771
+ − 1 ;;; lao-util.el --- utilities for Lao -*- coding: iso-2022-7bit; -*-
+ − 2
+ − 3 ;; Copyright (C) 1997 Electrotechnical Laboratory, JAPAN.
+ − 4 ;; Licensed to the Free Software Foundation.
+ − 5
+ − 6 ;; Keywords: multilingual, Lao
+ − 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
778
+ − 25 ;;; Synched up with: Emacs 21.1 (language/lao-util.el).
+ − 26
+ − 27 ;;; Commentary:
771
+ − 28
+ − 29 ;;; Code:
+ − 30
+ − 31 ;; Setting information of Thai characters.
+ − 32
+ − 33 (defconst lao-category-table (make-category-table))
+ − 34 (define-category ?c "Lao consonant" lao-category-table)
+ − 35 (define-category ?s "Lao semi-vowel" lao-category-table)
+ − 36 (define-category ?v "Lao upper/lower vowel" lao-category-table)
+ − 37 (define-category ?t "Lao tone" lao-category-table)
+ − 38
+ − 39 (let ((l '((?(1!(B consonant "LETTER KOR KAI'" "CHICKEN")
+ − 40 (?(1"(B consonant "LETTER KHOR KHAI'" "EGG")
+ − 41 (?(1#(B invalid nil)
+ − 42 (?(1$(B consonant "LETTER QHOR QHWARGN" "BUFFALO")
+ − 43 (?(1%(B invalid nil)
+ − 44 (? invalid nil)
+ − 45 (?(1'(B consonant "LETTER NGOR NGUU" "SNAKE")
+ − 46 (?(1((B consonant "LETTER JOR JUA" "BUDDHIST NOVICE")
+ − 47 (?(1)(B invalid nil)
+ − 48 (?(1*(B consonant "LETTER XOR X\"ARNG" "ELEPHANT")
+ − 49 (?(1+(B invalid nil)
+ − 50 (?(1,(B invalid nil)
+ − 51 (?(1-(B consonant "LETTER YOR YUNG" "MOSQUITO")
+ − 52 (?(1.(B invalid nil)
+ − 53 (?(1.(B invalid nil)
+ − 54 (?(1.(B invalid nil)
+ − 55 (?(1.(B invalid nil)
+ − 56 (?(1.(B invalid nil)
+ − 57 (?(1.(B invalid nil)
+ − 58 (?(14(B consonant "LETTER DOR DANG" "NOSE")
+ − 59 (?(15(B consonant "LETTER TOR TAR" "EYE")
+ − 60 (?(16(B consonant "LETTER THOR THUNG" "TO ASK,QUESTION")
+ − 61 (?(17(B consonant "LETTER DHOR DHARM" "FLAG")
+ − 62 (?(18(B invalid nil)
+ − 63 (?(19(B consonant "LETTER NOR NOK" "BIRD")
+ − 64 (?(1:(B consonant "LETTER BOR BED" "FISHHOOK")
+ − 65 (?(1;(B consonant "LETTER POR PAR" "FISH")
+ − 66 (?(1<(B consonant "LETTER HPOR HPER\"" "BEE")
+ − 67 (?(1=(B consonant "LETTER FHOR FHAR" "WALL")
+ − 68 (?(1>(B consonant "LETTER PHOR PHUU" "MOUNTAIN")
+ − 69 (?(1?(B consonant "LETTER FOR FAI" "FIRE")
+ − 70 (?(1@(B invalid nil)
+ − 71 (?(1A(B consonant "LETTER MOR MAR\"" "HORSE")
+ − 72 (?(1B(B consonant "LETTER GNOR GNAR" "MEDICINE")
+ − 73 (?(1C(B consonant "LETTER ROR ROD" "CAR")
+ − 74 (?(1D(B invalid nil)
+ − 75 (?(1E(B consonant "LETTER LOR LIING" "MONKEY")
+ − 76 (?(1F(B invalid nil)
+ − 77 (?(1G(B consonant "LETTER WOR WII" "HAND FAN")
+ − 78 (?(1H(B invalid nil)
+ − 79 (?(1I(B invalid nil)
+ − 80 (?(1J(B consonant "LETTER SOR SEA" "TIGER")
