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 ;; Copyright (C) 2001 Free Software Foundation, Inc.
   6
   7 ;; Keywords: multilingual, Lao, i18n
   8
   9 ;; This file is part of GNU Emacs.
  10
  11 ;; GNU Emacs is free software; you can redistribute it and/or modify
  12 ;; it under the terms of the GNU General Public License as published by
  13 ;; the Free Software Foundation; either version 2, or (at your option)
  14 ;; any later version.
  15
  16 ;; GNU Emacs is distributed in the hope that it will be useful,
  17 ;; but WITHOUT ANY WARRANTY; without even the implied warranty of
  18 ;; MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
  19 ;; GNU General Public License for more details.
  20
  21 ;; You should have received a copy of the GNU General Public License
  22 ;; along with GNU Emacs; see the file COPYING.  If not, write to the
  23 ;; Free Software Foundation, Inc., 59 Temple Place - Suite 330,
  24 ;; Boston, MA 02111-1307, USA.
  25
  26 ;;; Commentary:
  27
  28 ;;; Code:
  29
  30 ;; Setting information of Thai characters.
  31
  32 (defconst lao-category-table (make-category-table))
  33 (define-category ?c "Lao consonant" lao-category-table)
  34 (define-category ?s "Lao semi-vowel" lao-category-table)
  35 (define-category ?v "Lao upper/lower vowel" lao-category-table)
  36 (define-category ?t "Lao tone" lao-category-table)
  37
  38 ;; Fixme: Should `invalid' entries below be preserved?  (See also
  39 ;; thai-util.el.)
  40 (let ((l '((?\e(1!\e(B consonant "LETTER KOR  KAI'" "CHICKEN")
  41            (?\e(1"\e(B consonant "LETTER KHOR KHAI'" "EGG")
  42            (?\e(1#\e(B invalid nil)
  43            (?\e(1$\e(B consonant "LETTER QHOR QHWARGN" "BUFFALO")
  44            (?\e(1%\e(B invalid nil)
  45            (?  invalid nil)
  46            (?\e(1'\e(B consonant "LETTER NGOR NGUU" "SNAKE")
  47            (?\e(1(\e(B consonant "LETTER JOR JUA" "BUDDHIST NOVICE")
  48            (?\e(1)\e(B invalid nil)
  49            (?\e(1*\e(B consonant "LETTER XOR X\"ARNG" "ELEPHANT")
  50            (?\e(1+\e(B invalid nil)
  51            (?\e(1,\e(B invalid nil)
  52            (?\e(1-\e(B consonant "LETTER YOR YUNG" "MOSQUITO")
  53            (?\e(1.\e(B invalid nil)
  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(14\e(B consonant "LETTER DOR DANG" "NOSE")
  60            (?\e(15\e(B consonant "LETTER TOR TAR" "EYE")
  61            (?\e(16\e(B consonant "LETTER THOR THUNG" "TO ASK,QUESTION")
  62            (?\e(17\e(B consonant "LETTER DHOR DHARM" "FLAG")
  63            (?\e(18\e(B invalid nil)
  64            (?\e(19\e(B consonant "LETTER NOR NOK" "BIRD")
  65            (?\e(1:\e(B consonant "LETTER BOR BED" "FISHHOOK")
  66            (?\e(1;\e(B consonant "LETTER POR PAR" "FISH")
  67            (?\e(1<\e(B consonant "LETTER HPOR HPER\"" "BEE")
  68            (?\e(1=\e(B consonant "LETTER FHOR FHAR" "WALL")
  69            (?\e(1>\e(B consonant "LETTER PHOR PHUU" "MOUNTAIN")
  70            (?\e(1?\e(B consonant "LETTER FOR FAI" "FIRE")
  71            (?\e(1@\e(B invalid nil)
  72            (?\e(1A\e(B consonant "LETTER MOR MAR\"" "HORSE")
  73            (?\e(1B\e(B consonant "LETTER GNOR GNAR" "MEDICINE")
