code.delx.au - gnu-emacs/blob - lisp/language/lao-util.el

   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 GNU Emacs.
   9
  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.
  14
  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.
  19
  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.
  24
  25 ;;; Commentary:
  26
  27 ;;; Code:
  28
  29 ;; Setting information of Thai characters.
  30
  31 (defconst lao-category-table (make-category-table))
  32 (define-category ?c "Lao consonant" lao-category-table)
  33 (define-category ?s "Lao semi-vowel" lao-category-table)
  34 (define-category ?v "Lao upper/lower vowel" lao-category-table)
  35 (define-category ?t "Lao tone" lao-category-table)
  36
  37 (let ((l '((?\e(1!\e(B consonant "LETTER KOR  KAI'" "CHICKEN")
  38            (?\e(1"\e(B consonant "LETTER KHOR KHAI'" "EGG")
  39            (?\e(1#\e(B invalid nil)
  40            (?\e(1$\e(B consonant "LETTER QHOR QHWARGN" "BUFFALO")
  41            (?\e(1%\e(B invalid nil)
  42            (?  invalid nil)
  43            (?\e(1'\e(B consonant "LETTER NGOR NGUU" "SNAKE")
  44            (?\e(1(\e(B consonant "LETTER JOR JUA" "BUDDHIST NOVICE")
  45            (?\e(1)\e(B invalid nil)
  46            (?\e(1*\e(B consonant "LETTER XOR X\"ARNG" "ELEPHANT")
  47            (?\e(1+\e(B invalid nil)
  48            (?\e(1,\e(B invalid nil)
  49            (?\e(1-\e(B consonant "LETTER YOR YUNG" "MOSQUITO")
  50            (?\e(1.\e(B invalid nil)
  51            (?\e(1.\e(B invalid nil)
  52            (?\e(1.\e(B invalid nil)
  53            (?\e(1.\e(B invalid nil)
  54            (?\e(1.\e(B invalid nil)
  55            (?\e(1.\e(B invalid nil)
  56            (?\e(14\e(B consonant "LETTER DOR DANG" "NOSE")
  57            (?\e(15\e(B consonant "LETTER TOR TAR" "EYE")
  58            (?\e(16\e(B consonant "LETTER THOR THUNG" "TO ASK,QUESTION")
  59            (?\e(17\e(B consonant "LETTER DHOR DHARM" "FLAG")
  60            (?\e(18\e(B invalid nil)
  61            (?\e(19\e(B consonant "LETTER NOR NOK" "BIRD")
  62            (?\e(1:\e(B consonant "LETTER BOR BED" "FISHHOOK")
  63            (?\e(1;\e(B consonant "LETTER POR PAR" "FISH")
  64            (?\e(1<\e(B consonant "LETTER HPOR HPER\"" "BEE")
  65            (?\e(1=\e(B consonant "LETTER FHOR FHAR" "WALL")
  66            (?\e(1>\e(B consonant "LETTER PHOR PHUU" "MOUNTAIN")
  67            (?\e(1?\e(B consonant "LETTER FOR FAI" "FIRE")
  68            (?\e(1@\e(B invalid nil)
  69            (?\e(1A\e(B consonant "LETTER MOR MAR\"" "HORSE")
  70            (?\e(1B\e(B consonant "LETTER GNOR GNAR" "MEDICINE")
  71            (?\e(1C\e(B consonant "LETTER ROR ROD" "CAR")
  72            (?\e(1D\e(B invalid nil)
  73            (?\e(1E\e(B consonant "LETTER LOR LIING" "MONKEY")
  74            (?\e(1F\e(B invalid nil)
  75            (?\e(1G\e(B consonant "LETTER WOR WII" "HAND FAN")
  76            (?\e(1H\e(B invalid nil)
  77            (?\e(1I\e(B invalid nil)
  78            (?\e(1J\e(B consonant "LETTER SOR SEA" "TIGER")
  79            (?\e(1K\e(B consonant "LETTER HHOR HHAI" "JAR")
  80            (?\e(1L\e(B invalid nil)
