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, 2007, 2008
   4 ;;   Free Software Foundation, Inc.
   5 ;; Copyright (C) 1997, 1998, 1999, 2000, 2001, 2002, 2003, 2004, 2005, 2006,
   6 ;;   2007, 2008
   7 ;;   National Institute of Advanced Industrial Science and Technology (AIST)
   8 ;;   Registration Number H14PRO021
   9
  10 ;; Keywords: multilingual, Lao
  11
  12 ;; This file is part of GNU Emacs.
  13
  14 ;; GNU Emacs is free software; you can redistribute it and/or modify
  15 ;; it under the terms of the GNU General Public License as published by
  16 ;; the Free Software Foundation; either version 3, or (at your option)
  17 ;; any later version.
  18
  19 ;; GNU Emacs is distributed in the hope that it will be useful,
  20 ;; but WITHOUT ANY WARRANTY; without even the implied warranty of
  21 ;; MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
  22 ;; GNU General Public License for more details.
  23
  24 ;; You should have received a copy of the GNU General Public License
  25 ;; along with GNU Emacs; see the file COPYING.  If not, write to the
  26 ;; Free Software Foundation, Inc., 51 Franklin Street, Fifth Floor,
  27 ;; Boston, MA 02110-1301, USA.
  28
  29 ;;; Commentary:
  30
  31 ;;; Code:
  32
  33 ;; Setting information of Thai characters.
  34
  35 (defconst lao-category-table (make-category-table))
  36 (define-category ?c "Lao consonant" lao-category-table)
  37 (define-category ?s "Lao semi-vowel" lao-category-table)
  38 (define-category ?v "Lao upper/lower vowel" lao-category-table)
  39 (define-category ?t "Lao tone" lao-category-table)
  40
  41 (let ((l '((?\e(1!\e(B consonant "LETTER KOR  KAI'" "CHICKEN")
  42            (?\e(1"\e(B consonant "LETTER KHOR KHAI'" "EGG")
  43            (?\e(1#\e(B invalid nil)
  44            (?\e(1$\e(B consonant "LETTER QHOR QHWARGN" "BUFFALO")
  45            (?\e(1%\e(B invalid nil)
  46            (?  invalid nil)
  47            (?\e(1'\e(B consonant "LETTER NGOR NGUU" "SNAKE")
  48            (?\e(1(\e(B consonant "LETTER JOR JUA" "BUDDHIST NOVICE")
  49            (?\e(1)\e(B invalid nil)
  50            (?\e(1*\e(B consonant "LETTER XOR X\"ARNG" "ELEPHANT")
  51            (?\e(1+\e(B invalid nil)
  52            (?\e(1,\e(B invalid nil)
  53            (?\e(1-\e(B consonant "LETTER YOR YUNG" "MOSQUITO")
  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(1.\e(B invalid nil)
  60            (?\e(14\e(B consonant "LETTER DOR DANG" "NOSE")
  61            (?\e(15\e(B consonant "LETTER TOR TAR" "EYE")
  62            (?\e(16\e(B consonant "LETTER THOR THUNG" "TO ASK,QUESTION")
  63            (?\e(17\e(B consonant "LETTER DHOR DHARM" "FLAG")
  64            (?\e(18\e(B invalid nil)
  65            (?\e(19\e(B consonant "LETTER NOR NOK" "BIRD")
  66            (?\e(1:\e(B consonant "LETTER BOR BED" "FISHHOOK")
  67            (?\e(1;\e(B consonant "LETTER POR PAR" "FISH")
  68            (?\e(1<\e(B consonant "LETTER HPOR HPER\"" "BEE")
  69            (?\e(1=\e(B consonant "LETTER FHOR FHAR" "WALL")
  70            (?\e(1>\e(B consonant "LETTER PHOR PHUU" "MOUNTAIN")
