Commit | Line | Data |
---|---|---|
64fd2bb1 | 1 | ;;; lao-util.el --- utilities for Lao -*- coding: iso-2022-7bit; -*- |
d92d0128 | 2 | |
acaf905b | 3 | ;; Copyright (C) 2001-2012 Free Software Foundation, Inc. |
38141d20 | 4 | ;; Copyright (C) 1997, 1998, 1999, 2000, 2001, 2002, 2003, 2004, 2005, 2006, |
5df4f04c | 5 | ;; 2007, 2008, 2009, 2010, 2011 |
eaa61218 KH |
6 | ;; National Institute of Advanced Industrial Science and Technology (AIST) |
7 | ;; Registration Number H14PRO021 | |
8f924df7 KH |
8 | ;; Copyright (C) 2003 |
9 | ;; National Institute of Advanced Industrial Science and Technology (AIST) | |
10 | ;; Registration Number H13PRO009 | |
d92d0128 | 11 | |
ffdc0bca | 12 | ;; Keywords: multilingual, Lao, i18n |
d92d0128 KH |
13 | |
14 | ;; This file is part of GNU Emacs. | |
15 | ||
4936186e | 16 | ;; GNU Emacs is free software: you can redistribute it and/or modify |
d92d0128 | 17 | ;; it under the terms of the GNU General Public License as published by |
4936186e GM |
18 | ;; the Free Software Foundation, either version 3 of the License, or |
19 | ;; (at your option) any later version. | |
d92d0128 KH |
20 | |
21 | ;; GNU Emacs is distributed in the hope that it will be useful, | |
22 | ;; but WITHOUT ANY WARRANTY; without even the implied warranty of | |
23 | ;; MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the | |
24 | ;; GNU General Public License for more details. | |
25 | ||
26 | ;; You should have received a copy of the GNU General Public License | |
4936186e | 27 | ;; along with GNU Emacs. If not, see <http://www.gnu.org/licenses/>. |
d92d0128 | 28 | |
60370d40 PJ |
29 | ;;; Commentary: |
30 | ||
d92d0128 KH |
31 | ;;; Code: |
32 | ||
31ce7719 KH |
33 | ;; Setting information of Thai characters. |
34 | ||
35 | (defconst lao-category-table (make-category-table)) | |
36 | (define-category ?c "Lao consonant" lao-category-table) | |
37 | (define-category ?s "Lao semi-vowel" lao-category-table) | |
38 | (define-category ?v "Lao upper/lower vowel" lao-category-table) | |
39 | (define-category ?t "Lao tone" lao-category-table) | |
40 | ||
d92d0128 KH |
41 | (let ((l '((?\e(1!\e(B consonant "LETTER KOR KAI'" "CHICKEN") |
42 | (?\e(1"\e(B consonant "LETTER KHOR KHAI'" "EGG") | |
43 | (?\e(1#\e(B invalid nil) | |
44 | (?\e(1$\e(B consonant "LETTER QHOR QHWARGN" "BUFFALO") | |
45 | (?\e(1%\e(B invalid nil) | |
46 | (? invalid nil) | |
47 | (?\e(1'\e(B consonant "LETTER NGOR NGUU" "SNAKE") | |
48 | (?\e(1(\e(B consonant "LETTER JOR JUA" "BUDDHIST NOVICE") | |
49 | (?\e(1)\e(B invalid nil) | |
50 | (?\e(1*\e(B consonant "LETTER XOR X\"ARNG" "ELEPHANT") | |
51 | (?\e(1+\e(B invalid nil) | |
52 | (?\e(1,\e(B invalid nil) | |
53 | (?\e(1-\e(B consonant "LETTER YOR YUNG" "MOSQUITO") | |
54 | (?\e(1.\e(B invalid nil) | |