+ − 81 (?(1K(B consonant "LETTER HHOR HHAI" "JAR")
+ − 82 (?(1L(B invalid nil)
+ − 83 (?(1M(B consonant "LETTER OR OOW" "TAKE")
+ − 84 (?(1N(B consonant "LETTER HOR HEA" "BOAT")
+ − 85 (?(1O(B special "ELLIPSIS")
+ − 86 (?(1P(B vowel-base "VOWEL SIGN SARA A")
+ − 87 (?(1Q(B vowel-upper "VOWEL SIGN MAI KAN")
+ − 88 (?(1R(B vowel-base "VOWEL SIGN SARA AR")
+ − 89 (?(1S(B vowel-base "VOWEL SIGN SARA AM")
+ − 90 (?(1T(B vowel-upper "VOWEL SIGN SARA I")
+ − 91 (?(1U(B vowel-upper "VOWEL SIGN SARA II")
+ − 92 (?(1V(B vowel-upper "VOWEL SIGN SARA EU")
+ − 93 (?(1W(B vowel-upper "VOWEL SIGN SARA UR")
+ − 94 (?(1X(B vowel-lower "VOWEL SIGN SARA U")
+ − 95 (?(1Y(B vowel-lower "VOWEL SIGN SARA UU")
+ − 96 (?(1Z(B invalid nil)
+ − 97 (?(1[(B vowel-upper "VOWEL SIGN MAI KONG")
+ − 98 (?(1\(B semivowel-lower "SEMIVOWEL SIGN LO")
+ − 99 (?(1](B vowel-base "SEMIVOWEL SIGN SARA IA")
+ − 100 (?(1^(B invalid nil)
+ − 101 (?(1_(B invalid nil)
+ − 102 (?(1`(B vowel-base "VOWEL SIGN SARA EE")
+ − 103 (?(1a(B vowel-base "VOWEL SIGN SARA AA")
+ − 104 (?(1b(B vowel-base "VOWEL SIGN SARA OO")
+ − 105 (?(1c(B vowel-base "VOWEL SIGN SARA EI MAI MUAN\"")
+ − 106 (?(1d(B vowel-base "VOWEL SIGN SARA AI MAI MAY")
+ − 107 (?(1e(B invalid nil)
+ − 108 (?(1f(B special "KO LA (REPETITION)")
+ − 109 (?(1g(B invalid nil)
+ − 110 (?(1h(B tone "TONE MAI EK")
+ − 111 (?(1i(B tone "TONE MAI THO")
+ − 112 (?(1j(B tone "TONE MAI TI")
+ − 113 (?(1k(B tone "TONE MAI JADTAWAR")
+ − 114 (?(1l(B tone "CANCELLATION MARK")
+ − 115 (?(1m(B vowel-upper "VOWEL SIGN SARA OR")
+ − 116 (?(1n(B invalid nil)
+ − 117 (?(1o(B invalid nil)
+ − 118 (?(1p(B special "DIGIT ZERO")
+ − 119 (?(1q(B special "DIGIT ONE")
+ − 120 (?(1r(B special "DIGIT TWO")
+ − 121 (?(1s(B special "DIGIT THREE")
+ − 122 (?(1t(B special "DIGIT FOUR")
+ − 123 (?(1u(B special "DIGIT FIVE")
+ − 124 (?(1v(B special "DIGIT SIX")
+ − 125 (?(1w(B special "DIGIT SEVEN")
+ − 126 (?(1x(B special "DIGIT EIGHT")
+ − 127 (?(1y(B special "DIGIT NINE")
+ − 128 (?(1z(B invalid nil)
+ − 129 (?(1{(B invalid nil)
+ − 130 (?(1|(B consonant "LETTER NHOR NHUU" "MOUSE")
+ − 131 (?(1}(B consonant "LETTER MHOR MHAR" "DOG")
+ − 132 (?(1~(B invalid nil)
+ − 133 ))
+ − 134 elm)
+ − 135 (while l
+ − 136 (setq elm (car l) l (cdr l))
+ − 137 (let ((char (car elm))
+ − 138 (ptype (nth 1 elm)))
+ − 139 (cond ((eq ptype 'consonant)
+ − 140 (modify-category-entry char ?c lao-category-table))
+ − 141 ((memq ptype '(vowel-upper vowel-lower))
+ − 142 (modify-category-entry char ?v lao-category-table))
+ − 143 ((eq ptype 'semivowel-lower)
