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) 2001  Free Software Foundation, Inc.
   4 ;; Copyright (C) 1997, 1998, 1999, 2000
   5 ;;   National Institute of Advanced Industrial Science and Technology (AIST)
   6 ;;   Registration Number H14PRO021
   7
   8 ;; Keywords: multilingual, Lao
   9
  10 ;; This file is part of GNU Emacs.
  11
  12 ;; GNU Emacs is free software; you can redistribute it and/or modify
  13 ;; it under the terms of the GNU General Public License as published by
  14 ;; the Free Software Foundation; either version 2, or (at your option)
  15 ;; any later version.
  16
  17 ;; GNU Emacs is distributed in the hope that it will be useful,
  18 ;; but WITHOUT ANY WARRANTY; without even the implied warranty of
  19 ;; MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
  20 ;; GNU General Public License for more details.
  21
  22 ;; You should have received a copy of the GNU General Public License
  23 ;; along with GNU Emacs; see the file COPYING.  If not, write to the
  24 ;; Free Software Foundation, Inc., 51 Franklin Street, Fifth Floor,
  25 ;; Boston, MA 02110-1301, USA.
  26
  27 ;;; Commentary:
  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 '((?\e(1!\e(B consonant "LETTER KOR  KAI'" "CHICKEN")
  40            (?\e(1"\e(B consonant "LETTER KHOR KHAI'" "EGG")
  41            (?\e(1#\e(B invalid nil)
  42            (?\e(1$\e(B consonant "LETTER QHOR QHWARGN" "BUFFALO")
  43            (?\e(1%\e(B invalid nil)
  44            (?  invalid nil)
  45            (?\e(1'\e(B consonant "LETTER NGOR NGUU" "SNAKE")
  46            (?\e(1(\e(B consonant "LETTER JOR JUA" "BUDDHIST NOVICE")
  47            (?\e(1)\e(B invalid nil)
  48            (?\e(1*\e(B consonant "LETTER XOR X\"ARNG" "ELEPHANT")
  49            (?\e(1+\e(B invalid nil)
  50            (?\e(1,\e(B invalid nil)
  51            (?\e(1-\e(B consonant "LETTER YOR YUNG" "MOSQUITO")
  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(1.\e(B invalid nil)
  57            (?\e(1.\e(B invalid nil)
  58            (?\e(14\e(B consonant "LETTER DOR DANG" "NOSE")
  59            (?\e(15\e(B consonant "LETTER TOR TAR" "EYE")
  60            (?\e(16\e(B consonant "LETTER THOR THUNG" "TO ASK,QUESTION")
  61            (?\e(17\e(B consonant "LETTER DHOR DHARM" "FLAG")
  62            (?\e(18\e(B invalid nil)
  63            (?\e(19\e(B consonant "LETTER NOR NOK" "BIRD")
  64            (?\e(1:\e(B consonant "LETTER BOR BED" "FISHHOOK")
  65            (?\e(1;\e(B consonant "LETTER POR PAR" "FISH")
  66            (?\e(1<\e(B consonant "LETTER HPOR HPER\"" "BEE")
  67            (?\e(1=\e(B consonant "LETTER FHOR FHAR" "WALL")
  68            (?\e(1>\e(B consonant "LETTER PHOR PHUU" "MOUNTAIN")
  69            (?\e(1?\e(B consonant "LETTER FOR FAI" "FIRE")
  70            (?\e(1@\e(B invalid nil)
  71            (?\e(1A\e(B consonant "LETTER MOR MAR\"" "HORSE")
  72            (?\e(1B\e(B consonant "LETTER GNOR GNAR" "MEDICINE")
  73            (?\e(1C\e(B consonant "LETTER ROR ROD" "CAR")
  74            (?\e(1D\e(B invalid nil)
  75            (?\e(1E\e(B consonant "LETTER LOR LIING" "MONKEY")
  76            (?\e(1F\e(B invalid nil)
  77            (?\e(1G\e(B consonant "LETTER WOR WII" "HAND FAN")
  78            (?\e(1H\e(B invalid nil)
  79            (?\e(1I\e(B invalid nil)
  80            (?\e(1J\e(B consonant "LETTER SOR SEA" "TIGER")
  81            (?\e(1K\e(B consonant "LETTER HHOR HHAI" "JAR")
  82            (?\e(1L\e(B invalid nil)
  83            (?\e(1M\e(B consonant "LETTER OR OOW" "TAKE")