55 | (?\e(1.\e(B invalid nil) | |
56 | (?\e(1.\e(B invalid nil) | |
57 | (?\e(1.\e(B invalid nil) | |
58 | (?\e(1.\e(B invalid nil) | |
59 | (?\e(1.\e(B invalid nil) | |
60 | (?\e(14\e(B consonant "LETTER DOR DANG" "NOSE") | |
61 | (?\e(15\e(B consonant "LETTER TOR TAR" "EYE") | |
3518ac16 KH |
62 | (?\e(16\e(B consonant "LETTER THOR THUNG" "TO ASK,QUESTION") |
63 | (?\e(17\e(B consonant "LETTER DHOR DHARM" "FLAG") | |
d92d0128 KH |
64 | (?\e(18\e(B invalid nil) |
65 | (?\e(19\e(B consonant "LETTER NOR NOK" "BIRD") | |
66 | (?\e(1:\e(B consonant "LETTER BOR BED" "FISHHOOK") | |
67 | (?\e(1;\e(B consonant "LETTER POR PAR" "FISH") | |
68 | (?\e(1<\e(B consonant "LETTER HPOR HPER\"" "BEE") | |
69 | (?\e(1=\e(B consonant "LETTER FHOR FHAR" "WALL") | |
70 | (?\e(1>\e(B consonant "LETTER PHOR PHUU" "MOUNTAIN") | |
71 | (?\e(1?\e(B consonant "LETTER FOR FAI" "FIRE") | |
72 | (?\e(1@\e(B invalid nil) | |
73 | (?\e(1A\e(B consonant "LETTER MOR MAR\"" "HORSE") | |
74 | (?\e(1B\e(B consonant "LETTER GNOR GNAR" "MEDICINE") | |
75 | (?\e(1C\e(B consonant "LETTER ROR ROD" "CAR") | |
76 | (?\e(1D\e(B invalid nil) | |
77 | (?\e(1E\e(B consonant "LETTER LOR LIING" "MONKEY") | |
78 | (?\e(1F\e(B invalid nil) | |
79 | (?\e(1G\e(B consonant "LETTER WOR WII" "HAND FAN") | |
80 | (?\e(1H\e(B invalid nil) | |
81 | (?\e(1I\e(B invalid nil) | |
82 | (?\e(1J\e(B consonant "LETTER SOR SEA" "TIGER") | |
83 | (?\e(1K\e(B consonant "LETTER HHOR HHAI" "JAR") | |
84 | (?\e(1L\e(B invalid nil) | |
85 | (?\e(1M\e(B consonant "LETTER OR OOW" "TAKE") | |
86 | (?\e(1N\e(B consonant "LETTER HOR HEA" "BOAT") | |
87 | (?\e(1O\e(B special "ELLIPSIS") | |
88 | (?\e(1P\e(B vowel-base "VOWEL SIGN SARA A") | |
89 | (?\e(1Q\e(B vowel-upper "VOWEL SIGN MAI KAN") | |
90 | (?\e(1R\e(B vowel-base "VOWEL SIGN SARA AR") | |
91 | (?\e(1S\e(B vowel-base "VOWEL SIGN SARA AM") | |
a1506d29 | 92 | (?\e(1T\e(B vowel-upper "VOWEL SIGN SARA I") |
d92d0128 KH |
93 | (?\e(1U\e(B vowel-upper "VOWEL SIGN SARA II") |
94 | (?\e(1V\e(B vowel-upper "VOWEL SIGN SARA EU") | |
95 | (?\e(1W\e(B vowel-upper "VOWEL SIGN SARA UR") | |
96 | (?\e(1X\e(B vowel-lower "VOWEL SIGN SARA U") | |
97 | (?\e(1Y\e(B vowel-lower "VOWEL SIGN SARA UU") | |
98 | (?\e(1Z\e(B invalid nil) | |
99 | (?\e(1[\e(B vowel-upper "VOWEL SIGN MAI KONG") | |
100 | (?\e(1\\e(B semivowel-lower "SEMIVOWEL SIGN LO") | |
101 | (?\e(1]\e(B vowel-base "SEMIVOWEL SIGN SARA IA") | |
102 | (?\e(1^\e(B invalid nil) | |
103 | (?\e(1_\e(B invalid nil) | |
104 | (?\e(1`\e(B vowel-base "VOWEL SIGN SARA EE") | |
105 | (?\e(1a\e(B vowel-base "VOWEL SIGN SARA AA") | |
106 | (?\e(1b\e(B vowel-base "VOWEL SIGN SARA OO") | |
107 | (?\e(1c\e(B vowel-base "VOWEL SIGN SARA EI MAI MUAN\"") | |
108 | (?\e(1d\e(B vowel-base "VOWEL SIGN SARA AI MAI MAY") | |