+ − 144 (modify-category-entry char ?s lao-category-table))
+ − 145 ((eq ptype 'tone)
+ − 146 (modify-category-entry char ?t lao-category-table)))
+ − 147 (put-char-code-property char 'phonetic-type ptype)
+ − 148 (put-char-code-property char 'name (nth 2 elm))
+ − 149 (put-char-code-property char 'meaning (nth 3 elm)))))
+ − 150
+ − 151 ;; The general composing rules are as follows:
+ − 152 ;;
+ − 153 ;; T
+ − 154 ;; V T V T
+ − 155 ;; CV -> C, CT -> C, CVT -> C, Cv -> C, CvT -> C
+ − 156 ;; v v
+ − 157 ;; T
+ − 158 ;; V T V T
+ − 159 ;; CsV -> C, CsT -> C, CsVT -> C, Csv -> C, CvT -> C
+ − 160 ;; s s s s s
+ − 161 ;; v v
+ − 162
+ − 163
+ − 164 ;; where C: consonant, V: vowel upper, v: vowel lower,
+ − 165 ;; T: tone mark, s: semivowel lower
+ − 166
+ − 167 (defvar lao-composition-pattern
+ − 168 "\\cc\\(\\ct\\|\\cv\\ct?\\|\\cs\\(\\ct\\|\\cv\\ct?\\)?\\)"
+ − 169 "Regular expression matching a Lao composite sequence.")
+ − 170
+ − 171 ;;;###autoload
+ − 172 (defun lao-compose-string (str)
+ − 173 (with-category-table lao-category-table
+ − 174 (let ((idx 0))
+ − 175 (while (setq idx (string-match lao-composition-pattern str idx))
+ − 176 (compose-string str idx (match-end 0))
+ − 177 (setq idx (match-end 0))))
+ − 178 str))
+ − 179
+ − 180 ;;; LRT: Lao <-> Roman Transcription
+ − 181
+ − 182 ;; Upper vowels and tone-marks are put on the letter.
+ − 183 ;; Semi-vowel-sign-lo and lower vowels are put under the letter.
+ − 184
+ − 185 (defconst lao-transcription-consonant-alist
+ − 186 (sort '(;; single consonants
+ − 187 ("k" . "(1!(B")
+ − 188 ("kh" . "(1"(B")
+ − 189 ("qh" . "(1$(B")
+ − 190 ("ng" . "(1'(B")
+ − 191 ("j" . "(1((B")
+ − 192 ("s" . "(1J(B")
+ − 193 ("x" . "(1*(B")
+ − 194 ("y" . "(1-(B")
+ − 195 ("d" . "(14(B")
+ − 196 ("t" . "(15(B")
+ − 197 ("th" . "(16(B")
+ − 198 ("dh" . "(17(B")
+ − 199 ("n" . "(19(B")
+ − 200 ("b" . "(1:(B")
+ − 201 ("p" . "(1;(B")
+ − 202 ("hp" . "(1<(B")
+ − 203 ("fh" . "(1=(B")
+ − 204 ("ph" . "(1>(B")
+ − 205 ("f" . "(1?(B")
+ − 206 ("m" . "(1A(B")
+ − 207 ("gn" . "(1B(B")
+ − 208 ("l" . "(1E(B")
+ − 209 ("r" . "(1C(B")
+ − 210 ("v" . "(1G(B")
+ − 211 ("w" . "(1G(B")
+ − 212 ("hh" . "(1K(B")
+ − 213 ("O" . "(1M(B")
+ − 214 ("h" . "(1N(B")
+ − 215 ("nh" . "(1|(B")
+ − 216 ("mh" . "(1}(B")
+ − 217 ("lh" . ["(1K\(B"])
+ − 218 ;; double consonants
+ − 219 ("ngh" . ["(1K'(B"])
+ − 220 ("yh" . ["(1K](B"])
+ − 221 ("wh" . ["(1KG(B"])
+ − 222 ("hl" . ["(1KE(B"])
+ − 223 ("hy" . ["(1K-(B"])
+ − 224 ("hn" . ["(1K9(B"])
+ − 225 ("hm" . ["(1KA(B"])
+ − 226 )
+ − 227 (function (lambda (x y) (> (length (car x)) (length (car y)))))))