  71            (?\e(1?\e(B consonant "LETTER FOR FAI" "FIRE")
  72            (?\e(1@\e(B invalid nil)
  73            (?\e(1A\e(B consonant "LETTER MOR MAR\"" "HORSE")
  74            (?\e(1B\e(B consonant "LETTER GNOR GNAR" "MEDICINE")
  75            (?\e(1C\e(B consonant "LETTER ROR ROD" "CAR")
  76            (?\e(1D\e(B invalid nil)
  77            (?\e(1E\e(B consonant "LETTER LOR LIING" "MONKEY")
  78            (?\e(1F\e(B invalid nil)
  79            (?\e(1G\e(B consonant "LETTER WOR WII" "HAND FAN")
  80            (?\e(1H\e(B invalid nil)
  81            (?\e(1I\e(B invalid nil)
  82            (?\e(1J\e(B consonant "LETTER SOR SEA" "TIGER")
  83            (?\e(1K\e(B consonant "LETTER HHOR HHAI" "JAR")
  84            (?\e(1L\e(B invalid nil)
  85            (?\e(1M\e(B consonant "LETTER OR OOW" "TAKE")
  86            (?\e(1N\e(B consonant "LETTER HOR HEA" "BOAT")
  87            (?\e(1O\e(B special "ELLIPSIS")
  88            (?\e(1P\e(B vowel-base "VOWEL SIGN SARA A")
  89            (?\e(1Q\e(B vowel-upper "VOWEL SIGN MAI KAN")
  90            (?\e(1R\e(B vowel-base "VOWEL SIGN SARA AR")
  91            (?\e(1S\e(B vowel-base "VOWEL SIGN SARA AM")
  92            (?\e(1T\e(B vowel-upper "VOWEL SIGN SARA I")
  93            (?\e(1U\e(B vowel-upper "VOWEL SIGN SARA II")
  94            (?\e(1V\e(B vowel-upper "VOWEL SIGN SARA EU")
  95            (?\e(1W\e(B vowel-upper "VOWEL SIGN SARA UR")
  96            (?\e(1X\e(B vowel-lower "VOWEL SIGN SARA U")
  97            (?\e(1Y\e(B vowel-lower "VOWEL SIGN SARA UU")
  98            (?\e(1Z\e(B invalid nil)
  99            (?\e(1[\e(B vowel-upper "VOWEL SIGN MAI KONG")
 100            (?\e(1\\e(B semivowel-lower "SEMIVOWEL SIGN LO")
 101            (?\e(1]\e(B vowel-base "SEMIVOWEL SIGN SARA IA")
 102            (?\e(1^\e(B invalid nil)
 103            (?\e(1_\e(B invalid nil)
 104            (?\e(1`\e(B vowel-base "VOWEL SIGN SARA EE")
 105            (?\e(1a\e(B vowel-base "VOWEL SIGN SARA AA")
 106            (?\e(1b\e(B vowel-base "VOWEL SIGN SARA OO")
 107            (?\e(1c\e(B vowel-base "VOWEL SIGN SARA EI MAI MUAN\"")
 108            (?\e(1d\e(B vowel-base "VOWEL SIGN SARA AI MAI MAY")
 109            (?\e(1e\e(B invalid nil)
 110            (?\e(1f\e(B special "KO LA (REPETITION)")
 111            (?\e(1g\e(B invalid nil)
 112            (?\e(1h\e(B tone "TONE MAI EK")
 113            (?\e(1i\e(B tone "TONE MAI THO")
 114            (?\e(1j\e(B tone "TONE MAI TI")
 115            (?\e(1k\e(B tone "TONE MAI JADTAWAR")
 116            (?\e(1l\e(B tone "CANCELLATION MARK")
 117            (?\e(1m\e(B vowel-upper "VOWEL SIGN SARA OR")
 118            (?\e(1n\e(B invalid nil)
 119            (?\e(1o\e(B invalid nil)
 120            (?\e(1p\e(B special "DIGIT ZERO")
 121            (?\e(1q\e(B special "DIGIT ONE")
 122            (?\e(1r\e(B special "DIGIT TWO")
 123            (?\e(1s\e(B special "DIGIT THREE")
 124            (?\e(1t\e(B special "DIGIT FOUR")
 125            (?\e(1u\e(B special "DIGIT FIVE")
 126            (?\e(1v\e(B special "DIGIT SIX")