  84            (?\e(1N\e(B consonant "LETTER HOR HEA" "BOAT")
  85            (?\e(1O\e(B special "ELLIPSIS")
  86            (?\e(1P\e(B vowel-base "VOWEL SIGN SARA A")
  87            (?\e(1Q\e(B vowel-upper "VOWEL SIGN MAI KAN")
  88            (?\e(1R\e(B vowel-base "VOWEL SIGN SARA AR")
  89            (?\e(1S\e(B vowel-base "VOWEL SIGN SARA AM")
  90            (?\e(1T\e(B vowel-upper "VOWEL SIGN SARA I")
  91            (?\e(1U\e(B vowel-upper "VOWEL SIGN SARA II")
  92            (?\e(1V\e(B vowel-upper "VOWEL SIGN SARA EU")
  93            (?\e(1W\e(B vowel-upper "VOWEL SIGN SARA UR")
  94            (?\e(1X\e(B vowel-lower "VOWEL SIGN SARA U")
  95            (?\e(1Y\e(B vowel-lower "VOWEL SIGN SARA UU")
  96            (?\e(1Z\e(B invalid nil)
  97            (?\e(1[\e(B vowel-upper "VOWEL SIGN MAI KONG")
  98            (?\e(1\\e(B semivowel-lower "SEMIVOWEL SIGN LO")
  99            (?\e(1]\e(B vowel-base "SEMIVOWEL SIGN SARA IA")
 100            (?\e(1^\e(B invalid nil)
 101            (?\e(1_\e(B invalid nil)
 102            (?\e(1`\e(B vowel-base "VOWEL SIGN SARA EE")
 103            (?\e(1a\e(B vowel-base "VOWEL SIGN SARA AA")
 104            (?\e(1b\e(B vowel-base "VOWEL SIGN SARA OO")
 105            (?\e(1c\e(B vowel-base "VOWEL SIGN SARA EI MAI MUAN\"")
 106            (?\e(1d\e(B vowel-base "VOWEL SIGN SARA AI MAI MAY")
 107            (?\e(1e\e(B invalid nil)
 108            (?\e(1f\e(B special "KO LA (REPETITION)")
 109            (?\e(1g\e(B invalid nil)
 110            (?\e(1h\e(B tone "TONE MAI EK")
 111            (?\e(1i\e(B tone "TONE MAI THO")
 112            (?\e(1j\e(B tone "TONE MAI TI")
 113            (?\e(1k\e(B tone "TONE MAI JADTAWAR")
 114            (?\e(1l\e(B tone "CANCELLATION MARK")
 115            (?\e(1m\e(B vowel-upper "VOWEL SIGN SARA OR")
 116            (?\e(1n\e(B invalid nil)
 117            (?\e(1o\e(B invalid nil)
 118            (?\e(1p\e(B special "DIGIT ZERO")
 119            (?\e(1q\e(B special "DIGIT ONE")
 120            (?\e(1r\e(B special "DIGIT TWO")
 121            (?\e(1s\e(B special "DIGIT THREE")
 122            (?\e(1t\e(B special "DIGIT FOUR")
 123            (?\e(1u\e(B special "DIGIT FIVE")
 124            (?\e(1v\e(B special "DIGIT SIX")
 125            (?\e(1w\e(B special "DIGIT SEVEN")
 126            (?\e(1x\e(B special "DIGIT EIGHT")
 127            (?\e(1y\e(B special "DIGIT NINE")
 128            (?\e(1z\e(B invalid nil)
 129            (?\e(1{\e(B invalid nil)
 130            (?\e(1|\e(B consonant "LETTER NHOR NHUU" "MOUSE")
 131            (?\e(1}\e(B consonant "LETTER MHOR MHAR" "DOG")
 132            (?\e(1~\e(B invalid nil)
 133            ;; Unicode equivalents
 134            (?\e$,1D!\e(B consonant "LETTER KOR  KAI'" "CHICKEN")
 135            (?\e$,1D"\e(B consonant "LETTER KHOR KHAI'" "EGG")
 136            (?\e$,1D$\e(B consonant "LETTER QHOR QHWARGN" "BUFFALO")
 137            (?\e$,1D'\e(B consonant "LETTER NGOR NGUU" "SNAKE")
 138            (?\e$,1D(\e(B consonant "LETTER JOR JUA" "BUDDHIST NOVICE")
 139            (?\e$,1D*\e(B consonant "LETTER XOR X\"ARNG" "ELEPHANT")
 140            (?\e$,1D-\e(B consonant "LETTER YOR YUNG" "MOSQUITO")
 141            (?\e$,1D4\e(B consonant "LETTER DOR DANG" "NOSE")
 142            (?\e$,1D5\e(B consonant "LETTER TOR TAR" "EYE")
 143            (?\e$,1D6\e(B consonant "LETTER THOR THUNG" "TO ASK,QUESTION")
 144            (?\e$,1D7\e(B consonant "LETTER DHOR DHARM" "FLAG")
 145            (?\e$,1D9\e(B consonant "LETTER NOR NOK" "BIRD")