+ − 228
+ − 229 (defconst lao-transcription-semi-vowel-alist
+ − 230 '(("r" . "(1\(B")))
+ − 231
+ − 232 (defconst lao-transcription-vowel-alist
+ − 233 (sort '(("a" . "(1P(B")
+ − 234 ("ar" . "(1R(B")
+ − 235 ("i" . "(1T(B")
+ − 236 ("ii" . "(1U(B")
+ − 237 ("eu" . "(1V(B")
+ − 238 ("ur" . "(1W(B")
+ − 239 ("u" . "(1X(B")
+ − 240 ("uu" . "(1Y(B")
+ − 241 ("e" . ["(1`P(B"])
+ − 242 ("ee" . "(1`(B")
+ − 243 ("ae" . ["(1aP(B"])
+ − 244 ("aa" . "(1a(B")
+ − 245 ("o" . ["(1bP(B"])
+ − 246 ("oo" . "(1b(B")
+ − 247 ("oe" . ["(1`RP(B"])
+ − 248 ("or" . "(1m(B")
+ − 249 ("er" . ["(1`T(B"])
+ − 250 ("ir" . ["(1`U(B"])
+ − 251 ("ua" . ["(1[GP(B"])
+ − 252 ("uaa" . ["(1[G(B"])
+ − 253 ("ie" . ["(1`Q]P(B"])
+ − 254 ("ia" . ["(1`Q](B"])
+ − 255 ("ea" . ["(1`VM(B"])
+ − 256 ("eaa" . ["(1`WM(B"])
+ − 257 ("ai" . "(1d(B")
+ − 258 ("ei" . "(1c(B")
+ − 259 ("ao" . ["(1`[R(B"])
+ − 260 ("aM" . "(1S(B"))
+ − 261 (function (lambda (x y) (> (length (car x)) (length (car y)))))))
+ − 262
+ − 263 ;; Maa-sakod is put at the tail.
+ − 264 (defconst lao-transcription-maa-sakod-alist
+ − 265 '(("k" . "(1!(B")
+ − 266 ("g" . "(1'(B")
+ − 267 ("y" . "(1-(B")
+ − 268 ("d" . "(14(B")
+ − 269 ("n" . "(19(B")
+ − 270 ("b" . "(1:(B")
+ − 271 ("m" . "(1A(B")
+ − 272 ("v" . "(1G(B")
+ − 273 ("w" . "(1G(B")
+ − 274 ))
+ − 275
+ − 276 (defconst lao-transcription-tone-alist
+ − 277 '(("'" . "(1h(B")
+ − 278 ("\"" . "(1i(B")
+ − 279 ("^" . "(1j(B")
+ − 280 ("+" . "(1k(B")
+ − 281 ("~" . "(1l(B")))
+ − 282
+ − 283 (defconst lao-transcription-punctuation-alist
+ − 284 '(("\\0" . "(1p(B")
+ − 285 ("\\1" . "(1q(B")
+ − 286 ("\\2" . "(1r(B")
+ − 287 ("\\3" . "(1s(B")
+ − 288 ("\\4" . "(1t(B")
+ − 289 ("\\5" . "(1u(B")
+ − 290 ("\\6" . "(1v(B")
+ − 291 ("\\7" . "(1w(B")
+ − 292 ("\\8" . "(1x(B")
+ − 293 ("\\9" . "(1y(B")
+ − 294 ("\\\\" . "(1f(B")
+ − 295 ("\\$" . "(1O(B")))
+ − 296
+ − 297 (defconst lao-transcription-pattern
+ − 298 (concat
+ − 299 "\\("
+ − 300 (mapconcat 'car lao-transcription-consonant-alist "\\|")
+ − 301 "\\)\\("
+ − 302 (mapconcat 'car lao-transcription-semi-vowel-alist "\\|")
+ − 303 "\\)?\\(\\("
+ − 304 (mapconcat 'car lao-transcription-vowel-alist "\\|")
+ − 305 "\\)\\("
+ − 306 (mapconcat 'car lao-transcription-maa-sakod-alist "\\|")
+ − 307 "\\)?\\("
+ − 308 (mapconcat (lambda (x) (regexp-quote (car x)))
+ − 309 lao-transcription-tone-alist "\\|")
+ − 310 "\\)?\\)?\\|"
+ − 311 (mapconcat (lambda (x) (regexp-quote (car x)))
+ − 312 lao-transcription-punctuation-alist "\\|")
+ − 313 )
+ − 314 "Regexp of Roman transcription pattern for one Lao syllable.")