  74            (?\e(1C\e(B consonant "LETTER ROR ROD" "CAR")
  75            (?\e(1D\e(B invalid nil)
  76            (?\e(1E\e(B consonant "LETTER LOR LIING" "MONKEY")
  77            (?\e(1F\e(B invalid nil)
  78            (?\e(1G\e(B consonant "LETTER WOR WII" "HAND FAN")
  79            (?\e(1H\e(B invalid nil)
  80            (?\e(1I\e(B invalid nil)
  81            (?\e(1J\e(B consonant "LETTER SOR SEA" "TIGER")
  82            (?\e(1K\e(B consonant "LETTER HHOR HHAI" "JAR")
  83            (?\e(1L\e(B invalid nil)
  84            (?\e(1M\e(B consonant "LETTER OR OOW" "TAKE")
  85            (?\e(1N\e(B consonant "LETTER HOR HEA" "BOAT")
  86            (?\e(1O\e(B special "ELLIPSIS")
  87            (?\e(1P\e(B vowel-base "VOWEL SIGN SARA A")
  88            (?\e(1Q\e(B vowel-upper "VOWEL SIGN MAI KAN")
  89            (?\e(1R\e(B vowel-base "VOWEL SIGN SARA AR")
  90            (?\e(1S\e(B vowel-base "VOWEL SIGN SARA AM")
  91            (?\e(1T\e(B vowel-upper "VOWEL SIGN SARA I")
  92            (?\e(1U\e(B vowel-upper "VOWEL SIGN SARA II")
  93            (?\e(1V\e(B vowel-upper "VOWEL SIGN SARA EU")
  94            (?\e(1W\e(B vowel-upper "VOWEL SIGN SARA UR")
  95            (?\e(1X\e(B vowel-lower "VOWEL SIGN SARA U")
  96            (?\e(1Y\e(B vowel-lower "VOWEL SIGN SARA UU")
  97            (?\e(1Z\e(B invalid nil)
  98            (?\e(1[\e(B vowel-upper "VOWEL SIGN MAI KONG")
  99            (?\e(1\\e(B semivowel-lower "SEMIVOWEL SIGN LO")
 100            (?\e(1]\e(B vowel-base "SEMIVOWEL SIGN SARA IA")
 101            (?\e(1^\e(B invalid nil)
 102            (?\e(1_\e(B invalid nil)
 103            (?\e(1`\e(B vowel-base "VOWEL SIGN SARA EE")
 104            (?\e(1a\e(B vowel-base "VOWEL SIGN SARA AA")
 105            (?\e(1b\e(B vowel-base "VOWEL SIGN SARA OO")
 106            (?\e(1c\e(B vowel-base "VOWEL SIGN SARA EI MAI MUAN\"")
 107            (?\e(1d\e(B vowel-base "VOWEL SIGN SARA AI MAI MAY")
 108            (?\e(1e\e(B invalid nil)
 109            (?\e(1f\e(B special "KO LA (REPETITION)")
 110            (?\e(1g\e(B invalid nil)
 111            (?\e(1h\e(B tone "TONE MAI EK")
 112            (?\e(1i\e(B tone "TONE MAI THO")
 113            (?\e(1j\e(B tone "TONE MAI TI")
 114            (?\e(1k\e(B tone "TONE MAI JADTAWAR")
 115            (?\e(1l\e(B tone "CANCELLATION MARK")
 116            (?\e(1m\e(B vowel-upper "VOWEL SIGN SARA OR")
 117            (?\e(1n\e(B invalid nil)
 118            (?\e(1o\e(B invalid nil)
 119            (?\e(1p\e(B special "DIGIT ZERO")
 120            (?\e(1q\e(B special "DIGIT ONE")
 121            (?\e(1r\e(B special "DIGIT TWO")
 122            (?\e(1s\e(B special "DIGIT THREE")
 123            (?\e(1t\e(B special "DIGIT FOUR")
 124            (?\e(1u\e(B special "DIGIT FIVE")
 125            (?\e(1v\e(B special "DIGIT SIX")
 126            (?\e(1w\e(B special "DIGIT SEVEN")
 127            (?\e(1x\e(B special "DIGIT EIGHT")
 128            (?\e(1y\e(B special "DIGIT NINE")
 129            (?\e(1z\e(B invalid nil)
 130            (?\e(1{\e(B invalid nil)
 131            (?\e(1|\e(B consonant "LETTER NHOR NHUU" "MOUSE")
 132            (?\e(1}\e(B consonant "LETTER MHOR MHAR" "DOG")