 146            (?\e$,1D:\e(B consonant "LETTER BOR BED" "FISHHOOK")
 147            (?\e$,1D;\e(B consonant "LETTER POR PAR" "FISH")
 148            (?\e$,1D<\e(B consonant "LETTER HPOR HPER\"" "BEE")
 149            (?\e$,1D=\e(B consonant "LETTER FHOR FHAR" "WALL")
 150            (?\e$,1D>\e(B consonant "LETTER PHOR PHUU" "MOUNTAIN")
 151            (?\e$,1D?\e(B consonant "LETTER FOR FAI" "FIRE")
 152            (?\e$,1DA\e(B consonant "LETTER MOR MAR\"" "HORSE")
 153            (?\e$,1DB\e(B consonant "LETTER GNOR GNAR" "MEDICINE")
 154            (?\e$,1DC\e(B consonant "LETTER ROR ROD" "CAR")
 155            (?\e$,1DE\e(B consonant "LETTER LOR LIING" "MONKEY")
 156            (?\e$,1DG\e(B consonant "LETTER WOR WII" "HAND FAN")
 157            (?\e$,1DJ\e(B consonant "LETTER SOR SEA" "TIGER")
 158            (?\e$,1DK\e(B consonant "LETTER HHOR HHAI" "JAR")
 159            (?\e$,1DM\e(B consonant "LETTER OR OOW" "TAKE")
 160            (?\e$,1DN\e(B consonant "LETTER HOR HEA" "BOAT")
 161            (?\e$,1DO\e(B special "ELLIPSIS")
 162            (?\e$,1DP\e(B vowel-base "VOWEL SIGN SARA A")
 163            (?\e$,1DQ\e(B vowel-upper "VOWEL SIGN MAI KAN")
 164            (?\e$,1DR\e(B vowel-base "VOWEL SIGN SARA AR")
 165            (?\e$,1DS\e(B vowel-base "VOWEL SIGN SARA AM")
 166            (?\e$,1DT\e(B vowel-upper "VOWEL SIGN SARA I")
 167            (?\e$,1DU\e(B vowel-upper "VOWEL SIGN SARA II")
 168            (?\e$,1DV\e(B vowel-upper "VOWEL SIGN SARA EU")
 169            (?\e$,1DW\e(B vowel-upper "VOWEL SIGN SARA UR")
 170            (?\e$,1DX\e(B vowel-lower "VOWEL SIGN SARA U")
 171            (?\e$,1DY\e(B vowel-lower "VOWEL SIGN SARA UU")
 172            (?\e$,1D[\e(B vowel-upper "VOWEL SIGN MAI KONG")
 173            (?\e$,1D\\e(B semivowel-lower "SEMIVOWEL SIGN LO")
 174            (?\e$,1D]\e(B vowel-base "SEMIVOWEL SIGN SARA IA")
 175            (?\e$,1D`\e(B vowel-base "VOWEL SIGN SARA EE")
 176            (?\e$,1Da\e(B vowel-base "VOWEL SIGN SARA AA")
 177            (?\e$,1Db\e(B vowel-base "VOWEL SIGN SARA OO")
 178            (?\e$,1Dc\e(B vowel-base "VOWEL SIGN SARA EI MAI MUAN\"")
 179            (?\e$,1Dd\e(B vowel-base "VOWEL SIGN SARA AI MAI MAY")
 180            (?\e$,1Df\e(B special "KO LA (REPETITION)")
 181            (?\e$,1Dh\e(B tone "TONE MAI EK")
 182            (?\e$,1Di\e(B tone "TONE MAI THO")
 183            (?\e$,1Dj\e(B tone "TONE MAI TI")
 184            (?\e$,1Dk\e(B tone "TONE MAI JADTAWAR")
 185            (?\e$,1Dl\e(B tone "CANCELLATION MARK")
 186            (?\e$,1Dm\e(B vowel-upper "VOWEL SIGN SARA OR")
 187            (?\e$,1Dp\e(B special "DIGIT ZERO")
 188            (?\e$,1Dq\e(B special "DIGIT ONE")
 189            (?\e$,1Dr\e(B special "DIGIT TWO")
 190            (?\e$,1Ds\e(B special "DIGIT THREE")
 191            (?\e$,1Dt\e(B special "DIGIT FOUR")
 192            (?\e$,1Du\e(B special "DIGIT FIVE")
 193            (?\e$,1Dv\e(B special "DIGIT SIX")
 194            (?\e$,1Dw\e(B special "DIGIT SEVEN")
 195            (?\e$,1Dx\e(B special "DIGIT EIGHT")
 196            (?\e$,1Dy\e(B special "DIGIT NINE")
 197            (?\e$,1D|\e(B consonant "LETTER NHOR NHUU" "MOUSE")
 198            (?\e$,1D}\e(B consonant "LETTER MHOR MHAR" "DOG")))
 199       elm)
 200   (while l
 201     (setq elm (car l) l (cdr l))
 202     (let ((char (car elm))
 203           (ptype (nth 1 elm)))
 204       (cond ((eq ptype 'consonant)
 205              (modify-category-entry char ?c lao-category-table))
