)))))
246 ;;;###autoload
247 (defun thai-compose-region (beg end)
248 "Compose Thai characters in the region.
249 When called from a program, expects two arguments,
250 positions (integers or markers) specifying the region."
251 (interactive "r")
252 (save-restriction
253 (narrow-to-region beg end)
254 (goto-char (point-min))
255 (with-category-table thai-category-table
256 (while (re-search-forward thai-composition-pattern nil t)
257 (compose-region (match-beginning 0) (match-end 0))))))
259 ;;;###autoload
260 (defun thai-compose-string (string)
261 "Compose Thai characters in STRING and return the resulting string."
262 (with-category-table thai-category-table
263 (let ((idx 0))
264 (while (setq idx (string-match thai-composition-pattern string idx))
265 (compose-string string idx (match-end 0))
266 (setq idx (match-end 0)))))
267 string)
269 ;;;###autoload
270 (defun thai-compose-buffer ()
271 "Compose Thai characters in the current buffer."
272 (interactive)
273 (thai-compose-region (point-min) (point-max)))
275 ;;;###autoload
276 (defun thai-post-read-conversion (len)
277 (thai-compose-region (point) (+ (point) len))
278 len)
280 ;;;###autoload
281 (defun thai-composition-function (from to pattern &optional string)
282 "Compose Thai text in the region FROM and TO.
283 The text matches the regular expression PATTERN.
284 Optional 4th argument STRING, if non-nil, is a string containing text
285 to compose.
287 The return value is number of composed characters."
288 (if (< (1+ from) to)
289 (progn
290 (if string
291 (compose-string string from to)
292 (compose-region from to))
293 (- to from))))
296 (provide 'thai-util)
298 ;;; thai-util.el ends here