+ − 315
+ − 316 (defconst lao-transcription-pattern
+ − 317 (concat
+ − 318 "\\("
+ − 319 (regexp-opt (mapcar 'car lao-transcription-consonant-alist))
+ − 320 "\\)\\("
+ − 321 (regexp-opt (mapcar 'car lao-transcription-semi-vowel-alist))
+ − 322 "\\)?\\(\\("
+ − 323 (regexp-opt (mapcar 'car lao-transcription-vowel-alist))
+ − 324 "\\)\\("
+ − 325 (regexp-opt (mapcar 'car lao-transcription-maa-sakod-alist))
+ − 326 "\\)?\\("
+ − 327 (regexp-opt (mapcar 'car lao-transcription-tone-alist))
+ − 328 "\\)?\\)?\\|"
+ − 329 (regexp-opt (mapcar 'car lao-transcription-punctuation-alist))
+ − 330 )
+ − 331 "Regexp of Roman transcription pattern for one Lao syllable.")
+ − 332
+ − 333 (defconst lao-vowel-reordering-rule
+ − 334 '(("(1P(B" (0 ?(1P(B) (0 ?(1Q(B))
+ − 335 ("(1R(B" (0 ?(1R(B))
+ − 336 ("(1T(B" (0 ?(1U(B))
+ − 337 ("(1U(B" (0 ?(1U(B))
+ − 338 ("(1V(B" (0 ?(1V(B))
+ − 339 ("(1W(B" (0 ?(1W(B))
+ − 340 ("(1X(B" (0 ?(1X(B))
+ − 341 ("(1Y(B" (0 ?(1Y(B))
+ − 342 ("(1`P(B" (?(1`(B 0 ?(1P(B) (?(1`(B 0 ?(1Q(B))
+ − 343 ("(1`(B" (?(1`(B 0))
+ − 344 ("(1aP(B" (?(1a(B 0 ?(1P(B) (?(1a(B 0 ?(1Q(B))
+ − 345 ("(1a(B" (?(1a(B 0))
+ − 346 ("(1bP(B" (?(1b(B 0 ?(1P(B) (0 ?(1[(B) (?(1-(B ?(1b(B 0 ?(1Q(B) (?(1G(B ?(1b(B 0 ?(1Q(B))
+ − 347 ("(1b(B" (?(1b(B 0))
+ − 348 ("(1`RP(B" (?(1`(B 0 ?(1R(B ?(1P(B) (0 ?(1Q(B ?(1M(B))
+ − 349 ("(1m(B" (0 ?(1m(B) (0 ?(1M(B))
+ − 350 ("(1`T(B" (?(1`(B 0 ?(1T(B))
+ − 351 ("(1`U(B" (?(1`(B 0 ?(1U(B))
+ − 352 ("(1[GP(B" (0 ?(1[(B ?(1G(B ?(1P(B) (0 ?(1Q(B ?(1G(B))
+ − 353 ("(1[G(B" (0 ?(1[(B ?(1G(B) (0 ?(1G(B))
+ − 354 ("(1`Q]P(B" (?(1`(B 0 ?(1Q(B ?(1](B ?(1P(B) (0 ?(1Q(B ?(1](B))
+ − 355 ("(1`Q](B" (?(1`(B 0 ?(1Q(B ?(1](B) (0 ?(1](B))
+ − 356 ("(1`VM(B" (?(1`(B 0 ?(1V(B ?(1M(B))
+ − 357 ("(1`WM(B" (?(1`(B 0 ?(1W(B ?(1M(B))
+ − 358 ("(1d(B" (?(1d(B 0))
+ − 359 ("(1c(B" (?(1c(B 0))
+ − 360 ("(1`[R(B" (?(1`(B 0 ?(1[(B ?(1R(B))
+ − 361 ("(1S(B" (0 ?(1S(B)))
+ − 362 "Alist of Lao vowel string vs the corresponding re-ordering rule.