 206             ((memq ptype '(vowel-upper vowel-lower))
 207              (modify-category-entry char ?v lao-category-table))
 208             ((eq ptype 'semivowel-lower)
 209              (modify-category-entry char ?s lao-category-table))
 210             ((eq ptype 'tone)
 211              (modify-category-entry char ?t lao-category-table)))
 212       (put-char-code-property char 'phonetic-type ptype)
 213       (put-char-code-property char 'name (nth 2 elm))
 214       (put-char-code-property char 'meaning (nth 3 elm)))))
 215
 216 ;; The general composing rules are as follows:
 217 ;;
 218 ;;                          T
 219 ;;       V        T         V                  T
 220 ;; CV -> C, CT -> C, CVT -> C, Cv -> C, CvT -> C
 221 ;;                                   v         v
 222 ;;                             T
 223 ;;        V         T          V                   T
 224 ;; CsV -> C, CsT -> C, CsVT -> C, Csv -> C, CvT -> C
 225 ;;        s         s          s         s         s
 226 ;;                                       v         v
 227
 228
 229 ;; where C: consonant, V: vowel upper, v: vowel lower,
 230 ;;       T: tone mark, s: semivowel lower
 231
 232 (defvar lao-composition-pattern
 233   "\\cc\\(\\ct\\|\\cv\\ct?\\|\\cs\\(\\ct\\|\\cv\\ct?\\)?\\)"
 234   "Regular expression matching a Lao composite sequence.")
 235
 236 ;;;###autoload
 237 (defun lao-compose-string (str)
 238   (with-category-table lao-category-table
 239    (let ((idx 0))
 240      (while (setq idx (string-match lao-composition-pattern str idx))
 241        (compose-string str idx (match-end 0))
 242        (setq idx (match-end 0))))
 243    str))
 244
 245 ;;; LRT: Lao <-> Roman Transcription
 246
 247 ;; Upper vowels and tone-marks are put on the letter.
 248 ;; Semi-vowel-sign-lo and lower vowels are put under the letter.
 249
 250 (defconst lao-transcription-consonant-alist
 251   (sort '(;; single consonants
 252           ("k" . "\e(1!\e(B")
 253           ("kh" . "\e(1"\e(B")
 254           ("qh" . "\e(1$\e(B")
 255           ("ng" . "\e(1'\e(B")
 256           ("j" . "\e(1(\e(B")
 257           ("s" . "\e(1J\e(B")
 258           ("x" . "\e(1*\e(B")
 259           ("y" . "\e(1-\e(B")
 260           ("d" . "\e(14\e(B")
 261           ("t" . "\e(15\e(B")
 262           ("th" . "\e(16\e(B")
 263           ("dh" . "\e(17\e(B")
 264           ("n" . "\e(19\e(B")
 265           ("b" . "\e(1:\e(B")
 266           ("p" . "\e(1;\e(B")
 267           ("hp" . "\e(1<\e(B")
 268           ("fh" . "\e(1=\e(B")
 269           ("ph" . "\e(1>\e(B")
 270           ("f" . "\e(1?\e(B")
 271           ("m" . "\e(1A\e(B")
 272           ("gn" . "\e(1B\e(B")
 273           ("l" . "\e(1E\e(B")
 274           ("r" . "\e(1C\e(B")
 275           ("v" . "\e(1G\e(B")
 276           ("w" . "\e(1G\e(B")
 277           ("hh" . "\e(1K\e(B")
 278           ("O" . "\e(1M\e(B")
 279           ("h" . "\e(1N\e(B")
 280           ("nh" . "\e(1|\e(B")
 281           ("mh" . "\e(1}\e(B")
 282           ("lh" . ["\e(1K\\e(B"])
 283           ;; double consonants
 284           ("ngh" . ["\e(1K'\e(B"])
 285           ("yh" . ["\e(1K]\e(B"])
 286           ("wh" . ["\e(1KG\e(B"])
 287           ("hl" . ["\e(1KE\e(B"])
 288           ("hy" . ["\e(1K-\e(B"])
 289           ("hn" . ["\e(1K9\e(B"])
 290           ("hm" . ["\e(1KA\e(B"])
 291           )
 292         (function (lambda (x y) (> (length (car x)) (length (car y)))))))
 293
 294 (defconst lao-transcription-semi-vowel-alist
 295   '(("r" . "\e(1\\e(B")))
 296
