lisp/language/lao-util.el

   1 ;;; lao-util.el --- utilities for Lao -*- coding: iso-2022-7bit; -*-
   2
   3 ;; Copyright (C) 2001, 2002, 2003, 2004, 2005, 2006
   4 ;;   Free Software Foundation, Inc.
   5 ;; Copyright (C) 1997, 1998, 1999, 2000
   6 ;;   National Institute of Advanced Industrial Science and Technology (AIST)
   7 ;;   Registration Number H14PRO021
   8
   9 ;; Keywords: multilingual, Lao
  10
  11 ;; This file is part of GNU Emacs.
  12
  13 ;; GNU Emacs is free software; you can redistribute it and/or modify
  14 ;; it under the terms of the GNU General Public License as published by
  15 ;; the Free Software Foundation; either version 2, or (at your option)
  16 ;; any later version.
  17
  18 ;; GNU Emacs is distributed in the hope that it will be useful,
  19 ;; but WITHOUT ANY WARRANTY; without even the implied warranty of
  20 ;; MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
  21 ;; GNU General Public License for more details.
  22
  23 ;; You should have received a copy of the GNU General Public License
  24 ;; along with GNU Emacs; see the file COPYING.  If not, write to the
  25 ;; Free Software Foundation, Inc., 51 Franklin Street, Fifth Floor,
  26 ;; Boston, MA 02110-1301, USA.
  27
  28 ;;; Commentary:
  29
  30 ;;; Code:
  31
  32 ;; Setting information of Thai characters.
  33
  34 (defconst lao-category-table (make-category-table))
  35 (define-category ?c "Lao consonant" lao-category-table)
  36 (define-category ?s "Lao semi-vowel" lao-category-table)
  37 (define-category ?v "Lao upper/lower vowel" lao-category-table)
  38 (define-category ?t "Lao tone" lao-category-table)
  39
  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            ;; Unicode equivalents
 135            (?\e$,1D!\e(B consonant "LETTER KOR  KAI'" "CHICKEN")
 136            (?\e$,1D"\e(B consonant "LETTER KHOR KHAI'" "EGG")
 137            (?\e$,1D$\e(B consonant "LETTER QHOR QHWARGN" "BUFFALO")
 138            (?\e$,1D'\e(B consonant "LETTER NGOR NGUU" "SNAKE")
 139            (?\e$,1D(\e(B consonant "LETTER JOR JUA" "BUDDHIST NOVICE")
 140            (?\e$,1D*\e(B consonant "LETTER XOR X\"ARNG" "ELEPHANT")
 141            (?\e$,1D-\e(B consonant "LETTER YOR YUNG" "MOSQUITO")
 142            (?\e$,1D4\e(B consonant "LETTER DOR DANG" "NOSE")
 143            (?\e$,1D5\e(B consonant "LETTER TOR TAR" "EYE")
 144            (?\e$,1D6\e(B consonant "LETTER THOR THUNG" "TO ASK,QUESTION")
 145            (?\e$,1D7\e(B consonant "LETTER DHOR DHARM" "FLAG")
 146            (?\e$,1D9\e(B consonant "LETTER NOR NOK" "BIRD")
 147            (?\e$,1D:\e(B consonant "LETTER BOR BED" "FISHHOOK")
 148            (?\e$,1D;\e(B consonant "LETTER POR PAR" "FISH")