 133            (?\e(1~\e(B invalid nil)))
 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, CsvT -> 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" . "\e(1!\e(B")
 188           ("kh" . "\e(1"\e(B")
 189           ("qh" . "\e(1$\e(B")
 190           ("ng" . "\e(1'\e(B")
 191           ("j" . "\e(1(\e(B")
 192           ("s" . "\e(1J\e(B")
 193           ("x" . "\e(1*\e(B")
 194           ("y" . "\e(1-\e(B")
 195           ("d" . "\e(14\e(B")
 196           ("t" . "\e(15\e(B")
 197           ("th" . "\e(16\e(B")
 198           ("dh" . "\e(17\e(B")
 199           ("n" . "\e(19\e(B")
 200           ("b" . "\e(1:\e(B")
 201           ("p" . "\e(1;\e(B")
 202           ("hp" . "\e(1<\e(B")
 203           ("fh" . "\e(1=\e(B")
 204           ("ph" . "\e(1>\e(B")
 205           ("f" . "\e(1?\e(B")
 206           ("m" . "\e(1A\e(B")
 207           ("gn" . "\e(1B\e(B")
 208           ("l" . "\e(1E\e(B")
 209           ("r" . "\e(1C\e(B")
 210           ("v" . "\e(1G\e(B")
 211           ("w" . "\e(1G\e(B")
 212           ("hh" . "\e(1K\e(B")
 213           ("O" . "\e(1M\e(B")
 214           ("h" . "\e(1N\e(B")
 215           ("nh" . "\e(1|\e(B")
 216           ("mh" . "\e(1}\e(B")
 217           ("lh" . ["\e(1K\\e(B"])
 218           ;; double consonants
 219           ("ngh" . ["\e(1K'\e(B"])
 220           ("yh" . ["\e(1K]\e(B"])
 221           ("wh" . ["\e(1KG\e(B"])
 222           ("hl" . ["\e(1KE\e(B"])
 223           ("hy" . ["\e(1K-\e(B"])
 224           ("hn" . ["\e(1K9\e(B"])
 225           ("hm" . ["\e(1KA\e(B"])
 226           )
 227         (function (lambda (x y) (> (length (car x)) (length (car y)))))))
 228
 229 (defconst lao-transcription-semi-vowel-alist
 230   '(("r" . "\e(1\\e(B")))
 231
 232 (defconst lao-transcription-vowel-alist
 233   (sort '(("a" . "\e(1P\e(B")
 234           ("ar" . "\e(1R\e(B")
 235           ("i" . "\e(1T\e(B")
 236           ("ii" . "\e(1U\e(B")
 237           ("eu" . "\e(1V\e(B")
 238           ("ur" . "\e(1W\e(B")
 239           ("u" . "\e(1X\e(B")
 240           ("uu" . "\e(1Y\e(B")
 241           ("e" . ["\e(1`P\e(B"])
 242           ("ee" . "\e(1`\e(B")
 243           ("ae" . ["\e(1aP\e(B"])
 244           ("aa" . "\e(1a\e(B")
 245           ("o" . ["\e(1bP\e(B"])
 246           ("oo" . "\e(1b\e(B")
 247           ("oe" . ["\e(1`RP\e(B"])
 248           ("or" . "\e(1m\e(B")
 249           ("er" . ["\e(1`T\e(B"])
 250           ("ir" . ["\e(1`U\e(B"])
 251           ("ua" . ["\e(1[GP\e(B"])
 252           ("uaa" . ["\e(1[G\e(B"])
 253           ("ie" . ["\e(1`Q]P\e(B"])
 254           ("ia" . ["\e(1`Q]\e(B"])
 255           ("ea" . ["\e(1`VM\e(B"])
 256           ("eaa" . ["\e(1`WM\e(B"])
 257           ("ai" . "\e(1d\e(B")
 258           ("ei" . "\e(1c\e(B")
 259           ("ao" . ["\e(1`[R\e(B"])
 260           ("aM" . "\e(1S\e(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" . "\e(1!\e(B")
 266     ("g" . "\e(1'\e(B")
 267     ("y" . "\e(1-\e(B")
 268     ("d" . "\e(14\e(B")
 269     ("n" . "\e(19\e(B")
 270     ("b" . "\e(1:\e(B")
 271     ("m" . "\e(1A\e(B")
 272     ("v" . "\e(1G\e(B")
 273     ("w" . "\e(1G\e(B")
 274     ))
 275
 276 (defconst lao-transcription-tone-alist
 277   '(("'" . "\e(1h\e(B")