 297 (defconst lao-transcription-vowel-alist
 298   (sort '(("a" . "\e(1P\e(B")
 299           ("ar" . "\e(1R\e(B")
 300           ("i" . "\e(1T\e(B")
 301           ("ii" . "\e(1U\e(B")
 302           ("eu" . "\e(1V\e(B")
 303           ("ur" . "\e(1W\e(B")
 304           ("u" . "\e(1X\e(B")
 305           ("uu" . "\e(1Y\e(B")
 306           ("e" . ["\e(1`P\e(B"])
 307           ("ee" . "\e(1`\e(B")
 308           ("ae" . ["\e(1aP\e(B"])
 309           ("aa" . "\e(1a\e(B")
 310           ("o" . ["\e(1bP\e(B"])
 311           ("oo" . "\e(1b\e(B")
 312           ("oe" . ["\e(1`RP\e(B"])
 313           ("or" . "\e(1m\e(B")
 314           ("er" . ["\e(1`T\e(B"])
 315           ("ir" . ["\e(1`U\e(B"])
 316           ("ua" . ["\e(1[GP\e(B"])
 317           ("uaa" . ["\e(1[G\e(B"])
 318           ("ie" . ["\e(1`Q]P\e(B"])
 319           ("ia" . ["\e(1`Q]\e(B"])
 320           ("ea" . ["\e(1`VM\e(B"])
 321           ("eaa" . ["\e(1`WM\e(B"])
 322           ("ai" . "\e(1d\e(B")
 323           ("ei" . "\e(1c\e(B")
 324           ("ao" . ["\e(1`[R\e(B"])
 325           ("aM" . "\e(1S\e(B"))
 326         (function (lambda (x y) (> (length (car x)) (length (car y)))))))
 327
 328 ;; Maa-sakod is put at the tail.
 329 (defconst lao-transcription-maa-sakod-alist
 330   '(("k" . "\e(1!\e(B")
 331     ("g" . "\e(1'\e(B")
 332     ("y" . "\e(1-\e(B")
 333     ("d" . "\e(14\e(B")
 334     ("n" . "\e(19\e(B")
 335     ("b" . "\e(1:\e(B")
 336     ("m" . "\e(1A\e(B")
 337     ("v" . "\e(1G\e(B")
 338     ("w" . "\e(1G\e(B")
 339     ))
 340
 341 (defconst lao-transcription-tone-alist
 342   '(("'" . "\e(1h\e(B")
 343     ("\"" . "\e(1i\e(B")
 344     ("^" . "\e(1j\e(B")
 345     ("+" . "\e(1k\e(B")
 346     ("~" . "\e(1l\e(B")))
 347
 348 (defconst lao-transcription-punctuation-alist
 349   '(("\\0" . "\e(1p\e(B")
 350     ("\\1" . "\e(1q\e(B")
 351     ("\\2" . "\e(1r\e(B")
 352     ("\\3" . "\e(1s\e(B")
 353     ("\\4" . "\e(1t\e(B")
 354     ("\\5" . "\e(1u\e(B")
 355     ("\\6" . "\e(1v\e(B")
 356     ("\\7" . "\e(1w\e(B")
 357     ("\\8" . "\e(1x\e(B")
 358     ("\\9" . "\e(1y\e(B")
 359     ("\\\\" . "\e(1f\e(B")
 360     ("\\$" . "\e(1O\e(B")))
 361
 362 (defconst lao-transcription-pattern
 363   (concat
 364    "\\("
 365    (mapconcat 'car lao-transcription-consonant-alist "\\|")
 366    "\\)\\("
 367    (mapconcat 'car lao-transcription-semi-vowel-alist "\\|")
 368    "\\)?\\(\\("
 369    (mapconcat 'car lao-transcription-vowel-alist "\\|")
 370    "\\)\\("
 371    (mapconcat 'car lao-transcription-maa-sakod-alist "\\|")
 372    "\\)?\\("
 373    (mapconcat (lambda (x) (regexp-quote (car x)))
 374               lao-transcription-tone-alist "\\|")
 375    "\\)?\\)?\\|"
 376    (mapconcat (lambda (x) (regexp-quote (car x)))
 377               lao-transcription-punctuation-alist "\\|")
 378    )
 379   "Regexp of Roman transcription pattern for one Lao syllable.")
 380
 381 (defconst lao-transcription-pattern
 382   (concat
 383    "\\("
 384    (regexp-opt (mapcar 'car lao-transcription-consonant-alist))
 385    "\\)\\("
 386    (regexp-opt (mapcar 'car lao-transcription-semi-vowel-alist))
 387    "\\)?\\(\\("
 388    (regexp-opt (mapcar 'car lao-transcription-vowel-alist))
 389    "\\)\\("
 390    (regexp-opt (mapcar 'car lao-transcription-maa-sakod-alist))
 391    "\\)?\\("
 392    (regexp-opt (mapcar 'car lao-transcription-tone-alist))
 393    "\\)?\\)?\\|"
 394    (regexp-opt (mapcar 'car lao-transcription-punctuation-alist))
 395    )
 396   "Regexp of Roman transcription pattern for one Lao syllable.")