109 | (?\e(1e\e(B invalid nil) | |
a1506d29 | 110 | (?\e(1f\e(B special "KO LA (REPETITION)") |
d92d0128 KH |
111 | (?\e(1g\e(B invalid nil) |
112 | (?\e(1h\e(B tone "TONE MAI EK") | |
113 | (?\e(1i\e(B tone "TONE MAI THO") | |
114 | (?\e(1j\e(B tone "TONE MAI TI") | |
115 | (?\e(1k\e(B tone "TONE MAI JADTAWAR") | |
116 | (?\e(1l\e(B tone "CANCELLATION MARK") | |
117 | (?\e(1m\e(B vowel-upper "VOWEL SIGN SARA OR") | |
118 | (?\e(1n\e(B invalid nil) | |
119 | (?\e(1o\e(B invalid nil) | |
120 | (?\e(1p\e(B special "DIGIT ZERO") | |
121 | (?\e(1q\e(B special "DIGIT ONE") | |
122 | (?\e(1r\e(B special "DIGIT TWO") | |
123 | (?\e(1s\e(B special "DIGIT THREE") | |
124 | (?\e(1t\e(B special "DIGIT FOUR") | |
125 | (?\e(1u\e(B special "DIGIT FIVE") | |
126 | (?\e(1v\e(B special "DIGIT SIX") | |
127 | (?\e(1w\e(B special "DIGIT SEVEN") | |
128 | (?\e(1x\e(B special "DIGIT EIGHT") | |
129 | (?\e(1y\e(B special "DIGIT NINE") | |
130 | (?\e(1z\e(B invalid nil) | |
131 | (?\e(1{\e(B invalid nil) | |
132 | (?\e(1|\e(B consonant "LETTER NHOR NHUU" "MOUSE") | |
133 | (?\e(1}\e(B consonant "LETTER MHOR MHAR" "DOG") | |
ffdc0bca | 134 | (?\e(1~\e(B invalid nil))) |
d92d0128 KH |
135 | elm) |
136 | (while l | |
31ce7719 KH |
137 | (setq elm (car l) l (cdr l)) |
138 | (let ((char (car elm)) | |
139 | (ptype (nth 1 elm))) | |
140 | (cond ((eq ptype 'consonant) | |
141 | (modify-category-entry char ?c lao-category-table)) | |
142 | ((memq ptype '(vowel-upper vowel-lower)) | |
143 | (modify-category-entry char ?v lao-category-table)) | |
144 | ((eq ptype 'semivowel-lower) | |
145 | (modify-category-entry char ?s lao-category-table)) | |
146 | ((eq ptype 'tone) | |
147 | (modify-category-entry char ?t lao-category-table))) | |
148 | (put-char-code-property char 'phonetic-type ptype) | |
149 | (put-char-code-property char 'name (nth 2 elm)) | |
150 | (put-char-code-property char 'meaning (nth 3 elm))))) | |
151 | ||
152 | ;; The general composing rules are as follows: | |
153 | ;; | |
154 | ;; T | |
155 | ;; V T V T | |
156 | ;; CV -> C, CT -> C, CVT -> C, Cv -> C, CvT -> C | |
157 | ;; v v | |
158 | ;; T | |
908ac9e2 KH |
159 | ;; V T V T |
160 | ;; CsV -> C, CsT -> C, CsVT -> C, Csv -> C, CsvT -> C | |
161 | ;; s s s s s | |
162 | ;; v v | |
31ce7719 KH |
163 | |
164 | ||
165 | ;; where C: consonant, V: vowel upper, v: vowel lower, | |
166 | ;; T: tone mark, s: semivowel lower | |
167 | ||
168 | (defvar lao-composition-pattern | |
169 | "\\cc\\(\\ct\\|\\cv\\ct?\\|\\cs\\(\\ct\\|\\cv\\ct?\\)?\\)" | |
170 | "Regular expression matching a Lao composite sequence.") | |
171 | ||
172 | ;;;###autoload | |
173 | (defun lao-compose-string (str) | |
174 | (with-category-table lao-category-table | |
175 | (let ((idx 0)) | |