 127            (?\e(1w\e(B special "DIGIT SEVEN")
 128            (?\e(1x\e(B special "DIGIT EIGHT")
 129            (?\e(1y\e(B special "DIGIT NINE")
 130            (?\e(1z\e(B invalid nil)
 131            (?\e(1{\e(B invalid nil)
 132            (?\e(1|\e(B consonant "LETTER NHOR NHUU" "MOUSE")
 133            (?\e(1}\e(B consonant "LETTER MHOR MHAR" "DOG")
 134            (?\e(1~\e(B invalid nil)
 135            ;; Unicode equivalents
 136            (?\e$,1D!\e(B consonant "LETTER KOR  KAI'" "CHICKEN")
 137            (?\e$,1D"\e(B consonant "LETTER KHOR KHAI'" "EGG")
 138            (?\e$,1D$\e(B consonant "LETTER QHOR QHWARGN" "BUFFALO")
 139            (?\e$,1D'\e(B consonant "LETTER NGOR NGUU" "SNAKE")
 140            (?\e$,1D(\e(B consonant "LETTER JOR JUA" "BUDDHIST NOVICE")
 141            (?\e$,1D*\e(B consonant "LETTER XOR X\"ARNG" "ELEPHANT")
 142            (?\e$,1D-\e(B consonant "LETTER YOR YUNG" "MOSQUITO")
 143            (?\e$,1D4\e(B consonant "LETTER DOR DANG" "NOSE")
 144            (?\e$,1D5\e(B consonant "LETTER TOR TAR" "EYE")
 145            (?\e$,1D6\e(B consonant "LETTER THOR THUNG" "TO ASK,QUESTION")
 146            (?\e$,1D7\e(B consonant "LETTER DHOR DHARM" "FLAG")
 147            (?\e$,1D9\e(B consonant "LETTER NOR NOK" "BIRD")
 148            (?\e$,1D:\e(B consonant "LETTER BOR BED" "FISHHOOK")
 149            (?\e$,1D;\e(B consonant "LETTER POR PAR" "FISH")
 150            (?\e$,1D<\e(B consonant "LETTER HPOR HPER\"" "BEE")
 151            (?\e$,1D=\e(B consonant "LETTER FHOR FHAR" "WALL")
 152            (?\e$,1D>\e(B consonant "LETTER PHOR PHUU" "MOUNTAIN")
 153            (?\e$,1D?\e(B consonant "LETTER FOR FAI" "FIRE")
 154            (?\e$,1DA\e(B consonant "LETTER MOR MAR\"" "HORSE")
 155            (?\e$,1DB\e(B consonant "LETTER GNOR GNAR" "MEDICINE")
 156            (?\e$,1DC\e(B consonant "LETTER ROR ROD" "CAR")
 157            (?\e$,1DE\e(B consonant "LETTER LOR LIING" "MONKEY")
 158            (?\e$,1DG\e(B consonant "LETTER WOR WII" "HAND FAN")
 159            (?\e$,1DJ\e(B consonant "LETTER SOR SEA" "TIGER")
 160            (?\e$,1DK\e(B consonant "LETTER HHOR HHAI" "JAR")
 161            (?\e$,1DM\e(B consonant "LETTER OR OOW" "TAKE")
 162            (?\e$,1DN\e(B consonant "LETTER HOR HEA" "BOAT")
 163            (?\e$,1DO\e(B special "ELLIPSIS")
 164            (?\e$,1DP\e(B vowel-base "VOWEL SIGN SARA A")
 165            (?\e$,1DQ\e(B vowel-upper "VOWEL SIGN MAI KAN")
 166            (?\e$,1DR\e(B vowel-base "VOWEL SIGN SARA AR")
 167            (?\e$,1DS\e(B vowel-base "VOWEL SIGN SARA AM")
 168            (?\e$,1DT\e(B vowel-upper "VOWEL SIGN SARA I")
 169            (?\e$,1DU\e(B vowel-upper "VOWEL SIGN SARA II")
 170            (?\e$,1DV\e(B vowel-upper "VOWEL SIGN SARA EU")
 171            (?\e$,1DW\e(B vowel-upper "VOWEL SIGN SARA UR")
 172            (?\e$,1DX\e(B vowel-lower "VOWEL SIGN SARA U")
 173            (?\e$,1DY\e(B vowel-lower "VOWEL SIGN SARA UU")
 174            (?\e$,1D[\e(B vowel-upper "VOWEL SIGN MAI KONG")