 278     ("\"" . "\e(1i\e(B")
 279     ("^" . "\e(1j\e(B")
 280     ("+" . "\e(1k\e(B")
 281     ("~" . "\e(1l\e(B")))
 282
 283 (defconst lao-transcription-punctuation-alist
 284   '(("\\0" . "\e(1p\e(B")
 285     ("\\1" . "\e(1q\e(B")
 286     ("\\2" . "\e(1r\e(B")
 287     ("\\3" . "\e(1s\e(B")
 288     ("\\4" . "\e(1t\e(B")
 289     ("\\5" . "\e(1u\e(B")
 290     ("\\6" . "\e(1v\e(B")
 291     ("\\7" . "\e(1w\e(B")
 292     ("\\8" . "\e(1x\e(B")
 293     ("\\9" . "\e(1y\e(B")
 294     ("\\\\" . "\e(1f\e(B")
 295     ("\\$" . "\e(1O\e(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   '(("\e(1P\e(B" (0 ?\e(1P\e(B) (0 ?\e(1Q\e(B))
 335     ("\e(1R\e(B" (0 ?\e(1R\e(B))
 336     ("\e(1T\e(B" (0 ?\e(1U\e(B))
 337     ("\e(1U\e(B" (0 ?\e(1U\e(B))
 338     ("\e(1V\e(B" (0 ?\e(1V\e(B))
 339     ("\e(1W\e(B" (0 ?\e(1W\e(B))
 340     ("\e(1X\e(B" (0 ?\e(1X\e(B))
 341     ("\e(1Y\e(B" (0 ?\e(1Y\e(B))
 342     ("\e(1`P\e(B" (?\e(1`\e(B 0 ?\e(1P\e(B) (?\e(1`\e(B 0 ?\e(1Q\e(B))
 343     ("\e(1`\e(B" (?\e(1`\e(B 0))
 344     ("\e(1aP\e(B" (?\e(1a\e(B 0 ?\e(1P\e(B) (?\e(1a\e(B 0 ?\e(1Q\e(B))
 345     ("\e(1a\e(B" (?\e(1a\e(B 0))
 346     ("\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))
 347     ("\e(1b\e(B" (?\e(1b\e(B 0))
 348     ("\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))
 349     ("\e(1m\e(B" (0 ?\e(1m\e(B) (0 ?\e(1M\e(B))
 350     ("\e(1`T\e(B" (?\e(1`\e(B 0 ?\e(1T\e(B))
 351     ("\e(1`U\e(B" (?\e(1`\e(B 0 ?\e(1U\e(B))
 352     ("\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))
 353     ("\e(1[G\e(B" (0 ?\e(1[\e(B ?\e(1G\e(B) (0 ?\e(1G\e(B))
 354     ("\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))
 355     ("\e(1`Q]\e(B" (?\e(1`\e(B 0 ?\e(1Q\e(B ?\e(1]\e(B) (0 ?\e(1]\e(B))
 356     ("\e(1`VM\e(B" (?\e(1`\e(B 0 ?\e(1V\e(B ?\e(1M\e(B))
 357     ("\e(1`WM\e(B" (?\e(1`\e(B 0 ?\e(1W\e(B ?\e(1M\e(B))
 358     ("\e(1d\e(B" (?\e(1d\e(B 0))
 359     ("\e(1c\e(B" (?\e(1c\e(B 0))
 360     ("\e(1`[R\e(B" (?\e(1`\e(B 0 ?\e(1[\e(B ?\e(1R\e(B))
 361     ("\e(1S\e(B" (0 ?\e(1S\e(B))
 362
 363     ;; Unicode equivalents
 364     ("\e(1P\e(B" (0 ?\e(1P\e(B) (0 ?\e(1Q\e(B))
 365     ("\e(1R\e(B" (0 ?\e(1R\e(B))
 366     ("\e(1T\e(B" (0 ?\e(1U\e(B))
 367     ("\e(1U\e(B" (0 ?\e(1U\e(B))
 368     ("\e(1V\e(B" (0 ?\e(1V\e(B))
 369     ("\e(1W\e(B" (0 ?\e(1W\e(B))
 370     ("\e(1X\e(B" (0 ?\e(1X\e(B))
 371     ("\e(1Y\e(B" (0 ?\e(1Y\e(B))
 372     ("\e(1`P\e(B" (?\e(1`\e(B 0 ?\e(1P\e(B) (?\e(1`\e(B 0 ?\e(1Q\e(B))
 373     ("\e(1`\e(B" (?\e(1`\e(B 0))
 374     ("\e(1aP\e(B" (?\e(1a\e(B 0 ?\e(1P\e(B) (?\e(1a\e(B 0 ?\e(1Q\e(B))
 375     ("\e(1a\e(B" (?\e(1a\e(B 0))
 376     ("\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))
 377     ("\e(1b\e(B" (?\e(1b\e(B 0))
 378     ("\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))
 379     ("\e(1m\e(B" (0 ?\e(1m\e(B) (0 ?\e(1M\e(B))
 380     ("\e(1`T\e(B" (?\e(1`\e(B 0 ?\e(1T\e(B))
 381     ("\e(1`U\e(B" (?\e(1`\e(B 0 ?\e(1U\e(B))
 382     ("\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))
 383     ("\e(1[G\e(B" (0 ?\e(1[\e(B ?\e(1G\e(B) (0 ?\e(1G\e(B))
 384     ("\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))
 385     ("\e(1`Q]\e(B" (?\e(1`\e(B 0 ?\e(1Q\e(B ?\e(1]\e(B) (0 ?\e(1]\e(B))