  81            (?\e(1M\e(B consonant "LETTER OR OOW" "TAKE")
  82            (?\e(1N\e(B consonant "LETTER HOR HEA" "BOAT")
  83            (?\e(1O\e(B special "ELLIPSIS")
  84            (?\e(1P\e(B vowel-base "VOWEL SIGN SARA A")
  85            (?\e(1Q\e(B vowel-upper "VOWEL SIGN MAI KAN")
  86            (?\e(1R\e(B vowel-base "VOWEL SIGN SARA AR")
  87            (?\e(1S\e(B vowel-base "VOWEL SIGN SARA AM")
  88            (?\e(1T\e(B vowel-upper "VOWEL SIGN SARA I")
  89            (?\e(1U\e(B vowel-upper "VOWEL SIGN SARA II")
  90            (?\e(1V\e(B vowel-upper "VOWEL SIGN SARA EU")
  91            (?\e(1W\e(B vowel-upper "VOWEL SIGN SARA UR")
  92            (?\e(1X\e(B vowel-lower "VOWEL SIGN SARA U")
  93            (?\e(1Y\e(B vowel-lower "VOWEL SIGN SARA UU")
  94            (?\e(1Z\e(B invalid nil)
  95            (?\e(1[\e(B vowel-upper "VOWEL SIGN MAI KONG")
  96            (?\e(1\\e(B semivowel-lower "SEMIVOWEL SIGN LO")
  97            (?\e(1]\e(B vowel-base "SEMIVOWEL SIGN SARA IA")
  98            (?\e(1^\e(B invalid nil)
  99            (?\e(1_\e(B invalid nil)
 100            (?\e(1`\e(B vowel-base "VOWEL SIGN SARA EE")
 101            (?\e(1a\e(B vowel-base "VOWEL SIGN SARA AA")
 102            (?\e(1b\e(B vowel-base "VOWEL SIGN SARA OO")
 103            (?\e(1c\e(B vowel-base "VOWEL SIGN SARA EI MAI MUAN\"")
 104            (?\e(1d\e(B vowel-base "VOWEL SIGN SARA AI MAI MAY")
 105            (?\e(1e\e(B invalid nil)
 106            (?\e(1f\e(B special "KO LA (REPETITION)")
 107            (?\e(1g\e(B invalid nil)
 108            (?\e(1h\e(B tone "TONE MAI EK")
 109            (?\e(1i\e(B tone "TONE MAI THO")
 110            (?\e(1j\e(B tone "TONE MAI TI")
 111            (?\e(1k\e(B tone "TONE MAI JADTAWAR")
 112            (?\e(1l\e(B tone "CANCELLATION MARK")
 113            (?\e(1m\e(B vowel-upper "VOWEL SIGN SARA OR")
 114            (?\e(1n\e(B invalid nil)
 115            (?\e(1o\e(B invalid nil)
 116            (?\e(1p\e(B special "DIGIT ZERO")
 117            (?\e(1q\e(B special "DIGIT ONE")
 118            (?\e(1r\e(B special "DIGIT TWO")
 119            (?\e(1s\e(B special "DIGIT THREE")
 120            (?\e(1t\e(B special "DIGIT FOUR")
 121            (?\e(1u\e(B special "DIGIT FIVE")
 122            (?\e(1v\e(B special "DIGIT SIX")
 123            (?\e(1w\e(B special "DIGIT SEVEN")
 124            (?\e(1x\e(B special "DIGIT EIGHT")
 125            (?\e(1y\e(B special "DIGIT NINE")
 126            (?\e(1z\e(B invalid nil)
 127            (?\e(1{\e(B invalid nil)
 128            (?\e(1|\e(B consonant "LETTER NHOR NHUU" "MOUSE")
 129            (?\e(1}\e(B consonant "LETTER MHOR MHAR" "DOG")
 130            (?\e(1~\e(B invalid nil)
 131            ))
 132       elm)
 133   (while l
 134     (setq elm (car l) l (cdr l))
 135     (let ((char (car elm))
 136           (ptype (nth 1 elm)))
 137       (cond ((eq ptype 'consonant)
 138              (modify-category-entry char ?c lao-category-table))
 139             ((memq ptype '(vowel-upper vowel-lower))
 140              (modify-category-entry char ?v lao-category-table))