+ − 363 Each element has this form:
+ − 364 (VOWEL NO-MAA-SAKOD-RULE WITH-MAA-SAKOD-RULE (MAA-SAKOD-0 RULE-0) ...)
+ − 365
+ − 366 VOWEL is a vowel string (e.g. \"(1`Q]P(B\").
+ − 367
+ − 368 NO-MAA-SAKOD-RULE is a rule to re-order and modify VOWEL following a
+ − 369 consonant. It is a list vowel characters or 0. The element 0
+ − 370 indicate the place to embed a consonant.
+ − 371
+ − 372 Optional WITH-MAA-SAKOD-RULE is a rule to re-order and modify VOWEL
+ − 373 follwoing a consonant and preceding a maa-sakod character. If it is
+ − 374 nil, NO-MAA-SAKOD-RULE is used. The maa-sakod character is alwasy
+ − 375 appended at the tail.
+ − 376
+ − 377 For instance, rule `(\"(1`WM(B\" (?(1`(B t ?(1W(B ?(1M(B))' tells that this vowel
+ − 378 string following a consonant `(1!(B' should be re-ordered as \"(1`!WM(B\".
+ − 379
+ − 380 Optional (MAA-SAKOD-n RULE-n) are rules specially applied to maa-sakod
+ − 381 character MAA-SAKOD-n.")
+ − 382
+ − 383 ;;;###autoload
+ − 384 (defun lao-transcribe-single-roman-syllable-to-lao (from to &optional str)
+ − 385 "Transcribe a Romanized Lao syllable in the region FROM and TO to Lao string.
+ − 386 Only the first syllable is transcribed.
+ − 387 The value has the form: (START END LAO-STRING), where
+ − 388 START and END are the beggining and end positions of the Roman Lao syllable,
+ − 389 LAO-STRING is the Lao character transcription of it.
+ − 390
+ − 391 Optional 3rd arg STR, if non-nil, is a string to search for Roman Lao
+ − 392 syllable. In that case, FROM and TO are indexes to STR."
+ − 393 (if str
+ − 394 (if (setq from (string-match lao-transcription-pattern str from))
+ − 395 (progn
+ − 396 (if (>= from to)
+ − 397 (setq from nil)
+ − 398 (setq to (match-end 0)))))
+ − 399 (save-excursion
+ − 400 (goto-char from)
+ − 401 (if (setq to (re-search-forward lao-transcription-pattern to t))
+ − 402 (setq from (match-beginning 0))
+ − 403 (setq from nil))))
+ − 404 (if from
+ − 405 (let* ((consonant (match-string 1 str))
+ − 406 (semivowel (match-string 3 str))
+ − 407 (vowel (match-string 5 str))
+ − 408 (maa-sakod (match-string 8 str))
+ − 409 (tone (match-string 9 str))
+ − 410 lao-consonant lao-semivowel lao-vowel lao-maa-sakod lao-tone
+ − 411 clen cidx)
+ − 412 (setq to (match-end 0))
+ − 413 (if (not consonant)
+ − 414 (setq str (cdr (assoc (match-string 0 str)
+ − 415 lao-transcription-punctuation-alist)))
+ − 416 (setq lao-consonant
+ − 417 (cdr (assoc consonant lao-transcription-consonant-alist)))
+ − 418 (if (vectorp lao-consonant)
+ − 419 (setq lao-consonant (aref lao-consonant 0)))
+ − 420 (setq clen (length lao-consonant))
+ − 421 (if semivowel
+ − 422 ;; Include semivowel in STR.