 386     ("\e(1`VM\e(B" (?\e(1`\e(B 0 ?\e(1V\e(B ?\e(1M\e(B))
 387     ("\e(1`WM\e(B" (?\e(1`\e(B 0 ?\e(1W\e(B ?\e(1M\e(B))
 388     ("\e(1d\e(B" (?\e(1d\e(B 0))
 389     ("\e(1c\e(B" (?\e(1c\e(B 0))
 390     ("\e(1`[R\e(B" (?\e(1`\e(B 0 ?\e(1[\e(B ?\e(1R\e(B))
 391     ("\e(1S\e(B" (0 ?\e(1S\e(B)))
 392   "Alist of Lao vowel string vs the corresponding re-ordering rule.
 393 Each element has this form:
 394         (VOWEL NO-MAA-SAKOD-RULE WITH-MAA-SAKOD-RULE (MAA-SAKOD-0 RULE-0) ...)
 395
 396 VOWEL is a vowel string (e.g. \"\e(1`Q]P\e(B\").
 397
 398 NO-MAA-SAKOD-RULE is a rule to re-order and modify VOWEL following a
 399 consonant.  It is a list vowel characters or 0.  The element 0
 400 indicate the place to embed a consonant.
 401
 402 Optional WITH-MAA-SAKOD-RULE is a rule to re-order and modify VOWEL
 403 follwoing a consonant and preceding a maa-sakod character.  If it is
 404 nil, NO-MAA-SAKOD-RULE is used.  The maa-sakod character is alwasy
 405 appended at the tail.
 406
 407 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
 408 string following a consonant `\e(1!\e(B' should be re-ordered as \"\e(1`!WM\e(B\".
 409
 410 Optional (MAA-SAKOD-n RULE-n) are rules specially applied to maa-sakod
 411 character MAA-SAKOD-n.")
 412
 413 ;;;###autoload
 414 (defun lao-transcribe-single-roman-syllable-to-lao (from to &optional str)
 415   "Transcribe a Romanized Lao syllable in the region FROM and TO to Lao string.
 416 Only the first syllable is transcribed.
 417 The value has the form: (START END LAO-STRING), where
 418 START and END are the beggining and end positions of the Roman Lao syllable,
 419 LAO-STRING is the Lao character transcription of it.
 420
 421 Optional 3rd arg STR, if non-nil, is a string to search for Roman Lao
 422 syllable.  In that case, FROM and TO are indexes to STR."
 423   (if str
 424       (if (setq from (string-match lao-transcription-pattern str from))
 425           (progn
 426             (if (>= from to)
 427                 (setq from nil)
 428               (setq to (match-end 0)))))
 429     (save-excursion
 430       (goto-char from)
 431       (if (setq to (re-search-forward lao-transcription-pattern to t))
 432           (setq from (match-beginning 0))
 433         (setq from nil))))
 434   (if from
 435       (let* ((consonant (match-string 1 str))
 436              (semivowel (match-string 3 str))
 437              (vowel (match-string 5 str))
 438              (maa-sakod (match-string 8 str))
 439              (tone (match-string 9 str))
 440              lao-consonant lao-semivowel lao-vowel lao-maa-sakod lao-tone
 441              clen cidx)
 442         (setq to (match-end 0))
 443         (if (not consonant)
 444             (setq str (cdr (assoc (match-string 0 str)
 445                                   lao-transcription-punctuation-alist)))
 446           (setq lao-consonant
 447                 (cdr (assoc consonant lao-transcription-consonant-alist)))
 448           (if (vectorp lao-consonant)
 449               (setq lao-consonant (aref lao-consonant 0)))
 450           (setq clen (length lao-consonant))
 451           (if semivowel
 452               ;; Include semivowel in STR.