 141             ((eq ptype 'semivowel-lower)
 142              (modify-category-entry char ?s lao-category-table))
 143             ((eq ptype 'tone)
 144              (modify-category-entry char ?t lao-category-table)))
 145       (put-char-code-property char 'phonetic-type ptype)
 146       (put-char-code-property char 'name (nth 2 elm))
 147       (put-char-code-property char 'meaning (nth 3 elm)))))
 148
 149 ;; The general composing rules are as follows:
 150 ;;
 151 ;;                          T
 152 ;;       V        T         V                  T
 153 ;; CV -> C, CT -> C, CVT -> C, Cv -> C, CvT -> C
 154 ;;                                   v         v
 155 ;;                             T
 156 ;;        V         T          V                   T
 157 ;; CsV -> C, CsT -> C, CsVT -> C, Csv -> C, CvT -> C
 158 ;;        s         s          s         s         s
 159 ;;                                       v         v
 160
 161
 162 ;; where C: consonant, V: vowel upper, v: vowel lower,
 163 ;;       T: tone mark, s: semivowel lower
 164
 165 (defvar lao-composition-pattern
 166   "\\cc\\(\\ct\\|\\cv\\ct?\\|\\cs\\(\\ct\\|\\cv\\ct?\\)?\\)"
 167   "Regular expression matching a Lao composite sequence.")
 168
 169 ;;;###autoload
 170 (defun lao-compose-string (str)
 171   (with-category-table lao-category-table
 172    (let ((idx 0))
 173      (while (setq idx (string-match lao-composition-pattern str idx))
 174        (compose-string str idx (match-end 0))
 175        (setq idx (match-end 0))))
 176    str))
 177
 178 ;;; LRT: Lao <-> Roman Transcription
 179
 180 ;; Upper vowels and tone-marks are put on the letter.
 181 ;; Semi-vowel-sign-lo and lower vowels are put under the letter.
 182
 183 (defconst lao-transcription-consonant-alist
 184   (sort '(;; single consonants
 185           ("k" . "\e(1!\e(B")
 186           ("kh" . "\e(1"\e(B")
 187           ("qh" . "\e(1$\e(B")
 188           ("ng" . "\e(1'\e(B")
 189           ("j" . "\e(1(\e(B")
 190           ("s" . "\e(1J\e(B")
 191           ("x" . "\e(1*\e(B")
 192           ("y" . "\e(1-\e(B")
 193           ("d" . "\e(14\e(B")
 194           ("t" . "\e(15\e(B")
 195           ("th" . "\e(16\e(B")
 196           ("dh" . "\e(17\e(B")
 197           ("n" . "\e(19\e(B")
 198           ("b" . "\e(1:\e(B")
 199           ("p" . "\e(1;\e(B")
 200           ("hp" . "\e(1<\e(B")
 201           ("fh" . "\e(1=\e(B")
 202           ("ph" . "\e(1>\e(B")
 203           ("f" . "\e(1?\e(B")
 204           ("m" . "\e(1A\e(B")
 205           ("gn" . "\e(1B\e(B")
 206           ("l" . "\e(1E\e(B")
 207           ("r" . "\e(1C\e(B")
 208           ("v" . "\e(1G\e(B")
 209           ("w" . "\e(1G\e(B")
 210           ("hh" . "\e(1K\e(B")
 211           ("O" . "\e(1M\e(B")
 212           ("h" . "\e(1N\e(B")
 213           ("nh" . "\e(1|\e(B")
 214           ("mh" . "\e(1}\e(B")
 215           ("lh" . ["\e(1K\\e(B"])
 216           ;; double consonants
 217           ("ngh" . ["\e(1K'\e(B"])
 218           ("yh" . ["\e(1K]\e(B"])
 219           ("wh" . ["\e(1KG\e(B"])
 220           ("hl" . ["\e(1KE\e(B"])
 221           ("hy" . ["\e(1K-\e(B"])
 222           ("hn" . ["\e(1K9\e(B"])