 149            (?\e$,1D<\e(B consonant "LETTER HPOR HPER\"" "BEE")
 150            (?\e$,1D=\e(B consonant "LETTER FHOR FHAR" "WALL")
 151            (?\e$,1D>\e(B consonant "LETTER PHOR PHUU" "MOUNTAIN")
 152            (?\e$,1D?\e(B consonant "LETTER FOR FAI" "FIRE")
 153            (?\e$,1DA\e(B consonant "LETTER MOR MAR\"" "HORSE")
 154            (?\e$,1DB\e(B consonant "LETTER GNOR GNAR" "MEDICINE")
 155            (?\e$,1DC\e(B consonant "LETTER ROR ROD" "CAR")
 156            (?\e$,1DE\e(B consonant "LETTER LOR LIING" "MONKEY")
 157            (?\e$,1DG\e(B consonant "LETTER WOR WII" "HAND FAN")
 158            (?\e$,1DJ\e(B consonant "LETTER SOR SEA" "TIGER")
 159            (?\e$,1DK\e(B consonant "LETTER HHOR HHAI" "JAR")
 160            (?\e$,1DM\e(B consonant "LETTER OR OOW" "TAKE")
 161            (?\e$,1DN\e(B consonant "LETTER HOR HEA" "BOAT")
 162            (?\e$,1DO\e(B special "ELLIPSIS")
 163            (?\e$,1DP\e(B vowel-base "VOWEL SIGN SARA A")
 164            (?\e$,1DQ\e(B vowel-upper "VOWEL SIGN MAI KAN")
 165            (?\e$,1DR\e(B vowel-base "VOWEL SIGN SARA AR")
 166            (?\e$,1DS\e(B vowel-base "VOWEL SIGN SARA AM")
 167            (?\e$,1DT\e(B vowel-upper "VOWEL SIGN SARA I")
 168            (?\e$,1DU\e(B vowel-upper "VOWEL SIGN SARA II")
 169            (?\e$,1DV\e(B vowel-upper "VOWEL SIGN SARA EU")
 170            (?\e$,1DW\e(B vowel-upper "VOWEL SIGN SARA UR")
 171            (?\e$,1DX\e(B vowel-lower "VOWEL SIGN SARA U")
 172            (?\e$,1DY\e(B vowel-lower "VOWEL SIGN SARA UU")
 173            (?\e$,1D[\e(B vowel-upper "VOWEL SIGN MAI KONG")
 174            (?\e$,1D\\e(B semivowel-lower "SEMIVOWEL SIGN LO")
 175            (?\e$,1D]\e(B vowel-base "SEMIVOWEL SIGN SARA IA")
 176            (?\e$,1D`\e(B vowel-base "VOWEL SIGN SARA EE")
 177            (?\e$,1Da\e(B vowel-base "VOWEL SIGN SARA AA")
 178            (?\e$,1Db\e(B vowel-base "VOWEL SIGN SARA OO")
 179            (?\e$,1Dc\e(B vowel-base "VOWEL SIGN SARA EI MAI MUAN\"")
 180            (?\e$,1Dd\e(B vowel-base "VOWEL SIGN SARA AI MAI MAY")
 181            (?\e$,1Df\e(B special "KO LA (REPETITION)")
 182            (?\e$,1Dh\e(B tone "TONE MAI EK")
 183            (?\e$,1Di\e(B tone "TONE MAI THO")
 184            (?\e$,1Dj\e(B tone "TONE MAI TI")
 185            (?\e$,1Dk\e(B tone "TONE MAI JADTAWAR")
 186            (?\e$,1Dl\e(B tone "CANCELLATION MARK")
 187            (?\e$,1Dm\e(B vowel-upper "VOWEL SIGN SARA OR")
 188            (?\e$,1Dp\e(B special "DIGIT ZERO")
 189            (?\e$,1Dq\e(B special "DIGIT ONE")
 190            (?\e$,1Dr\e(B special "DIGIT TWO")
 191            (?\e$,1Ds\e(B special "DIGIT THREE")
 192            (?\e$,1Dt\e(B special "DIGIT FOUR")
 193            (?\e$,1Du\e(B special "DIGIT FIVE")