176 | (while (setq idx (string-match lao-composition-pattern str idx)) | |
177 | (compose-string str idx (match-end 0)) | |
178 | (setq idx (match-end 0)))) | |
179 | str)) | |
180 | ||
181 | ;;; LRT: Lao <-> Roman Transcription | |
182 | ||
183 | ;; Upper vowels and tone-marks are put on the letter. | |
184 | ;; Semi-vowel-sign-lo and lower vowels are put under the letter. | |
185 | ||
186 | (defconst lao-transcription-consonant-alist | |
187 | (sort '(;; single consonants | |
188 | ("k" . "\e(1!\e(B") | |
189 | ("kh" . "\e(1"\e(B") | |
190 | ("qh" . "\e(1$\e(B") | |
191 | ("ng" . "\e(1'\e(B") | |
192 | ("j" . "\e(1(\e(B") | |
193 | ("s" . "\e(1J\e(B") | |
194 | ("x" . "\e(1*\e(B") | |
195 | ("y" . "\e(1-\e(B") | |
196 | ("d" . "\e(14\e(B") | |
197 | ("t" . "\e(15\e(B") | |
198 | ("th" . "\e(16\e(B") | |
199 | ("dh" . "\e(17\e(B") | |
200 | ("n" . "\e(19\e(B") | |
201 | ("b" . "\e(1:\e(B") | |
202 | ("p" . "\e(1;\e(B") | |
203 | ("hp" . "\e(1<\e(B") | |
204 | ("fh" . "\e(1=\e(B") | |
205 | ("ph" . "\e(1>\e(B") | |
206 | ("f" . "\e(1?\e(B") | |
207 | ("m" . "\e(1A\e(B") | |
208 | ("gn" . "\e(1B\e(B") | |
209 | ("l" . "\e(1E\e(B") | |
210 | ("r" . "\e(1C\e(B") | |
211 | ("v" . "\e(1G\e(B") | |
212 | ("w" . "\e(1G\e(B") | |
213 | ("hh" . "\e(1K\e(B") | |
214 | ("O" . "\e(1M\e(B") | |
215 | ("h" . "\e(1N\e(B") | |
216 | ("nh" . "\e(1|\e(B") | |
217 | ("mh" . "\e(1}\e(B") | |
218 | ("lh" . ["\e(1K\\e(B"]) | |
219 | ;; double consonants | |
220 | ("ngh" . ["\e(1K'\e(B"]) | |
221 | ("yh" . ["\e(1K]\e(B"]) | |
222 | ("wh" . ["\e(1KG\e(B"]) | |
223 | ("hl" . ["\e(1KE\e(B"]) | |
224 | ("hy" . ["\e(1K-\e(B"]) | |
225 | ("hn" . ["\e(1K9\e(B"]) | |
226 | ("hm" . ["\e(1KA\e(B"]) | |
227 | ) | |
228 | (function (lambda (x y) (> (length (car x)) (length (car y))))))) | |
229 | ||
230 | (defconst lao-transcription-semi-vowel-alist | |
231 | '(("r" . "\e(1\\e(B"))) | |
232 | ||
233 | (defconst lao-transcription-vowel-alist | |
234 | (sort '(("a" . "\e(1P\e(B") | |
235 | ("ar" . "\e(1R\e(B") | |
236 | ("i" . "\e(1T\e(B") | |
237 | ("ii" . "\e(1U\e(B") | |
238 | ("eu" . "\e(1V\e(B") | |
239 | ("ur" . "\e(1W\e(B") | |
240 | ("u" . "\e(1X\e(B") | |
241 | ("uu" . "\e(1Y\e(B") | |
242 | ("e" . ["\e(1`P\e(B"]) | |
243 | ("ee" . "\e(1`\e(B") | |
244 | ("ae" . ["\e(1aP\e(B"]) | |
245 | ("aa" . "\e(1a\e(B") | |
246 | ("o" . ["\e(1bP\e(B"]) | |
247 | ("oo" . "\e(1b\e(B") | |
248 | ("oe" . ["\e(1`RP\e(B"]) | |
249 | ("or" . "\e(1m\e(B") | |
250 | ("er" . ["\e(1`T\e(B"]) | |
251 | ("ir" . ["\e(1`U\e(B"]) | |
252 | ("ua" . ["\e(1[GP\e(B"]) | |
253 | ("uaa" . ["\e(1[G\e(B"]) | |
254 | ("ie" . ["\e(1`Q]P\e(B"]) | |
255 | ("ia" . ["\e(1`Q]\e(B"]) | |
256 | ("ea" . ["\e(1`VM\e(B"]) | |
257 | ("eaa" . ["\e(1`WM\e(B"]) | |
258 | ("ai" . "\e(1d\e(B") | |