 223           ("hm" . ["\e(1KA\e(B"])
 224           )
 225         (function (lambda (x y) (> (length (car x)) (length (car y)))))))
 226
 227 (defconst lao-transcription-semi-vowel-alist
 228   '(("r" . "\e(1\\e(B")))
 229
 230 (defconst lao-transcription-vowel-alist
 231   (sort '(("a" . "\e(1P\e(B")
 232           ("ar" . "\e(1R\e(B")
 233           ("i" . "\e(1T\e(B")
 234           ("ii" . "\e(1U\e(B")
 235           ("eu" . "\e(1V\e(B")
 236           ("ur" . "\e(1W\e(B")
 237           ("u" . "\e(1X\e(B")
 238           ("uu" . "\e(1Y\e(B")
 239           ("e" . ["\e(1`P\e(B"])
 240           ("ee" . "\e(1`\e(B")
 241           ("ae" . ["\e(1aP\e(B"])
 242           ("aa" . "\e(1a\e(B")
 243           ("o" . ["\e(1bP\e(B"])
 244           ("oo" . "\e(1b\e(B")
 245           ("oe" . ["\e(1`RP\e(B"])
 246           ("or" . "\e(1m\e(B")
 247           ("er" . ["\e(1`T\e(B"])
 248           ("ir" . ["\e(1`U\e(B"])
 249           ("ua" . ["\e(1[GP\e(B"])
 250           ("uaa" . ["\e(1[G\e(B"])
 251           ("ie" . ["\e(1`Q]P\e(B"])
 252           ("ia" . ["\e(1`Q]\e(B"])
 253           ("ea" . ["\e(1`VM\e(B"])
 254           ("eaa" . ["\e(1`WM\e(B"])
 255           ("ai" . "\e(1d\e(B")
 256           ("ei" . "\e(1c\e(B")
 257           ("ao" . ["\e(1`[R\e(B"])
 258           ("aM" . "\e(1S\e(B"))
 259         (function (lambda (x y) (> (length (car x)) (length (car y)))))))
 260
 261 ;; Maa-sakod is put at the tail.
 262 (defconst lao-transcription-maa-sakod-alist
 263   '(("k" . "\e(1!\e(B")
 264     ("g" . "\e(1'\e(B")
 265     ("y" . "\e(1-\e(B")
 266     ("d" . "\e(14\e(B")
 267     ("n" . "\e(19\e(B")
 268     ("b" . "\e(1:\e(B")
 269     ("m" . "\e(1A\e(B")
 270     ("v" . "\e(1G\e(B")
 271     ("w" . "\e(1G\e(B")
 272     ))
 273
 274 (defconst lao-transcription-tone-alist
 275   '(("'" . "\e(1h\e(B")
 276     ("\"" . "\e(1i\e(B")
 277     ("^" . "\e(1j\e(B")
 278     ("+" . "\e(1k\e(B")
 279     ("~" . "\e(1l\e(B")))
 280
 281 (defconst lao-transcription-punctuation-alist
 282   '(("\\0" . "\e(1p\e(B")
 283     ("\\1" . "\e(1q\e(B")
 284     ("\\2" . "\e(1r\e(B")
 285     ("\\3" . "\e(1s\e(B")
 286     ("\\4" . "\e(1t\e(B")
 287     ("\\5" . "\e(1u\e(B")
 288     ("\\6" . "\e(1v\e(B")
 289     ("\\7" . "\e(1w\e(B")
 290     ("\\8" . "\e(1x\e(B")
 291     ("\\9" . "\e(1y\e(B")
 292     ("\\\\" . "\e(1f\e(B")
 293     ("\\$" . "\e(1O\e(B")))
 294
 295 (defconst lao-transcription-pattern
 296   (concat
 297    "\\("
 298    (mapconcat 'car lao-transcription-consonant-alist "\\|")
 299    "\\)\\("
 300    (mapconcat 'car lao-transcription-semi-vowel-alist "\\|")
 301    "\\)?\\(\\("
 302    (mapconcat 'car lao-transcription-vowel-alist "\\|")
 303    "\\)\\("
 304    (mapconcat 'car lao-transcription-maa-sakod-alist "\\|")
 305    "\\)?\\("
 306    (mapconcat (lambda (x) (regexp-quote (car x)))
 307               lao-transcription-tone-alist "\\|")
 308    "\\)?\\)?\\|"
 309    (mapconcat (lambda (x) (regexp-quote (car x)))
 310               lao-transcription-punctuation-alist "\\|")
 311    )
 312   "Regexp of Roman transcription pattern for one Lao syllable.")