 194            (?\e$,1Dv\e(B special "DIGIT SIX")
 195            (?\e$,1Dw\e(B special "DIGIT SEVEN")
 196            (?\e$,1Dx\e(B special "DIGIT EIGHT")
 197            (?\e$,1Dy\e(B special "DIGIT NINE")
 198            (?\e$,1D|\e(B consonant "LETTER NHOR NHUU" "MOUSE")
 199            (?\e$,1D}\e(B consonant "LETTER MHOR MHAR" "DOG")))
 200       elm)
 201   (while l
 202     (setq elm (car l) l (cdr l))
 203     (let ((char (car elm))
 204           (ptype (nth 1 elm)))
 205       (cond ((eq ptype 'consonant)
 206              (modify-category-entry char ?c lao-category-table))
 207             ((memq ptype '(vowel-upper vowel-lower))
 208              (modify-category-entry char ?v lao-category-table))
 209             ((eq ptype 'semivowel-lower)
 210              (modify-category-entry char ?s lao-category-table))
 211             ((eq ptype 'tone)
 212              (modify-category-entry char ?t lao-category-table)))
 213       (put-char-code-property char 'phonetic-type ptype)
 214       (put-char-code-property char 'name (nth 2 elm))
 215       (put-char-code-property char 'meaning (nth 3 elm)))))
 216
 217 ;; The general composing rules are as follows:
 218 ;;
 219 ;;                          T
 220 ;;       V        T         V                  T
 221 ;; CV -> C, CT -> C, CVT -> C, Cv -> C, CvT -> C
 222 ;;                                   v         v
 223 ;;                             T
 224 ;;        V         T          V                   T
 225 ;; CsV -> C, CsT -> C, CsVT -> C, Csv -> C, CvT -> C
 226 ;;        s         s          s         s         s
 227 ;;                                       v         v
 228
 229
 230 ;; where C: consonant, V: vowel upper, v: vowel lower,
 231 ;;       T: tone mark, s: semivowel lower
 232
 233 (defvar lao-composition-pattern
 234   "\\cc\\(\\ct\\|\\cv\\ct?\\|\\cs\\(\\ct\\|\\cv\\ct?\\)?\\)"
 235   "Regular expression matching a Lao composite sequence.")
 236
 237 ;;;###autoload
 238 (defun lao-compose-string (str)
 239   (with-category-table lao-category-table
 240    (let ((idx 0))
 241      (while (setq idx (string-match lao-composition-pattern str idx))
 242        (compose-string str idx (match-end 0))
 243        (setq idx (match-end 0))))
 244    str))
 245
 246 ;;; LRT: Lao <-> Roman Transcription
 247
 248 ;; Upper vowels and tone-marks are put on the letter.
 249 ;; Semi-vowel-sign-lo and lower vowels are put under the letter.
 250
 251 (defconst lao-transcription-consonant-alist
 252   (sort '(;; single consonants
 253           ("k" . "\e(1!\e(B")
 254           ("kh" . "\e(1"\e(B")
 255           ("qh" . "\e(1$\e(B")
 256           ("ng" . "\e(1'\e(B")
 257           ("j" . "\e(1(\e(B")