 397
 398 (defconst lao-vowel-reordering-rule
 399   '(("\e(1P\e(B" (0 ?\e(1P\e(B) (0 ?\e(1Q\e(B))
 400     ("\e(1R\e(B" (0 ?\e(1R\e(B))
 401     ("\e(1T\e(B" (0 ?\e(1U\e(B))
 402     ("\e(1U\e(B" (0 ?\e(1U\e(B))
 403     ("\e(1V\e(B" (0 ?\e(1V\e(B))
 404     ("\e(1W\e(B" (0 ?\e(1W\e(B))
 405     ("\e(1X\e(B" (0 ?\e(1X\e(B))
 406     ("\e(1Y\e(B" (0 ?\e(1Y\e(B))
 407     ("\e(1`P\e(B" (?\e(1`\e(B 0 ?\e(1P\e(B) (?\e(1`\e(B 0 ?\e(1Q\e(B))
 408     ("\e(1`\e(B" (?\e(1`\e(B 0))
 409     ("\e(1aP\e(B" (?\e(1a\e(B 0 ?\e(1P\e(B) (?\e(1a\e(B 0 ?\e(1Q\e(B))
 410     ("\e(1a\e(B" (?\e(1a\e(B 0))
 411     ("\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))
 412     ("\e(1b\e(B" (?\e(1b\e(B 0))
 413     ("\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))
 414     ("\e(1m\e(B" (0 ?\e(1m\e(B) (0 ?\e(1M\e(B))
 415     ("\e(1`T\e(B" (?\e(1`\e(B 0 ?\e(1T\e(B))
 416     ("\e(1`U\e(B" (?\e(1`\e(B 0 ?\e(1U\e(B))
 417     ("\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))
 418     ("\e(1[G\e(B" (0 ?\e(1[\e(B ?\e(1G\e(B) (0 ?\e(1G\e(B))
 419     ("\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))
 420     ("\e(1`Q]\e(B" (?\e(1`\e(B 0 ?\e(1Q\e(B ?\e(1]\e(B) (0 ?\e(1]\e(B))
 421     ("\e(1`VM\e(B" (?\e(1`\e(B 0 ?\e(1V\e(B ?\e(1M\e(B))
 422     ("\e(1`WM\e(B" (?\e(1`\e(B 0 ?\e(1W\e(B ?\e(1M\e(B))
 423     ("\e(1d\e(B" (?\e(1d\e(B 0))
 424     ("\e(1c\e(B" (?\e(1c\e(B 0))
 425     ("\e(1`[R\e(B" (?\e(1`\e(B 0 ?\e(1[\e(B ?\e(1R\e(B))
 426     ("\e(1S\e(B" (0 ?\e(1S\e(B))
 427
 428     ;; Unicode equivalents
 429     ("\e$,1DP\e(B" (0 ?\e$,1DP\e(B) (0 ?\e$,1DQ\e(B))
 430     ("\e$,1DR\e(B" (0 ?\e$,1DR\e(B))
 431     ("\e$,1DT\e(B" (0 ?\e$,1DU\e(B))
 432     ("\e$,1DU\e(B" (0 ?\e$,1DU\e(B))
 433     ("\e$,1DV\e(B" (0 ?\e$,1DV\e(B))
 434     ("\e$,1DW\e(B" (0 ?\e$,1DW\e(B))
 435     ("\e$,1DX\e(B" (0 ?\e$,1DX\e(B))
 436     ("\e$,1DY\e(B" (0 ?\e$,1DY\e(B))
 437     ("\e$,1D`DP\e(B" (?\e$,1D`\e(B 0 ?\e$,1DP\e(B) (?\e$,1D`\e(B 0 ?\e$,1DQ\e(B))
 438     ("\e$,1D`\e(B" (?\e$,1D`\e(B 0))
 439     ("\e$,1DaDP\e(B" (?\e$,1Da\e(B 0 ?\e$,1DP\e(B) (?\e$,1Da\e(B 0 ?\e$,1DQ\e(B))
 440     ("\e$,1Da\e(B" (?\e$,1Da\e(B 0))
 441     ("\e$,1DbDP\e(B" (?\e$,1Db\e(B 0 ?\e$,1DP\e(B) (0 ?\e$,1D[\e(B) (?\e$,1D-\e(B ?\e$,1Db\e(B 0 ?\e$,1DQ\e(B) (?\e$,1DG\e(B ?\e$,1Db\e(B 0 ?\e$,1DQ\e(B))
 442     ("\e$,1Db\e(B" (?\e$,1Db\e(B 0))
 443     ("\e$,1D`DRDP\e(B" (?\e$,1D`\e(B 0 ?\e$,1DR\e(B ?\e$,1DP\e(B) (0 ?\e$,1DQ\e(B ?\e$,1DM\e(B))
 444     ("\e$,1Dm\e(B" (0 ?\e$,1Dm\e(B) (0 ?\e$,1DM\e(B))
 445     ("\e$,1D`DT\e(B" (?\e$,1D`\e(B 0 ?\e$,1DT\e(B))
 446     ("\e$,1D`DU\e(B" (?\e$,1D`\e(B 0 ?\e$,1DU\e(B))
 447     ("\e$,1D[DGDP\e(B" (0 ?\e$,1D[\e(B ?\e$,1DG\e(B ?\e$,1DP\e(B) (0 ?\e$,1DQ\e(B ?\e$,1DG\e(B))
 448     ("\e$,1D[DG\e(B" (0 ?\e$,1D[\e(B ?\e$,1DG\e(B) (0 ?\e$,1DG\e(B))
 449     ("\e$,1D`DQD]DP\e(B" (?\e$,1D`\e(B 0 ?\e$,1DQ\e(B ?\e$,1D]\e(B ?\e$,1DP\e(B) (0 ?\e$,1DQ\e(B ?\e$,1D]\e(B))
 450     ("\e$,1D`DQD]\e(B" (?\e$,1D`\e(B 0 ?\e$,1DQ\e(B ?\e$,1D]\e(B) (0 ?\e$,1D]\e(B))
 451     ("\e$,1D`DVDM\e(B" (?\e$,1D`\e(B 0 ?\e$,1DV\e(B ?\e$,1DM\e(B))
 452     ("\e$,1D`DWDM\e(B" (?\e$,1D`\e(B 0 ?\e$,1DW\e(B ?\e$,1DM\e(B))