 313
 314 (defconst lao-transcription-pattern
 315   (concat
 316    "\\("
 317    (regexp-opt (mapcar 'car lao-transcription-consonant-alist))
 318    "\\)\\("
 319    (regexp-opt (mapcar 'car lao-transcription-semi-vowel-alist))
 320    "\\)?\\(\\("
 321    (regexp-opt (mapcar 'car lao-transcription-vowel-alist))
 322    "\\)\\("
 323    (regexp-opt (mapcar 'car lao-transcription-maa-sakod-alist))
 324    "\\)?\\("
 325    (regexp-opt (mapcar 'car lao-transcription-tone-alist))
 326    "\\)?\\)?\\|"
 327    (regexp-opt (mapcar 'car lao-transcription-punctuation-alist))
 328    )
 329   "Regexp of Roman transcription pattern for one Lao syllable.")
 330
 331 (defconst lao-vowel-reordering-rule
 332   '(("\e(1P\e(B" (0 ?\e(1P\e(B) (0 ?\e(1Q\e(B))
 333     ("\e(1R\e(B" (0 ?\e(1R\e(B))
 334     ("\e(1T\e(B" (0 ?\e(1U\e(B))
 335     ("\e(1U\e(B" (0 ?\e(1U\e(B))
 336     ("\e(1V\e(B" (0 ?\e(1V\e(B))
 337     ("\e(1W\e(B" (0 ?\e(1W\e(B))
 338     ("\e(1X\e(B" (0 ?\e(1X\e(B))
 339     ("\e(1Y\e(B" (0 ?\e(1Y\e(B))
 340     ("\e(1`P\e(B" (?\e(1`\e(B 0 ?\e(1P\e(B) (?\e(1`\e(B 0 ?\e(1Q\e(B))
 341     ("\e(1`\e(B" (?\e(1`\e(B 0))
 342     ("\e(1aP\e(B" (?\e(1a\e(B 0 ?\e(1P\e(B) (?\e(1a\e(B 0 ?\e(1Q\e(B))
 343     ("\e(1a\e(B" (?\e(1a\e(B 0))
 344     ("\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))
 345     ("\e(1b\e(B" (?\e(1b\e(B 0))
 346     ("\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))
 347     ("\e(1m\e(B" (0 ?\e(1m\e(B) (0 ?\e(1M\e(B))
 348     ("\e(1`T\e(B" (?\e(1`\e(B 0 ?\e(1T\e(B))
 349     ("\e(1`U\e(B" (?\e(1`\e(B 0 ?\e(1U\e(B))
 350     ("\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))
 351     ("\e(1[G\e(B" (0 ?\e(1[\e(B ?\e(1G\e(B) (0 ?\e(1G\e(B))
 352     ("\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))
 353     ("\e(1`Q]\e(B" (?\e(1`\e(B 0 ?\e(1Q\e(B ?\e(1]\e(B) (0 ?\e(1]\e(B))
 354     ("\e(1`VM\e(B" (?\e(1`\e(B 0 ?\e(1V\e(B ?\e(1M\e(B))
 355     ("\e(1`WM\e(B" (?\e(1`\e(B 0 ?\e(1W\e(B ?\e(1M\e(B))
 356     ("\e(1d\e(B" (?\e(1d\e(B 0))
 357     ("\e(1c\e(B" (?\e(1c\e(B 0))
 358     ("\e(1`[R\e(B" (?\e(1`\e(B 0 ?\e(1[\e(B ?\e(1R\e(B))
 359     ("\e(1S\e(B" (0 ?\e(1S\e(B)))
 360   "Alist of Lao vowel string vs the corresponding re-ordering rule.