 258           ("s" . "\e(1J\e(B")
 259           ("x" . "\e(1*\e(B")
 260           ("y" . "\e(1-\e(B")
 261           ("d" . "\e(14\e(B")
 262           ("t" . "\e(15\e(B")
 263           ("th" . "\e(16\e(B")
 264           ("dh" . "\e(17\e(B")
 265           ("n" . "\e(19\e(B")
 266           ("b" . "\e(1:\e(B")
 267           ("p" . "\e(1;\e(B")
 268           ("hp" . "\e(1<\e(B")
 269           ("fh" . "\e(1=\e(B")
 270           ("ph" . "\e(1>\e(B")
 271           ("f" . "\e(1?\e(B")
 272           ("m" . "\e(1A\e(B")
 273           ("gn" . "\e(1B\e(B")
 274           ("l" . "\e(1E\e(B")
 275           ("r" . "\e(1C\e(B")
 276           ("v" . "\e(1G\e(B")
 277           ("w" . "\e(1G\e(B")
 278           ("hh" . "\e(1K\e(B")
 279           ("O" . "\e(1M\e(B")
 280           ("h" . "\e(1N\e(B")
 281           ("nh" . "\e(1|\e(B")
 282           ("mh" . "\e(1}\e(B")
 283           ("lh" . ["\e(1K\\e(B"])
 284           ;; double consonants
 285           ("ngh" . ["\e(1K'\e(B"])
 286           ("yh" . ["\e(1K]\e(B"])
 287           ("wh" . ["\e(1KG\e(B"])
 288           ("hl" . ["\e(1KE\e(B"])
 289           ("hy" . ["\e(1K-\e(B"])
 290           ("hn" . ["\e(1K9\e(B"])
 291           ("hm" . ["\e(1KA\e(B"])
 292           )
 293         (function (lambda (x y) (> (length (car x)) (length (car y)))))))
 294
 295 (defconst lao-transcription-semi-vowel-alist
 296   '(("r" . "\e(1\\e(B")))
 297
 298 (defconst lao-transcription-vowel-alist
 299   (sort '(("a" . "\e(1P\e(B")
 300           ("ar" . "\e(1R\e(B")
 301           ("i" . "\e(1T\e(B")
 302           ("ii" . "\e(1U\e(B")
 303           ("eu" . "\e(1V\e(B")
 304           ("ur" . "\e(1W\e(B")
 305           ("u" . "\e(1X\e(B")
 306           ("uu" . "\e(1Y\e(B")
 307           ("e" . ["\e(1`P\e(B"])
 308           ("ee" . "\e(1`\e(B")
 309           ("ae" . ["\e(1aP\e(B"])
 310           ("aa" . "\e(1a\e(B")
 311           ("o" . ["\e(1bP\e(B"])
 312           ("oo" . "\e(1b\e(B")
 313           ("oe" . ["\e(1`RP\e(B"])
 314           ("or" . "\e(1m\e(B")
 315           ("er" . ["\e(1`T\e(B"])
 316           ("ir" . ["\e(1`U\e(B"])
 317           ("ua" . ["\e(1[GP\e(B"])
 318           ("uaa" . ["\e(1[G\e(B"])
 319           ("ie" . ["\e(1`Q]P\e(B"])
 320           ("ia" . ["\e(1`Q]\e(B"])
 321           ("ea" . ["\e(1`VM\e(B"])
 322           ("eaa" . ["\e(1`WM\e(B"])
 323           ("ai" . "\e(1d\e(B")
 324           ("ei" . "\e(1c\e(B")
 325           ("ao" . ["\e(1`[R\e(B"])
 326           ("aM" . "\e(1S\e(B"))
 327         (function (lambda (x y) (> (length (car x)) (length (car y)))))))
 328
 329 ;; Maa-sakod is put at the tail.
 330 (defconst lao-transcription-maa-sakod-alist
 331   '(("k" . "\e(1!\e(B")
 332     ("g" . "\e(1'\e(B")