 453     ("\e$,1Dd\e(B" (?\e$,1Dd\e(B 0))
 454     ("\e$,1Dc\e(B" (?\e$,1Dc\e(B 0))
 455     ("\e$,1D`D[DR\e(B" (?\e$,1D`\e(B 0 ?\e$,1D[\e(B ?\e$,1DR\e(B))
 456     ("\e$,1DS\e(B" (0 ?\e$,1DS\e(B)))
 457   "Alist of Lao vowel string vs the corresponding re-ordering rule.
 458 Each element has this form:
 459         (VOWEL NO-MAA-SAKOD-RULE WITH-MAA-SAKOD-RULE (MAA-SAKOD-0 RULE-0) ...)
 460
 461 VOWEL is a vowel string (e.g. \"\e(1`Q]P\e(B\").
 462
 463 NO-MAA-SAKOD-RULE is a rule to re-order and modify VOWEL following a
 464 consonant.  It is a list vowel characters or 0.  The element 0
 465 indicate the place to embed a consonant.
 466
 467 Optional WITH-MAA-SAKOD-RULE is a rule to re-order and modify VOWEL
 468 follwoing a consonant and preceding a maa-sakod character.  If it is
 469 nil, NO-MAA-SAKOD-RULE is used.  The maa-sakod character is alwasy
 470 appended at the tail.
 471
 472 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
 473 string following a consonant `\e(1!\e(B' should be re-ordered as \"\e(1`!WM\e(B\".
 474
 475 Optional (MAA-SAKOD-n RULE-n) are rules specially applied to maa-sakod
 476 character MAA-SAKOD-n.")
 477
 478 ;;;###autoload
 479 (defun lao-transcribe-single-roman-syllable-to-lao (from to &optional str)
 480   "Transcribe a Romanized Lao syllable in the region FROM and TO to Lao string.
 481 Only the first syllable is transcribed.
 482 The value has the form: (START END LAO-STRING), where
 483 START and END are the beggining and end positions of the Roman Lao syllable,
 484 LAO-STRING is the Lao character transcription of it.
 485
 486 Optional 3rd arg STR, if non-nil, is a string to search for Roman Lao
 487 syllable.  In that case, FROM and TO are indexes to STR."
 488   (if str
 489       (if (setq from (string-match lao-transcription-pattern str from))
 490           (progn
 491             (if (>= from to)
 492                 (setq from nil)
 493               (setq to (match-end 0)))))
 494     (save-excursion
 495       (goto-char from)
 496       (if (setq to (re-search-forward lao-transcription-pattern to t))
 497           (setq from (match-beginning 0))
 498         (setq from nil))))
 499   (if from
 500       (let* ((consonant (match-string 1 str))
 501              (semivowel (match-string 3 str))
 502              (vowel (match-string 5 str))
 503              (maa-sakod (match-string 8 str))
 504              (tone (match-string 9 str))
 505              lao-consonant lao-semivowel lao-vowel lao-maa-sakod lao-tone
 506              clen cidx)
 507         (setq to (match-end 0))
 508         (if (not consonant)
 509             (setq str (cdr (assoc (match-string 0 str)
 510                                   lao-transcription-punctuation-alist)))
 511           (setq lao-consonant
 512                 (cdr (assoc consonant lao-transcription-consonant-alist)))
 513           (if (vectorp lao-consonant)
 514               (setq lao-consonant (aref lao-consonant 0)))
 515           (setq clen (length lao-consonant))
 516           (if semivowel
 517               ;; Include semivowel in STR.