259 | ("ei" . "\e(1c\e(B") | |
260 | ("ao" . ["\e(1`[R\e(B"]) | |
261 | ("aM" . "\e(1S\e(B")) | |
262 | (function (lambda (x y) (> (length (car x)) (length (car y))))))) | |
263 | ||
264 | ;; Maa-sakod is put at the tail. | |
265 | (defconst lao-transcription-maa-sakod-alist | |
266 | '(("k" . "\e(1!\e(B") | |
267 | ("g" . "\e(1'\e(B") | |
268 | ("y" . "\e(1-\e(B") | |
269 | ("d" . "\e(14\e(B") | |
270 | ("n" . "\e(19\e(B") | |
271 | ("b" . "\e(1:\e(B") | |
272 | ("m" . "\e(1A\e(B") | |
273 | ("v" . "\e(1G\e(B") | |
274 | ("w" . "\e(1G\e(B") | |
275 | )) | |
276 | ||
277 | (defconst lao-transcription-tone-alist | |
278 | '(("'" . "\e(1h\e(B") | |
279 | ("\"" . "\e(1i\e(B") | |
280 | ("^" . "\e(1j\e(B") | |
281 | ("+" . "\e(1k\e(B") | |
282 | ("~" . "\e(1l\e(B"))) | |
283 | ||
284 | (defconst lao-transcription-punctuation-alist | |
285 | '(("\\0" . "\e(1p\e(B") | |
286 | ("\\1" . "\e(1q\e(B") | |
287 | ("\\2" . "\e(1r\e(B") | |
288 | ("\\3" . "\e(1s\e(B") | |
289 | ("\\4" . "\e(1t\e(B") | |
290 | ("\\5" . "\e(1u\e(B") | |
291 | ("\\6" . "\e(1v\e(B") | |
292 | ("\\7" . "\e(1w\e(B") | |
293 | ("\\8" . "\e(1x\e(B") | |
294 | ("\\9" . "\e(1y\e(B") | |
295 | ("\\\\" . "\e(1f\e(B") | |
296 | ("\\$" . "\e(1O\e(B"))) | |
297 | ||
298 | (defconst lao-transcription-pattern | |
299 | (concat | |
300 | "\\(" | |
301 | (mapconcat 'car lao-transcription-consonant-alist "\\|") | |
302 | "\\)\\(" | |
303 | (mapconcat 'car lao-transcription-semi-vowel-alist "\\|") | |
304 | "\\)?\\(\\(" | |
305 | (mapconcat 'car lao-transcription-vowel-alist "\\|") | |
306 | "\\)\\(" | |
307 | (mapconcat 'car lao-transcription-maa-sakod-alist "\\|") | |
308 | "\\)?\\(" | |
309 | (mapconcat (lambda (x) (regexp-quote (car x))) | |
310 | lao-transcription-tone-alist "\\|") | |
311 | "\\)?\\)?\\|" | |
312 | (mapconcat (lambda (x) (regexp-quote (car x))) | |
313 | lao-transcription-punctuation-alist "\\|") | |
314 | ) | |
315 | "Regexp of Roman transcription pattern for one Lao syllable.") | |
316 | ||
317 | (defconst lao-transcription-pattern | |
318 | (concat | |
319 | "\\(" | |
320 | (regexp-opt (mapcar 'car lao-transcription-consonant-alist)) | |
321 | "\\)\\(" | |
322 | (regexp-opt (mapcar 'car lao-transcription-semi-vowel-alist)) | |
323 | "\\)?\\(\\(" | |
324 | (regexp-opt (mapcar 'car lao-transcription-vowel-alist)) | |
325 | "\\)\\(" | |
326 | (regexp-opt (mapcar 'car lao-transcription-maa-sakod-alist)) | |
327 | "\\)?\\(" | |
328 | (regexp-opt (mapcar 'car lao-transcription-tone-alist)) | |
329 | "\\)?\\)?\\|" | |
330 | (regexp-opt (mapcar 'car lao-transcription-punctuation-alist)) | |
331 | ) | |
332 | "Regexp of Roman transcription pattern for one Lao syllable.") | |
333 | ||
334 | (defconst lao-vowel-reordering-rule | |