 333     ("y" . "\e(1-\e(B")
 334     ("d" . "\e(14\e(B")
 335     ("n" . "\e(19\e(B")
 336     ("b" . "\e(1:\e(B")
 337     ("m" . "\e(1A\e(B")
 338     ("v" . "\e(1G\e(B")
 339     ("w" . "\e(1G\e(B")
 340     ))
 341
 342 (defconst lao-transcription-tone-alist
 343   '(("'" . "\e(1h\e(B")
 344     ("\"" . "\e(1i\e(B")
 345     ("^" . "\e(1j\e(B")
 346     ("+" . "\e(1k\e(B")
 347     ("~" . "\e(1l\e(B")))
 348
 349 (defconst lao-transcription-punctuation-alist
 350   '(("\\0" . "\e(1p\e(B")
 351     ("\\1" . "\e(1q\e(B")
 352     ("\\2" . "\e(1r\e(B")
 353     ("\\3" . "\e(1s\e(B")
 354     ("\\4" . "\e(1t\e(B")
 355     ("\\5" . "\e(1u\e(B")
 356     ("\\6" . "\e(1v\e(B")
 357     ("\\7" . "\e(1w\e(B")
 358     ("\\8" . "\e(1x\e(B")
 359     ("\\9" . "\e(1y\e(B")
 360     ("\\\\" . "\e(1f\e(B")
 361     ("\\$" . "\e(1O\e(B")))
 362
 363 (defconst lao-transcription-pattern
 364   (concat
 365    "\\("
 366    (mapconcat 'car lao-transcription-consonant-alist "\\|")
 367    "\\)\\("
 368    (mapconcat 'car lao-transcription-semi-vowel-alist "\\|")
 369    "\\)?\\(\\("
 370    (mapconcat 'car lao-transcription-vowel-alist "\\|")
 371    "\\)\\("
 372    (mapconcat 'car lao-transcription-maa-sakod-alist "\\|")
 373    "\\)?\\("
 374    (mapconcat (lambda (x) (regexp-quote (car x)))
 375               lao-transcription-tone-alist "\\|")
 376    "\\)?\\)?\\|"
 377    (mapconcat (lambda (x) (regexp-quote (car x)))
 378               lao-transcription-punctuation-alist "\\|")
 379    )
 380   "Regexp of Roman transcription pattern for one Lao syllable.")
 381
 382 (defconst lao-transcription-pattern
 383   (concat
 384    "\\("
 385    (regexp-opt (mapcar 'car lao-transcription-consonant-alist))
 386    "\\)\\("
 387    (regexp-opt (mapcar 'car lao-transcription-semi-vowel-alist))
 388    "\\)?\\(\\("
 389    (regexp-opt (mapcar 'car lao-transcription-vowel-alist))
 390    "\\)\\("
 391    (regexp-opt (mapcar 'car lao-transcription-maa-sakod-alist))
 392    "\\)?\\("
 393    (regexp-opt (mapcar 'car lao-transcription-tone-alist))
 394    "\\)?\\)?\\|"
 395    (regexp-opt (mapcar 'car lao-transcription-punctuation-alist))
 396    )
 397   "Regexp of Roman transcription pattern for one Lao syllable.")
 398
 399 (defconst lao-vowel-reordering-rule
 400   '(("\e(1P\e(B" (0 ?\e(1P\e(B) (0 ?\e(1Q\e(B))
 401     ("\e(1R\e(B" (0 ?\e(1R\e(B))
 402     ("\e(1T\e(B" (0 ?\e(1U\e(B))
 403     ("\e(1U\e(B" (0 ?\e(1U\e(B))
 404     ("\e(1V\e(B" (0 ?\e(1V\e(B))
 405     ("\e(1W\e(B" (0 ?\e(1W\e(B))
 406     ("\e(1X\e(B" (0 ?\e(1X\e(B))
 407     ("\e(1Y\e(B" (0 ?\e(1Y\e(B))
 408     ("\e(1`P\e(B" (?\e(1`\e(B 0 ?\e(1P\e(B) (?\e(1`\e(B 0 ?\e(1Q\e(B))
 409     ("\e(1`\e(B" (?\e(1`\e(B 0))
 410     ("\e(1aP\e(B" (?\e(1a\e(B 0 ?\e(1P\e(B) (?\e(1a\e(B 0 ?\e(1Q\e(B))