 453               (setq lao-semivowel
 454                     (cdr (assoc semivowel lao-transcription-semi-vowel-alist))
 455                     str (if (= clen 1)
 456                             (concat lao-consonant lao-semivowel)
 457                           (concat (substring lao-consonant 0 1) lao-semivowel
 458                                   (substring lao-consonant 1))))
 459             (setq str lao-consonant))
 460           (if vowel
 461               (let (rule)
 462                 (setq lao-vowel
 463                       (cdr (assoc vowel lao-transcription-vowel-alist)))
 464                 (if (vectorp lao-vowel)
 465                     (setq lao-vowel (aref lao-vowel 0)))
 466                 (setq rule (assoc lao-vowel lao-vowel-reordering-rule))
 467                 (if (null maa-sakod)
 468                     (setq rule (nth 1 rule))
 469                   (setq lao-maa-sakod
 470                         (cdr (assoc maa-sakod lao-transcription-maa-sakod-alist))
 471                         rule
 472                         (or (cdr (assq (aref lao-maa-sakod 0) (nthcdr 2 rule)))
 473                             (nth 2 rule)
 474                             (nth 1 rule))))
 475                 (or rule
 476                     (error "Lao vowel %S has no re-ordering rule" lao-vowel))
 477                 (setq lao-consonant str str "")
 478                 (while rule
 479                   (if (= (car rule) 0)
 480                       (setq str (concat str lao-consonant)
 481                             cidx (length str))
 482                     (setq str (concat str (list (car rule)))))
 483                   (setq rule (cdr rule)))
 484                 (or cidx
 485                     (error "Lao vowel %S has malformed re-ordering rule" vowel))
 486                 ;; Set CIDX to after upper or lower vowel if any.
 487                 (let ((len (length str)))
 488                   (while (and (< cidx len)
 489                               (memq (get-char-code-property (aref str cidx)
 490                                                             'phonetic-type)
 491                                     '(vowel-lower vowel-upper)))
 492                     (setq cidx (1+ cidx))))
 493                 (if lao-maa-sakod
 494                     (setq str (concat str lao-maa-sakod)))
 495                 (if tone
 496                     (setq lao-tone
 497                           (cdr (assoc tone lao-transcription-tone-alist))
 498                           str (concat (substring str 0 cidx) lao-tone
 499                                       (substring str cidx)))))))
 500         (list from to (lao-compose-string str)))))
 501
 502 ;;;###autoload
 503 (defun lao-transcribe-roman-to-lao-string (str)
 504   "Transcribe Romanized Lao string STR to Lao character string."
 505   (let ((from 0)
 506         (to (length str))
 507         (lao-str "")
 508         val)
 509     (while (setq val (lao-transcribe-single-roman-syllable-to-lao from to str))
 510       (let ((start (car val))
 511             (end (nth 1 val))
 512             (lao (nth 2 val)))
 513         (if (> start from)
 514             (setq lao-str (concat lao-str (substring str from start) lao))
 515           (setq lao-str (concat lao-str lao)))
 516         (setq from end)))
 517     (if (< from to)
 518         (concat lao-str (substring str from to))
 519       lao-str)))
 520
 521 ;;;###autoload
 522 (defun lao-composition-function (pos &optional string)
 523   (setq pos (1- pos))
 524   (with-category-table lao-category-table
 525     (if string
 526         (if (and (>= pos 0)
 527                  (eq (string-match lao-composition-pattern string pos) pos))
 528             (prog1 (match-end 0)
 529               (compose-string string pos (match-end 0))))
 530       (if (>= pos (point-min))
 531           (save-excursion
 532             (goto-char pos)
 533             (if (looking-at lao-composition-pattern)
 534                 (prog1 (match-end 0)
 535                   (compose-region pos (match-end 0)))))))))
 536
 537 ;;;###autoload
 538 (defun lao-compose-region (from to)
 539   (interactive "r")
 540   (save-restriction
 541     (narrow-to-region from to)
 542     (goto-char (point-min))
 543     (with-category-table lao-category-table
 544       (while (re-search-forward lao-composition-pattern nil t)
 545         (compose-region (match-beginning 0) (point))))))
 546
 547 ;;
 548 (provide 'lao-util)
 549
 550 ;;; lao-util.el ends here