335 | '(("\e(1P\e(B" (0 ?\e(1P\e(B) (0 ?\e(1Q\e(B)) | |
336 | ("\e(1R\e(B" (0 ?\e(1R\e(B)) | |
337 | ("\e(1T\e(B" (0 ?\e(1U\e(B)) | |
338 | ("\e(1U\e(B" (0 ?\e(1U\e(B)) | |
339 | ("\e(1V\e(B" (0 ?\e(1V\e(B)) | |
340 | ("\e(1W\e(B" (0 ?\e(1W\e(B)) | |
341 | ("\e(1X\e(B" (0 ?\e(1X\e(B)) | |
342 | ("\e(1Y\e(B" (0 ?\e(1Y\e(B)) | |
343 | ("\e(1`P\e(B" (?\e(1`\e(B 0 ?\e(1P\e(B) (?\e(1`\e(B 0 ?\e(1Q\e(B)) | |
344 | ("\e(1`\e(B" (?\e(1`\e(B 0)) | |
345 | ("\e(1aP\e(B" (?\e(1a\e(B 0 ?\e(1P\e(B) (?\e(1a\e(B 0 ?\e(1Q\e(B)) | |
346 | ("\e(1a\e(B" (?\e(1a\e(B 0)) | |
347 | ("\e(1bP\e(B" (?\e(1b\e(B 0 ?\e(1P\e(B) (0 ?\e(1[\e(B) (?\e(1-\e(B ?\e(1b\e(B 0 ?\e(1Q\e(B) (?\e(1G\e(B ?\e(1b\e(B 0 ?\e(1Q\e(B)) | |
348 | ("\e(1b\e(B" (?\e(1b\e(B 0)) | |
349 | ("\e(1`RP\e(B" (?\e(1`\e(B 0 ?\e(1R\e(B ?\e(1P\e(B) (0 ?\e(1Q\e(B ?\e(1M\e(B)) | |
350 | ("\e(1m\e(B" (0 ?\e(1m\e(B) (0 ?\e(1M\e(B)) | |
351 | ("\e(1`T\e(B" (?\e(1`\e(B 0 ?\e(1T\e(B)) | |
352 | ("\e(1`U\e(B" (?\e(1`\e(B 0 ?\e(1U\e(B)) | |
353 | ("\e(1[GP\e(B" (0 ?\e(1[\e(B ?\e(1G\e(B ?\e(1P\e(B) (0 ?\e(1Q\e(B ?\e(1G\e(B)) | |
354 | ("\e(1[G\e(B" (0 ?\e(1[\e(B ?\e(1G\e(B) (0 ?\e(1G\e(B)) | |
355 | ("\e(1`Q]P\e(B" (?\e(1`\e(B 0 ?\e(1Q\e(B ?\e(1]\e(B ?\e(1P\e(B) (0 ?\e(1Q\e(B ?\e(1]\e(B)) | |
356 | ("\e(1`Q]\e(B" (?\e(1`\e(B 0 ?\e(1Q\e(B ?\e(1]\e(B) (0 ?\e(1]\e(B)) | |
357 | ("\e(1`VM\e(B" (?\e(1`\e(B 0 ?\e(1V\e(B ?\e(1M\e(B)) | |
358 | ("\e(1`WM\e(B" (?\e(1`\e(B 0 ?\e(1W\e(B ?\e(1M\e(B)) | |
359 | ("\e(1d\e(B" (?\e(1d\e(B 0)) | |
360 | ("\e(1c\e(B" (?\e(1c\e(B 0)) | |
361 | ("\e(1`[R\e(B" (?\e(1`\e(B 0 ?\e(1[\e(B ?\e(1R\e(B)) | |
ffdc0bca | 362 | ("\e(1S\e(B" (0 ?\e(1S\e(B))) |
31ce7719 KH |
363 | "Alist of Lao vowel string vs the corresponding re-ordering rule. |
364 | Each element has this form: | |
365 | (VOWEL NO-MAA-SAKOD-RULE WITH-MAA-SAKOD-RULE (MAA-SAKOD-0 RULE-0) ...) | |
366 | ||
367 | VOWEL is a vowel string (e.g. \"\e(1`Q]P\e(B\"). | |
368 | ||
369 | NO-MAA-SAKOD-RULE is a rule to re-order and modify VOWEL following a | |
370 | consonant. It is a list vowel characters or 0. The element 0 | |
371 | indicate the place to embed a consonant. | |
372 | ||
373 | Optional WITH-MAA-SAKOD-RULE is a rule to re-order and modify VOWEL | |
c5e87d10 | 374 | following a consonant and preceding a maa-sakod character. If it is |
91af3942 | 375 | nil, NO-MAA-SAKOD-RULE is used. The maa-sakod character is always |
31ce7719 KH |
376 | appended at the tail. |
377 | ||
378 | For instance, rule `(\"\e(1`WM\e(B\" (?\e(1`\e(B t ?\e(1W\e(B ?\e(1M\e(B))' tells that this vowel | |
379 | string following a consonant `\e(1!\e(B' should be re-ordered as \"\e(1`!WM\e(B\". | |
380 | ||
381 | Optional (MAA-SAKOD-n RULE-n) are rules specially applied to maa-sakod | |
382 | character MAA-SAKOD-n.") | |
383 | ||
384 | ;;;###autoload | |