 411     ("\e(1a\e(B" (?\e(1a\e(B 0))
 412     ("\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))
 413     ("\e(1b\e(B" (?\e(1b\e(B 0))
 414     ("\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))
 415     ("\e(1m\e(B" (0 ?\e(1m\e(B) (0 ?\e(1M\e(B))
 416     ("\e(1`T\e(B" (?\e(1`\e(B 0 ?\e(1T\e(B))
 417     ("\e(1`U\e(B" (?\e(1`\e(B 0 ?\e(1U\e(B))
 418     ("\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))
 419     ("\e(1[G\e(B" (0 ?\e(1[\e(B ?\e(1G\e(B) (0 ?\e(1G\e(B))
 420     ("\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))
 421     ("\e(1`Q]\e(B" (?\e(1`\e(B 0 ?\e(1Q\e(B ?\e(1]\e(B) (0 ?\e(1]\e(B))
 422     ("\e(1`VM\e(B" (?\e(1`\e(B 0 ?\e(1V\e(B ?\e(1M\e(B))
 423     ("\e(1`WM\e(B" (?\e(1`\e(B 0 ?\e(1W\e(B ?\e(1M\e(B))
 424     ("\e(1d\e(B" (?\e(1d\e(B 0))
 425     ("\e(1c\e(B" (?\e(1c\e(B 0))
 426     ("\e(1`[R\e(B" (?\e(1`\e(B 0 ?\e(1[\e(B ?\e(1R\e(B))
 427     ("\e(1S\e(B" (0 ?\e(1S\e(B))
 428
 429     ;; Unicode equivalents
 430     ("\e$,1DP\e(B" (0 ?\e$,1DP\e(B) (0 ?\e$,1DQ\e(B))
 431     ("\e$,1DR\e(B" (0 ?\e$,1DR\e(B))
 432     ("\e$,1DT\e(B" (0 ?\e$,1DU\e(B))
 433     ("\e$,1DU\e(B" (0 ?\e$,1DU\e(B))
 434     ("\e$,1DV\e(B" (0 ?\e$,1DV\e(B))
 435     ("\e$,1DW\e(B" (0 ?\e$,1DW\e(B))
 436     ("\e$,1DX\e(B" (0 ?\e$,1DX\e(B))
 437     ("\e$,1DY\e(B" (0 ?\e$,1DY\e(B))
 438     ("\e$,1D`DP\e(B" (?\e$,1D`\e(B 0 ?\e$,1DP\e(B) (?\e$,1D`\e(B 0 ?\e$,1DQ\e(B))
 439     ("\e$,1D`\e(B" (?\e$,1D`\e(B 0))
 440     ("\e$,1DaDP\e(B" (?\e$,1Da\e(B 0 ?\e$,1DP\e(B) (?\e$,1Da\e(B 0 ?\e$,1DQ\e(B))
 441     ("\e$,1Da\e(B" (?\e$,1Da\e(B 0))
 442     ("\e$,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$,1DQ\e(B))
 443     ("\e$,1Db\e(B" (?\e$,1Db\e(B 0))
 444     ("\e$,1D`DRDP\e(B" (?\e$,1D`\e(B 0 ?\e$,1DR\e(B ?\e$,1DP\e(B) (0 ?\e$,1DQ\e(B ?\e$,1DM\e(B))
 445     ("\e$,1Dm\e(B" (0 ?\e$,1Dm\e(B) (0 ?\e$,1DM\e(B))
 446     ("\e$,1D`DT\e(B" (?\e$,1D`\e(B 0 ?\e$,1DT\e(B))
 447     ("\e$,1D`DU\e(B" (?\e$,1D`\e(B 0 ?\e$,1DU\e(B))
 448     ("\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))
 449     ("\e$,1D[DG\e(B" (0 ?\e$,1D[\e(B ?\e$,1DG\e(B) (0 ?\e$,1DG\e(B))
 450     ("\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$,1D]\e(B))
 451     ("\e$,1D`DQD]\e(B" (?\e$,1D`\e(B 0 ?\e$,1DQ\e(B ?\e$,1D]\e(B) (0 ?\e$,1D]\e(B))
 452     ("\e$,1D`DVDM\e(B" (?\e$,1D`\e(B 0 ?\e$,1DV\e(B ?\e$,1DM\e(B))
 453     ("\e$,1D`DWDM\e(B" (?\e$,1D`\e(B 0 ?\e$,1DW\e(B ?\e$,1DM\e(B))
 454     ("\e$,1Dd\e(B" (?\e$,1Dd\e(B 0))
 455     ("\e$,1Dc\e(B" (?\e$,1Dc\e(B 0))
 456     ("\e$,1D`D[DR\e(B" (?\e$,1D`\e(B 0 ?\e$,1D[\e(B ?\e$,1DR\e(B))
 457     ("\e$,1DS\e(B" (0 ?\e$,1DS\e(B)))
 458   "Alist of Lao vowel string vs the corresponding re-ordering rule.