 361 Each element has this form:
 362         (VOWEL NO-MAA-SAKOD-RULE WITH-MAA-SAKOD-RULE (MAA-SAKOD-0 RULE-0) ...)
 363
 364 VOWEL is a vowel string (e.g. \"\e(1`Q]P\e(B\").
 365
 366 NO-MAA-SAKOD-RULE is a rule to re-order and modify VOWEL following a
 367 consonant.  It is a list vowel characters or 0.  The element 0
 368 indicate the place to embed a consonant.
 369
 370 Optional WITH-MAA-SAKOD-RULE is a rule to re-order and modify VOWEL
 371 follwoing a consonant and preceding a maa-sakod character.  If it is
 372 nil, NO-MAA-SAKOD-RULE is used.  The maa-sakod character is alwasy
 373 appended at the tail.
 374
 375 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
 376 string following a consonant `\e(1!\e(B' should be re-ordered as \"\e(1`!WM\e(B\".
 377
 378 Optional (MAA-SAKOD-n RULE-n) are rules specially applied to maa-sakod
 379 character MAA-SAKOD-n.")
 380
 381 ;;;###autoload
 382 (defun lao-transcribe-single-roman-syllable-to-lao (from to &optional str)
 383   "Transcribe a Romanized Lao syllable in the region FROM and TO to Lao string.
 384 Only the first syllable is transcribed.
 385 The value has the form: (START END LAO-STRING), where
 386 START and END are the beggining and end positions of the Roman Lao syllable,
 387 LAO-STRING is the Lao character transcription of it.
 388
 389 Optional 3rd arg STR, if non-nil, is a string to search for Roman Lao
 390 syllable.  In that case, FROM and TO are indexes to STR."
 391   (if str
 392       (if (setq from (string-match lao-transcription-pattern str from))
 393           (progn
 394             (if (>= from to)
 395                 (setq from nil)
 396               (setq to (match-end 0)))))
 397     (save-excursion
 398       (goto-char from)
 399       (if (setq to (re-search-forward lao-transcription-pattern to t))
 400           (setq from (match-beginning 0))
 401         (setq from nil))))
 402   (if from
 403       (let* ((consonant (match-string 1 str))
 404              (semivowel (match-string 3 str))
 405              (vowel (match-string 5 str))
 406              (maa-sakod (match-string 8 str))
 407              (tone (match-string 9 str))
 408              lao-consonant lao-semivowel lao-vowel lao-maa-sakod lao-tone
 409              clen cidx)
 410         (setq to (match-end 0))
 411         (if (not consonant)
 412             (setq str (cdr (assoc (match-string 0 str)
 413                                   lao-transcription-punctuation-alist)))
 414           (setq lao-consonant
 415                 (cdr (assoc consonant lao-transcription-consonant-alist)))
 416           (if (vectorp lao-consonant)
 417               (setq lao-consonant (aref lao-consonant 0)))
 418           (setq clen (length lao-consonant))
 419           (if semivowel
 420               ;; Include semivowel in STR.