385 | (defun lao-transcribe-single-roman-syllable-to-lao (from to &optional str) | |
386 | "Transcribe a Romanized Lao syllable in the region FROM and TO to Lao string. | |
387 | Only the first syllable is transcribed. | |
388 | The value has the form: (START END LAO-STRING), where | |
fa463103 | 389 | START and END are the beginning and end positions of the Roman Lao syllable, |
31ce7719 KH |
390 | LAO-STRING is the Lao character transcription of it. |
391 | ||
392 | Optional 3rd arg STR, if non-nil, is a string to search for Roman Lao | |
393 | syllable. In that case, FROM and TO are indexes to STR." | |
394 | (if str | |
395 | (if (setq from (string-match lao-transcription-pattern str from)) | |
396 | (progn | |
397 | (if (>= from to) | |
398 | (setq from nil) | |
399 | (setq to (match-end 0))))) | |
400 | (save-excursion | |
401 | (goto-char from) | |
402 | (if (setq to (re-search-forward lao-transcription-pattern to t)) | |
403 | (setq from (match-beginning 0)) | |
404 | (setq from nil)))) | |
405 | (if from | |
406 | (let* ((consonant (match-string 1 str)) | |
407 | (semivowel (match-string 3 str)) | |
408 | (vowel (match-string 5 str)) | |
409 | (maa-sakod (match-string 8 str)) | |
410 | (tone (match-string 9 str)) | |
411 | lao-consonant lao-semivowel lao-vowel lao-maa-sakod lao-tone | |
412 | clen cidx) | |
413 | (setq to (match-end 0)) | |
414 | (if (not consonant) | |
415 | (setq str (cdr (assoc (match-string 0 str) | |
416 | lao-transcription-punctuation-alist))) | |
417 | (setq lao-consonant | |
418 | (cdr (assoc consonant lao-transcription-consonant-alist))) | |
419 | (if (vectorp lao-consonant) | |
420 | (setq lao-consonant (aref lao-consonant 0))) | |
421 | (setq clen (length lao-consonant)) | |
422 | (if semivowel | |
423 | ;; Include semivowel in STR. | |
424 | (setq lao-semivowel | |
425 | (cdr (assoc semivowel lao-transcription-semi-vowel-alist)) | |
426 | str (if (= clen 1) | |
427 | (concat lao-consonant lao-semivowel) | |
428 | (concat (substring lao-consonant 0 1) lao-semivowel | |
429 | (substring lao-consonant 1)))) | |
430 | (setq str lao-consonant)) | |
431 | (if vowel | |
432 | (let (rule) | |
433 | (setq lao-vowel | |
434 | (cdr (assoc vowel lao-transcription-vowel-alist))) | |
435 | (if (vectorp lao-vowel) | |
436 | (setq lao-vowel (aref lao-vowel 0))) | |
437 | (setq rule (assoc lao-vowel lao-vowel-reordering-rule)) | |
438 | (if (null maa-sakod) | |
439 | (setq rule (nth 1 rule)) | |
440 | (setq lao-maa-sakod | |
441 | (cdr (assoc maa-sakod lao-transcription-maa-sakod-alist)) | |
442 | rule | |
443 | (or (cdr (assq (aref lao-maa-sakod 0) (nthcdr 2 rule))) | |
444 | (nth 2 rule) | |
445 | (nth 1 rule)))) | |
446 | (or rule | |
447 | (error "Lao vowel %S has no re-ordering rule" lao-vowel)) | |