 459 Each element has this form:
 460         (VOWEL NO-MAA-SAKOD-RULE WITH-MAA-SAKOD-RULE (MAA-SAKOD-0 RULE-0) ...)
 461
 462 VOWEL is a vowel string (e.g. \"\e(1`Q]P\e(B\").
 463
 464 NO-MAA-SAKOD-RULE is a rule to re-order and modify VOWEL following a
 465 consonant.  It is a list vowel characters or 0.  The element 0
 466 indicate the place to embed a consonant.
 467
 468 Optional WITH-MAA-SAKOD-RULE is a rule to re-order and modify VOWEL
 469 follwoing a consonant and preceding a maa-sakod character.  If it is
 470 nil, NO-MAA-SAKOD-RULE is used.  The maa-sakod character is alwasy
 471 appended at the tail.
 472
 473 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
 474 string following a consonant `\e(1!\e(B' should be re-ordered as \"\e(1`!WM\e(B\".
 475
 476 Optional (MAA-SAKOD-n RULE-n) are rules specially applied to maa-sakod
 477 character MAA-SAKOD-n.")
 478
 479 ;;;###autoload
 480 (defun lao-transcribe-single-roman-syllable-to-lao (from to &optional str)
 481   "Transcribe a Romanized Lao syllable in the region FROM and TO to Lao string.
 482 Only the first syllable is transcribed.
 483 The value has the form: (START END LAO-STRING), where
 484 START and END are the beggining and end positions of the Roman Lao syllable,
 485 LAO-STRING is the Lao character transcription of it.
 486
 487 Optional 3rd arg STR, if non-nil, is a string to search for Roman Lao
 488 syllable.  In that case, FROM and TO are indexes to STR."
 489   (if str
 490       (if (setq from (string-match lao-transcription-pattern str from))
 491           (progn
 492             (if (>= from to)
 493                 (setq from nil)
 494               (setq to (match-end 0)))))
 495     (save-excursion
 496       (goto-char from)
 497       (if (setq to (re-search-forward lao-transcription-pattern to t))
 498           (setq from (match-beginning 0))
 499         (setq from nil))))
 500   (if from
 501       (let* ((consonant (match-string 1 str))
 502              (semivowel (match-string 3 str))
 503              (vowel (match-string 5 str))
 504              (maa-sakod (match-string 8 str))
 505              (tone (match-string 9 str))
 506              lao-consonant lao-semivowel lao-vowel lao-maa-sakod lao-tone
 507              clen cidx)
 508         (setq to (match-end 0))
 509         (if (not consonant)
 510             (setq str (cdr (assoc (match-string 0 str)
 511                                   lao-transcription-punctuation-alist)))
 512           (setq lao-consonant
 513                 (cdr (assoc consonant lao-transcription-consonant-alist)))
 514           (if (vectorp lao-consonant)
 515               (setq lao-consonant (aref lao-consonant 0)))
 516           (setq clen (length lao-consonant))
 517           (if semivowel
 518               ;; Include semivowel in STR.