 518               (setq lao-semivowel
 519                     (cdr (assoc semivowel lao-transcription-semi-vowel-alist))
 520                     str (if (= clen 1)
 521                             (concat lao-consonant lao-semivowel)
 522                           (concat (substring lao-consonant 0 1) lao-semivowel
 523                                   (substring lao-consonant 1))))
 524             (setq str lao-consonant))
 525           (if vowel
 526               (let (rule)
 527                 (setq lao-vowel
 528                       (cdr (assoc vowel lao-transcription-vowel-alist)))
 529                 (if (vectorp lao-vowel)
 530                     (setq lao-vowel (aref lao-vowel 0)))
 531                 (setq rule (assoc lao-vowel lao-vowel-reordering-rule))
 532                 (if (null maa-sakod)
 533                     (setq rule (nth 1 rule))
 534                   (setq lao-maa-sakod
 535                         (cdr (assoc maa-sakod lao-transcription-maa-sakod-alist))
 536                         rule
 537                         (or (cdr (assq (aref lao-maa-sakod 0) (nthcdr 2 rule)))
 538                             (nth 2 rule)
 539                             (nth 1 rule))))
 540                 (or rule
 541                     (error "Lao vowel %S has no re-ordering rule" lao-vowel))
 542                 (setq lao-consonant str str "")
 543                 (while rule
 544                   (if (= (car rule) 0)
 545                       (setq str (concat str lao-consonant)
 546                             cidx (length str))
 547                     (setq str (concat str (list (car rule)))))
 548                   (setq rule (cdr rule)))
 549                 (or cidx
 550                     (error "Lao vowel %S has malformed re-ordering rule" vowel))
 551                 ;; Set CIDX to after upper or lower vowel if any.
 552                 (let ((len (length str)))
 553                   (while (and (< cidx len)
 554                               (memq (get-char-code-property (aref str cidx)
 555                                                             'phonetic-type)
 556                                     '(vowel-lower vowel-upper)))
 557                     (setq cidx (1+ cidx))))
 558                 (if lao-maa-sakod
 559                     (setq str (concat str lao-maa-sakod)))
 560                 (if tone
 561                     (setq lao-tone
 562                           (cdr (assoc tone lao-transcription-tone-alist))
 563                           str (concat (substring str 0 cidx) lao-tone
 564                                       (substring str cidx)))))))
 565         (list from to (lao-compose-string str)))))
 566
 567 ;;;###autoload
 568 (defun lao-transcribe-roman-to-lao-string (str)
 569   "Transcribe Romanized Lao string STR to Lao character string."
 570   (let ((from 0)
 571         (to (length str))
 572         (lao-str "")
 573         val)
 574     (while (setq val (lao-transcribe-single-roman-syllable-to-lao from to str))
 575       (let ((start (car val))
 576             (end (nth 1 val))
 577             (lao (nth 2 val)))
 578         (if (> start from)
 579             (setq lao-str (concat lao-str (substring str from start) lao))
 580           (setq lao-str (concat lao-str lao)))
 581         (setq from end)))
 582     (if (< from to)
 583         (concat lao-str (substring str from to))
 584       lao-str)))
 585
 586 ;;;###autoload
 587 (defun lao-post-read-conversion (len)
 588   (lao-compose-region (point) (+ (point) len))
 589   len)
 590
 591 ;;;###autoload
 592 (defun lao-composition-function (from to pattern &optional string)
 593   "Compose Lao text in the region FROM and TO.
 594 The text matches the regular expression PATTERN.
 595 Optional 4th argument STRING, if non-nil, is a string containing text
 596 to compose.
 597
 598 The return value is number of composed characters."
 599   (if (< (1+ from) to)
 600       (progn
 601         (if string
 602             (compose-string string from to)
 603           (compose-region from to))
 604         (- to from))))
 605
 606 ;;;###autoload
 607 (defun lao-compose-region (from to)
 608   (interactive "r")
 609   (save-restriction
 610     (narrow-to-region from to)
 611     (goto-char (point-min))
 612     (with-category-table lao-category-table
 613       (while (re-search-forward lao-composition-pattern nil t)
 614         (compose-region (match-beginning 0) (point))))))
 615
 616 ;;
 617 (provide 'lao-util)
 618
 619 ;;; arch-tag: 1f828781-3cb8-4695-88af-8f33222338ce
 620 ;;; lao-util.el ends here