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