448 | (setq lao-consonant str str "") | |
449 | (while rule | |
450 | (if (= (car rule) 0) | |
451 | (setq str (concat str lao-consonant) | |
452 | cidx (length str)) | |
453 | (setq str (concat str (list (car rule))))) | |
454 | (setq rule (cdr rule))) | |
455 | (or cidx | |
456 | (error "Lao vowel %S has malformed re-ordering rule" vowel)) | |
457 | ;; Set CIDX to after upper or lower vowel if any. | |
458 | (let ((len (length str))) | |
459 | (while (and (< cidx len) | |
460 | (memq (get-char-code-property (aref str cidx) | |
461 | 'phonetic-type) | |
462 | '(vowel-lower vowel-upper))) | |
463 | (setq cidx (1+ cidx)))) | |
464 | (if lao-maa-sakod | |
465 | (setq str (concat str lao-maa-sakod))) | |
466 | (if tone | |
467 | (setq lao-tone | |
468 | (cdr (assoc tone lao-transcription-tone-alist)) | |
469 | str (concat (substring str 0 cidx) lao-tone | |
470 | (substring str cidx))))))) | |
471 | (list from to (lao-compose-string str))))) | |
472 | ||
473 | ;;;###autoload | |
474 | (defun lao-transcribe-roman-to-lao-string (str) | |
475 | "Transcribe Romanized Lao string STR to Lao character string." | |
476 | (let ((from 0) | |
477 | (to (length str)) | |
478 | (lao-str "") | |
479 | val) | |
480 | (while (setq val (lao-transcribe-single-roman-syllable-to-lao from to str)) | |
481 | (let ((start (car val)) | |
482 | (end (nth 1 val)) | |
483 | (lao (nth 2 val))) | |
484 | (if (> start from) | |
485 | (setq lao-str (concat lao-str (substring str from start) lao)) | |
486 | (setq lao-str (concat lao-str lao))) | |
487 | (setq from end))) | |
488 | (if (< from to) | |
489 | (concat lao-str (substring str from to)) | |
490 | lao-str))) | |
491 | ||
a549727e | 492 | ;;;###autoload |
70a6526c KH |
493 | (defun lao-composition-function (gstring) |
494 | (if (= (lgstring-char-len gstring) 1) | |
39a7775d | 495 | (compose-gstring-for-graphic gstring) |
70a6526c | 496 | (or (font-shape-gstring gstring) |
39a7775d KH |
497 | (let ((glyph-len (lgstring-glyph-len gstring)) |
498 | (i 0) | |
499 | glyph) | |
500 | (while (and (< i glyph-len) | |
501 | (setq glyph (lgstring-glyph gstring i))) | |
502 | (setq i (1+ i))) | |
70a6526c | 503 | (compose-glyph-string-relative gstring 0 i 0.1))))) |
d92d0128 | 504 | |
9b21ecc2 KH |
505 | ;;;###autoload |
506 | (defun lao-compose-region (from to) | |
507 | (interactive "r") | |
508 | (save-restriction | |
509 | (narrow-to-region from to) | |
510 | (goto-char (point-min)) | |
511 | (with-category-table lao-category-table | |
512 | (while (re-search-forward lao-composition-pattern nil t) | |
513 | (compose-region (match-beginning 0) (point)))))) | |
514 | ||
d92d0128 | 515 | ;; |
650e8505 | 516 | (provide 'lao-util) |
d92d0128 KH |
517 | |
518 | ;;; lao-util.el ends here |