 421               (setq lao-semivowel
 422                     (cdr (assoc semivowel lao-transcription-semi-vowel-alist))
 423                     str (if (= clen 1)
 424                             (concat lao-consonant lao-semivowel)
 425                           (concat (substring lao-consonant 0 1) lao-semivowel
 426                                   (substring lao-consonant 1))))
 427             (setq str lao-consonant))
 428           (if vowel
 429               (let (rule)
 430                 (setq lao-vowel
 431                       (cdr (assoc vowel lao-transcription-vowel-alist)))
 432                 (if (vectorp lao-vowel)
 433                     (setq lao-vowel (aref lao-vowel 0)))
 434                 (setq rule (assoc lao-vowel lao-vowel-reordering-rule))
 435                 (if (null maa-sakod)
 436                     (setq rule (nth 1 rule))
 437                   (setq lao-maa-sakod
 438                         (cdr (assoc maa-sakod lao-transcription-maa-sakod-alist))
 439                         rule
 440                         (or (cdr (assq (aref lao-maa-sakod 0) (nthcdr 2 rule)))
 441                             (nth 2 rule)
 442                             (nth 1 rule))))
 443                 (or rule
 444                     (error "Lao vowel %S has no re-ordering rule" lao-vowel))
 445                 (setq lao-consonant str str "")
 446                 (while rule
 447                   (if (= (car rule) 0)
 448                       (setq str (concat str lao-consonant)
 449                             cidx (length str))
 450                     (setq str (concat str (list (car rule)))))
 451                   (setq rule (cdr rule)))
 452                 (or cidx
 453                     (error "Lao vowel %S has malformed re-ordering rule" vowel))
 454                 ;; Set CIDX to after upper or lower vowel if any.
 455                 (let ((len (length str)))
 456                   (while (and (< cidx len)
 457                               (memq (get-char-code-property (aref str cidx)
 458                                                             'phonetic-type)
 459                                     '(vowel-lower vowel-upper)))
 460                     (setq cidx (1+ cidx))))
 461                 (if lao-maa-sakod
 462                     (setq str (concat str lao-maa-sakod)))
 463                 (if tone
 464                     (setq lao-tone
 465                           (cdr (assoc tone lao-transcription-tone-alist))
 466                           str (concat (substring str 0 cidx) lao-tone
 467                                       (substring str cidx)))))))
 468         (list from to (lao-compose-string str)))))
 469
 470 ;;;###autoload
 471 (defun lao-transcribe-roman-to-lao-string (str)
 472   "Transcribe Romanized Lao string STR to Lao character string."
 473   (let ((from 0)
 474         (to (length str))
 475         (lao-str "")
 476         val)
 477     (while (setq val (lao-transcribe-single-roman-syllable-to-lao from to str))
 478       (let ((start (car val))
 479             (end (nth 1 val))
 480             (lao (nth 2 val)))
 481         (if (> start from)
 482             (setq lao-str (concat lao-str (substring str from start) lao))
 483           (setq lao-str (concat lao-str lao)))
 484         (setq from end)))
 485     (if (< from to)
 486         (concat lao-str (substring str from to))
 487       lao-str)))
 488
 489 ;;;###autoload
 490 (defun lao-composition-function (from to pattern &optional string)
 491   "Compose Lao text in the region FROM and TO.
 492 The text matches the regular expression PATTERN.
 493 Optional 4th argument STRING, if non-nil, is a string containing text
 494 to compose.
 495
 496 The return value is number of composed characters."
 497   (if (< (1+ from) to)
 498       (prog1 (- to from)
 499         (if string
 500             (compose-string string from to)
 501           (compose-region from to))
 502         (- to from))))
 503
 504 ;;;###autoload
 505 (defun lao-compose-region (from to)
 506   (interactive "r")
 507   (save-restriction
 508     (narrow-to-region from to)
 509     (goto-char (point-min))
 510     (with-category-table lao-category-table
 511       (while (re-search-forward lao-composition-pattern nil t)
 512         (compose-region (match-beginning 0) (point))))))
 513
 514 ;;
 515 (provide 'lao-util)
 516
 517 ;;; lao-util.el ends here