 175            (?\e$,1D\\e(B semivowel-lower "SEMIVOWEL SIGN LO")
 176            (?\e$,1D]\e(B vowel-base "SEMIVOWEL SIGN SARA IA")
 177            (?\e$,1D`\e(B vowel-base "VOWEL SIGN SARA EE")
 178            (?\e$,1Da\e(B vowel-base "VOWEL SIGN SARA AA")
 179            (?\e$,1Db\e(B vowel-base "VOWEL SIGN SARA OO")
 180            (?\e$,1Dc\e(B vowel-base "VOWEL SIGN SARA EI MAI MUAN\"")
 181            (?\e$,1Dd\e(B vowel-base "VOWEL SIGN SARA AI MAI MAY")
 182            (?\e$,1Df\e(B special "KO LA (REPETITION)")
 183            (?\e$,1Dh\e(B tone "TONE MAI EK")
 184            (?\e$,1Di\e(B tone "TONE MAI THO")
 185            (?\e$,1Dj\e(B tone "TONE MAI TI")
 186            (?\e$,1Dk\e(B tone "TONE MAI JADTAWAR")
 187            (?\e$,1Dl\e(B tone "CANCELLATION MARK")
 188            (?\e$,1Dm\e(B vowel-upper "VOWEL SIGN SARA OR")
 189            (?\e$,1Dp\e(B special "DIGIT ZERO")
 190            (?\e$,1Dq\e(B special "DIGIT ONE")
 191            (?\e$,1Dr\e(B special "DIGIT TWO")
 192            (?\e$,1Ds\e(B special "DIGIT THREE")
 193            (?\e$,1Dt\e(B special "DIGIT FOUR")
 194            (?\e$,1Du\e(B special "DIGIT FIVE")
 195            (?\e$,1Dv\e(B special "DIGIT SIX")
 196            (?\e$,1Dw\e(B special "DIGIT SEVEN")
 197            (?\e$,1Dx\e(B special "DIGIT EIGHT")
 198            (?\e$,1Dy\e(B special "DIGIT NINE")
 199            (?\e$,1D|\e(B consonant "LETTER NHOR NHUU" "MOUSE")
 200            (?\e$,1D}\e(B consonant "LETTER MHOR MHAR" "DOG")))
 201       elm)
 202   (while l
 203     (setq elm (car l) l (cdr l))
 204     (let ((char (car elm))
 205           (ptype (nth 1 elm)))
 206       (cond ((eq ptype 'consonant)
 207              (modify-category-entry char ?c lao-category-table))
 208             ((memq ptype '(vowel-upper vowel-lower))
 209              (modify-category-entry char ?v lao-category-table))
 210             ((eq ptype 'semivowel-lower)
 211              (modify-category-entry char ?s lao-category-table))
 212             ((eq ptype 'tone)
 213              (modify-category-entry char ?t lao-category-table)))
 214       (put-char-code-property char 'phonetic-type ptype)
 215       (put-char-code-property char 'name (nth 2 elm))
 216       (put-char-code-property char 'meaning (nth 3 elm)))))
 217
 218 ;; The general composing rules are as follows:
 219 ;;
 220 ;;                          T
 221 ;;       V        T         V                  T
 222 ;; CV -> C, CT -> C, CVT -> C, Cv -> C, CvT -> C
 223 ;;                                   v         v
 224 ;;                             T
 225 ;;        V         T          V                   T
 226 ;; CsV -> C, CsT -> C, CsVT -> C, Csv -> C, CvT -> C
 227 ;;        s         s          s         s         s
 228 ;;                                       v         v
 229
 230
 231 ;; where C: consonant, V: vowel upper, v: vowel lower,
 232 ;;       T: tone mark, s: semivowel lower
 233
 234 (defvar lao-composition-pattern
 235   "\\cc\\(\\ct\\|\\cv\\ct?\\|\\cs\\(\\ct\\|\\cv\\ct?\\)?\\)"
 236   "Regular expression matching a Lao composite sequence.")