 519               (setq lao-semivowel
 520                     (cdr (assoc semivowel lao-transcription-semi-vowel-alist))
 521                     str (if (= clen 1)
 522                             (concat lao-consonant lao-semivowel)
 523                           (concat (substring lao-consonant 0 1) lao-semivowel
 524                                   (substring lao-consonant 1))))
 525             (setq str lao-consonant))
 526           (if vowel
 527               (let (rule)
 528                 (setq lao-vowel
 529                       (cdr (assoc vowel lao-transcription-vowel-alist)))
 530                 (if (vectorp lao-vowel)
 531                     (setq lao-vowel (aref lao-vowel 0)))
 532                 (setq rule (assoc lao-vowel lao-vowel-reordering-rule))
 533                 (if (null maa-sakod)
 534                     (setq rule (nth 1 rule))
 535                   (setq lao-maa-sakod
 536                         (cdr (assoc maa-sakod lao-transcription-maa-sakod-alist))
 537                         rule
 538                         (or (cdr (assq (aref lao-maa-sakod 0) (nthcdr 2 rule)))
 539                             (nth 2 rule)
 540                             (nth 1 rule))))
 541                 (or rule
 542                     (error "Lao vowel %S has no re-ordering rule" lao-vowel))
 543                 (setq lao-consonant str str "")
 544                 (while rule
 545                   (if (= (car rule) 0)
 546                       (setq str (concat str lao-consonant)
 547                             cidx (length str))
 548                     (setq str (concat str (list (car rule)))))
 549                   (setq rule (cdr rule)))
 550                 (or cidx
 551                     (error "Lao vowel %S has malformed re-ordering rule" vowel))
 552                 ;; Set CIDX to after upper or lower vowel if any.
 553                 (let ((len (length str)))
 554                   (while (and (< cidx len)
 555                               (memq (get-char-code-property (aref str cidx)
 556                                                             'phonetic-type)
 557                                     '(vowel-lower vowel-upper)))
 558                     (setq cidx (1+ cidx))))
 559                 (if lao-maa-sakod
 560                     (setq str (concat str lao-maa-sakod)))
 561                 (if tone
 562                     (setq lao-tone
 563                           (cdr (assoc tone lao-transcription-tone-alist))
 564                           str (concat (substring str 0 cidx) lao-tone
 565                                       (substring str cidx)))))))
 566         (list from to (lao-compose-string str)))))
 567
 568 ;;;###autoload
 569 (defun lao-transcribe-roman-to-lao-string (str)
 570   "Transcribe Romanized Lao string STR to Lao character string."
 571   (let ((from 0)
 572         (to (length str))
 573         (lao-str "")
 574         val)
 575     (while (setq val (lao-transcribe-single-roman-syllable-to-lao from to str))
 576       (let ((start (car val))
 577             (end (nth 1 val))
 578             (lao (nth 2 val)))
 579         (if (> start from)
 580             (setq lao-str (concat lao-str (substring str from start) lao))
 581           (setq lao-str (concat lao-str lao)))
 582         (setq from end)))
 583     (if (< from to)
 584         (concat lao-str (substring str from to))
 585       lao-str)))
 586
 587 ;;;###autoload
 588 (defun lao-post-read-conversion (len)
 589   (lao-compose-region (point) (+ (point) len))
 590   len)
 591
 592 ;;;###autoload
 593 (defun lao-composition-function (from to pattern &optional string)
 594   "Compose Lao text in the region FROM and TO.
 595 The text matches the regular expression PATTERN.
 596 Optional 4th argument STRING, if non-nil, is a string containing text
 597 to compose.
 598
 599 The return value is number of composed characters."
 600   (if (< (1+ from) to)
 601       (progn
 602         (if string
 603             (compose-string string from to)
 604           (compose-region from to))
 605         (- to from))))
 606
 607 ;;;###autoload
 608 (defun lao-compose-region (from to)
 609   (interactive "r")
 610   (save-restriction
 611     (narrow-to-region from to)
 612     (goto-char (point-min))
 613     (with-category-table lao-category-table
 614       (while (re-search-forward lao-composition-pattern nil t)
 615         (compose-region (match-beginning 0) (point))))))
 616
 617 ;;
 618 (provide 'lao-util)
 619
 620 ;;; arch-tag: 1f828781-3cb8-4695-88af-8f33222338ce
 621 ;;; lao-util.el ends here