+ − 423 (setq lao-semivowel
+ − 424 (cdr (assoc semivowel lao-transcription-semi-vowel-alist))
+ − 425 str (if (= clen 1)
+ − 426 (concat lao-consonant lao-semivowel)
+ − 427 (concat (substring lao-consonant 0 1) lao-semivowel
+ − 428 (substring lao-consonant 1))))
+ − 429 (setq str lao-consonant))
+ − 430 (if vowel
+ − 431 (let (rule)
+ − 432 (setq lao-vowel
+ − 433 (cdr (assoc vowel lao-transcription-vowel-alist)))
+ − 434 (if (vectorp lao-vowel)
+ − 435 (setq lao-vowel (aref lao-vowel 0)))
+ − 436 (setq rule (assoc lao-vowel lao-vowel-reordering-rule))
+ − 437 (if (null maa-sakod)
+ − 438 (setq rule (nth 1 rule))
+ − 439 (setq lao-maa-sakod
+ − 440 (cdr (assoc maa-sakod lao-transcription-maa-sakod-alist))
+ − 441 rule
+ − 442 (or (cdr (assq (aref lao-maa-sakod 0) (nthcdr 2 rule)))
+ − 443 (nth 2 rule)
+ − 444 (nth 1 rule))))
+ − 445 (or rule
+ − 446 (error "Lao vowel %S has no re-ordering rule" lao-vowel))
+ − 447 (setq lao-consonant str str "")
+ − 448 (while rule
+ − 449 (if (= (car rule) 0)
+ − 450 (setq str (concat str lao-consonant)
+ − 451 cidx (length str))
+ − 452 (setq str (concat str (list (car rule)))))
+ − 453 (setq rule (cdr rule)))
+ − 454 (or cidx
+ − 455 (error "Lao vowel %S has malformed re-ordering rule" vowel))
+ − 456 ;; Set CIDX to after upper or lower vowel if any.
+ − 457 (let ((len (length str)))
+ − 458 (while (and (< cidx len)
+ − 459 (memq (get-char-code-property (aref str cidx)
+ − 460 'phonetic-type)
+ − 461 '(vowel-lower vowel-upper)))
+ − 462 (setq cidx (1+ cidx))))
+ − 463 (if lao-maa-sakod
+ − 464 (setq str (concat str lao-maa-sakod)))
+ − 465 (if tone
+ − 466 (setq lao-tone
+ − 467 (cdr (assoc tone lao-transcription-tone-alist))
+ − 468 str (concat (substring str 0 cidx) lao-tone
+ − 469 (substring str cidx)))))))
+ − 470 (list from to (lao-compose-string str)))))
+ − 471
+ − 472 ;;;###autoload
+ − 473 (defun lao-transcribe-roman-to-lao-string (str)
+ − 474 "Transcribe Romanized Lao string STR to Lao character string."
+ − 475 (let ((from 0)
+ − 476 (to (length str))
+ − 477 (lao-str "")
+ − 478 val)
+ − 479 (while (setq val (lao-transcribe-single-roman-syllable-to-lao from to str))
+ − 480 (let ((start (car val))
+ − 481 (end (nth 1 val))
+ − 482 (lao (nth 2 val)))
+ − 483 (if (> start from)
+ − 484 (setq lao-str (concat lao-str (substring str from start) lao))
+ − 485 (setq lao-str (concat lao-str lao)))
+ − 486 (setq from end)))
+ − 487 (if (< from to)
+ − 488 (concat lao-str (substring str from to))
+ − 489 lao-str)))
+ − 490
+ − 491 ;;;###autoload
+ − 492 (defun lao-composition-function (from to pattern &optional string)
+ − 493 "Compose Lao text in the region FROM and TO.
+ − 494 The text matches the regular expression PATTERN.
+ − 495 Optional 4th argument STRING, if non-nil, is a string containing text
+ − 496 to compose.
+ − 497
+ − 498 The return value is number of composed characters."
+ − 499 (if (< (1+ from) to)
+ − 500 (prog1 (- to from)
+ − 501 (if string
+ − 502 (compose-string string from to)
+ − 503 (compose-region from to))
+ − 504 (- to from))))
+ − 505
+ − 506 ;;;###autoload
+ − 507 (defun lao-compose-region (from to)
+ − 508 (interactive "r")
+ − 509 (save-restriction
+ − 510 (narrow-to-region from to)
+ − 511 (goto-char (point-min))
+ − 512 (with-category-table lao-category-table
+ − 513 (while (re-search-forward lao-composition-pattern nil t)
+ − 514 (compose-region (match-beginning 0) (point))))))
+ − 515
+ − 516 ;;
+ − 517 (provide 'lao-util)
+ − 518
+ − 519 ;;; lao-util.el ends here