 237
 238 ;;;###autoload
 239 (defun lao-compose-string (str)
 240   (with-category-table lao-category-table
 241    (let ((idx 0))
 242      (while (setq idx (string-match lao-composition-pattern str idx))
 243        (compose-string str idx (match-end 0))
 244        (setq idx (match-end 0))))
 245    str))
 246
 247 ;;; LRT: Lao <-> Roman Transcription
 248
 249 ;; Upper vowels and tone-marks are put on the letter.
 250 ;; Semi-vowel-sign-lo and lower vowels are put under the letter.
 251
 252 (defconst lao-transcription-consonant-alist
 253   (sort '(;; single consonants
 254           ("k" . "\e(1!\e(B")
 255           ("kh" . "\e(1"\e(B")
 256           ("qh" . "\e(1$\e(B")
 257           ("ng" . "\e(1'\e(B")
 258           ("j" . "\e(1(\e(B")
 259           ("s" . "\e(1J\e(B")
 260           ("x" . "\e(1*\e(B")
 261           ("y" . "\e(1-\e(B")
 262           ("d" . "\e(14\e(B")
 263           ("t" . "\e(15\e(B")
 264           ("th" . "\e(16\e(B")
 265           ("dh" . "\e(17\e(B")
 266           ("n" . "\e(19\e(B")
 267           ("b" . "\e(1:\e(B")
 268           ("p" . "\e(1;\e(B")
 269           ("hp" . "\e(1<\e(B")
 270           ("fh" . "\e(1=\e(B")
 271           ("ph" . "\e(1>\e(B")
 272           ("f" . "\e(1?\e(B")
 273           ("m" . "\e(1A\e(B")
 274           ("gn" . "\e(1B\e(B")
 275           ("l" . "\e(1E\e(B")
 276           ("r" . "\e(1C\e(B")
 277           ("v" . "\e(1G\e(B")
 278           ("w" . "\e(1G\e(B")
 279           ("hh" . "\e(1K\e(B")
 280           ("O" . "\e(1M\e(B")
 281           ("h" . "\e(1N\e(B")
 282           ("nh" . "\e(1|\e(B")
 283           ("mh" . "\e(1}\e(B")
 284           ("lh" . ["\e(1K\\e(B"])
 285           ;; double consonants
 286           ("ngh" . ["\e(1K'\e(B"])
 287           ("yh" . ["\e(1K]\e(B"])
 288           ("wh" . ["\e(1KG\e(B"])
 289           ("hl" . ["\e(1KE\e(B"])
 290           ("hy" . ["\e(1K-\e(B"])
 291           ("hn" . ["\e(1K9\e(B"])
 292           ("hm" . ["\e(1KA\e(B"])
 293           )
 294         (function (lambda (x y) (> (length (car x)) (length (car y)))))))
 295
 296 (defconst lao-transcription-semi-vowel-alist
 297   '(("r" . "\e(1\\e(B")))
 298
 299 (defconst lao-transcription-vowel-alist
 300   (sort '(("a" . "\e(1P\e(B")
 301           ("ar" . "\e(1R\e(B")
 302           ("i" . "\e(1T\e(B")
 303           ("ii" . "\e(1U\e(B")
 304           ("eu" . "\e(1V\e(B")
 305           ("ur" . "\e(1W\e(B")
 306           ("u" . "\e(1X\e(B")
 307           ("uu" . "\e(1Y\e(B")
 308           ("e" . ["\e(1`P\e(B"])
 309           ("ee" . "\e(1`\e(B")
 310           ("ae" . ["\e(1aP\e(B"])
 311           ("aa" . "\e(1a\e(B")
 312           ("o" . ["\e(1bP\e(B"])
 313           ("oo" . "\e(1b\e(B")
 314           ("oe" . ["\e(1`RP\e(B"])
 315           ("or" . "\e(1m\e(B")
 316           ("er" . ["\e(1`T\e(B"])
 317           ("ir" . ["\e(1`U\e(B"])
 318           ("ua" . ["\e(1[GP\e(B"])
 319           ("uaa" . ["\e(1[G\e(B"])
 320           ("ie" . ["\e(1`Q]P\e(B"])
 321           ("ia" . ["\e(1`Q]\e(B"])
 322           ("ea" . ["\e(1`VM\e(B"])
 323           ("eaa" . ["\e(1`WM\e(B"])
 324           ("ai" . "\e(1d\e(B")
 325           ("ei" . "\e(1c\e(B")
 326           ("ao" . ["\e(1`[R\e(B"])
 327           ("aM" . "\e(1S\e(B"))
 328         (function (lambda (x y) (> (length (car x)) (length (car y)))))))
 329
 330 ;; Maa-sakod is put at the tail.
 331 (defconst lao-transcription-maa-sakod-alist
 332   '(("k" . "\e(1!\e(B")
 333     ("g" . "\e(1'\e(B")
 334     ("y" . "\e(1-\e(B")
 335     ("d" . "\e(14\e(B")
 336     ("n" . "\e(19\e(B")
 337     ("b" . "\e(1:\e(B")
 338     ("m" . "\e(1A\e(B")
 339     ("v" . "\e(1G\e(B")
 340     ("w" . "\e(1G\e(B")
 341     ))
 342
 343 (defconst lao-transcription-tone-alist
 344   '(("'" . "\e(1h\e(B")
 345     ("\"" . "\e(1i\e(B")
 346     ("^" . "\e(1j\e(B")
 347     ("+" . "\e(1k\e(B")
 348     ("~" . "\e(1l\e(B")))
 349
 350 (defconst lao-transcription-punctuation-alist
 351   '(("\\0" . "\e(1p\e(B")
 352     ("\\1" . "\e(1q\e(B")
 353     ("\\2" . "\e(1r\e(B")
 354     ("\\3" . "\e(1s\e(B")
 355     ("\\4" . "\e(1t\e(B")
 356     ("\\5" . "\e(1u\e(B")
 357     ("\\6" . "\e(1v\e(B")
 358     ("\\7" . "\e(1w\e(B")
 359     ("\\8" . "\e(1x\e(B")
 360     ("\\9" . "\e(1y\e(B")
 361     ("\\\\" . "\e(1f\e(B")
 362     ("\\$" . "\e(1O\e(B")))
 363
 364 (defconst lao-transcription-pattern
 365   (concat
 366    "\\("
 367    (mapconcat 'car lao-transcription-consonant-alist "\\|")
 368    "\\)\\("
 369    (mapconcat 'car lao-transcription-semi-vowel-alist "\\|")
 370    "\\)?\\(\\("
 371    (mapconcat 'car lao-transcription-vowel-alist "\\|")
 372    "\\)\\("
 373    (mapconcat 'car lao-transcription-maa-sakod-alist "\\|")
 374    "\\)?\\("
 375    (mapconcat (lambda (x) (regexp-quote (car x)))
 376               lao-transcription-tone-alist "\\|")
 377    "\\)?\\)?\\|"
 378    (mapconcat (lambda (x) (regexp-quote (car x)))
 379               lao-transcription-punctuation-alist "\\|")
 380    )
 381   "Regexp of Roman transcription pattern for one Lao syllable.")
 382
 383 (defconst lao-transcription-pattern
 384   (concat
 385    "\\("
 386    (regexp-opt (mapcar 'car lao-transcription-consonant-alist))
 387    "\\)\\("
 388    (regexp-opt (mapcar 'car lao-transcription-semi-vowel-alist))
 389    "\\)?\\(\\("
 390    (regexp-opt (mapcar 'car lao-transcription-vowel-alist))
 391    "\\)\\("
 392    (regexp-opt (mapcar 'car lao-transcription-maa-sakod-alist))
 393    "\\)?\\("
 394    (regexp-opt (mapcar 'car lao-transcription-tone-alist))
 395    "\\)?\\)?\\|"
 396    (regexp-opt (mapcar 'car lao-transcription-punctuation-alist))
 397    )
 398   "Regexp of Roman transcription pattern for one Lao syllable.")
 399
 400 (defconst lao-vowel-reordering-rule
 401   '(("\e(1P\e(B" (0 ?\e(1P\e(B) (0 ?\e(1Q\e(B))
 402     ("\e(1R\e(B" (0 ?\e(1R\e(B))
 403     ("\e(1T\e(B" (0 ?\e(1U\e(B))
 404     ("\e(1U\e(B" (0 ?\e(1U\e(B))
 405     ("\e(1V\e(B" (0 ?\e(1V\e(B))
 406     ("\e(1W\e(B" (0 ?\e(1W\e(B))
 407     ("\e(1X\e(B" (0 ?\e(1X\e(B))
 408     ("\e(1Y\e(B" (0 ?\e(1Y\e(B))
 409     ("\e(1`P\e(B" (?\e(1`\e(B 0 ?\e(1P\e(B) (?\e(1`\e(B 0 ?\e(1Q\e(B))
 410     ("\e(1`\e(B" (?\e(1`\e(B 0))
 411     ("\e(1aP\e(B" (?\e(1a\e(B 0 ?\e(1P\e(B) (?\e(1a\e(B 0 ?\e(1Q\e(B))
 412     ("\e(1a\e(B" (?\e(1a\e(B 0))
 413     ("\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))
 414     ("\e(1b\e(B" (?\e(1b\e(B 0))
 415     ("\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))
 416     ("\e(1m\e(B" (0 ?\e(1m\e(B) (0 ?\e(1M\e(B))
 417     ("\e(1`T\e(B" (?\e(1`\e(B 0 ?\e(1T\e(B))
 418     ("\e(1`U\e(B" (?\e(1`\e(B 0 ?\e(1U\e(B))
 419     ("\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))
 420     ("\e(1[G\e(B" (0 ?\e(1[\e(B ?\e(1G\e(B) (0 ?\e(1G\e(B))
 421     ("\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))
 422     ("\e(1`Q]\e(B" (?\e(1`\e(B 0 ?\e(1Q\e(B ?\e(1]\e(B) (0 ?\e(1]\e(B))
 423     ("\e(1`VM\e(B" (?\e(1`\e(B 0 ?\e(1V\e(B ?\e(1M\e(B))
 424     ("\e(1`WM\e(B" (?\e(1`\e(B 0 ?\e(1W\e(B ?\e(1M\e(B))
 425     ("\e(1d\e(B" (?\e(1d\e(B 0))
 426     ("\e(1c\e(B" (?\e(1c\e(B 0))
 427     ("\e(1`[R\e(B" (?\e(1`\e(B 0 ?\e(1[\e(B ?\e(1R\e(B))
 428     ("\e(1S\e(B" (0 ?\e(1S\e(B))
 429
 430     ;; Unicode equivalents
 431     ("\e$,1DP\e(B" (0 ?\e$,1DP\e(B) (0 ?\e$,1DQ\e(B))
 432     ("\e$,1DR\e(B" (0 ?\e$,1DR\e(B))
 433     ("\e$,1DT\e(B" (0 ?\e$,1DU\e(B))
 434     ("\e$,1DU\e(B" (0 ?\e$,1DU\e(B))
 435     ("\e$,1DV\e(B" (0 ?\e$,1DV\e(B))
 436     ("\e$,1DW\e(B" (0 ?\e$,1DW\e(B))
 437     ("\e$,1DX\e(B" (0 ?\e$,1DX\e(B))
 438     ("\e$,1DY\e(B" (0 ?\e$,1DY\e(B))
 439     ("\e$,1D`DP\e(B" (?\e$,1D`\e(B 0 ?\e$,1DP\e(B) (?\e$,1D`\e(B 0 ?\e$,1DQ\e(B))
 440     ("\e$,1D`\e(B" (?\e$,1D`\e(B 0))
 441     ("\e$,1DaDP\e(B" (?\e$,1Da\e(B 0 ?\e$,1DP\e(B) (?\e$,1Da\e(B 0 ?\e$,1DQ\e(B))
 442     ("\e$,1Da\e(B" (?\e$,1Da\e(B 0))
 443     ("\e$,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$,1DQ\e(B))
 444     ("\e$,1Db\e(B" (?\e$,1Db\e(B 0))
 445     ("\e$,1D`DRDP\e(B" (?\e$,1D`\e(B 0 ?\e$,1DR\e(B ?\e$,1DP\e(B) (0 ?\e$,1DQ\e(B ?\e$,1DM\e(B))
 446     ("\e$,1Dm\e(B" (0 ?\e$,1Dm\e(B) (0 ?\e$,1DM\e(B))
 447     ("\e$,1D`DT\e(B" (?\e$,1D`\e(B 0 ?\e$,1DT\e(B))
 448     ("\e$,1D`DU\e(B" (?\e$,1D`\e(B 0 ?\e$,1DU\e(B))
 449     ("\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))
 450     ("\e$,1D[DG\e(B" (0 ?\e$,1D[\e(B ?\e$,1DG\e(B) (0 ?\e$,1DG\e(B))
 451     ("\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$,1D]\e(B))
 452     ("\e$,1D`DQD]\e(B" (?\e$,1D`\e(B 0 ?\e$,1DQ\e(B ?\e$,1D]\e(B) (0 ?\e$,1D]\e(B))
 453     ("\e$,1D`DVDM\e(B" (?\e$,1D`\e(B 0 ?\e$,1DV\e(B ?\e$,1DM\e(B))
 454     ("\e$,1D`DWDM\e(B" (?\e$,1D`\e(B 0 ?\e$,1DW\e(B ?\e$,1DM\e(B))
 455     ("\e$,1Dd\e(B" (?\e$,1Dd\e(B 0))
 456     ("\e$,1Dc\e(B" (?\e$,1Dc\e(B 0))
 457     ("\e$,1D`D[DR\e(B" (?\e$,1D`\e(B 0 ?\e$,1D[\e(B ?\e$,1DR\e(B))
 458     ("\e$,1DS\e(B" (0 ?\e$,1DS\e(B)))
 459   "Alist of Lao vowel string vs the corresponding re-ordering rule.
 460 Each element has this form:
 461         (VOWEL NO-MAA-SAKOD-RULE WITH-MAA-SAKOD-RULE (MAA-SAKOD-0 RULE-0) ...)
 462
 463 VOWEL is a vowel string (e.g. \"\e(1`Q]P\e(B\").
 464
 465 NO-MAA-SAKOD-RULE is a rule to re-order and modify VOWEL following a
 466 consonant.  It is a list vowel characters or 0.  The element 0
 467 indicate the place to embed a consonant.
 468
 469 Optional WITH-MAA-SAKOD-RULE is a rule to re-order and modify VOWEL
 470 follwoing a consonant and preceding a maa-sakod character.  If it is
 471 nil, NO-MAA-SAKOD-RULE is used.  The maa-sakod character is alwasy
 472 appended at the tail.
 473
 474 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
 475 string following a consonant `\e(1!\e(B' should be re-ordered as \"\e(1`!WM\e(B\".
 476
 477 Optional (MAA-SAKOD-n RULE-n) are rules specially applied to maa-sakod
 478 character MAA-SAKOD-n.")
 479
 480 ;;;###autoload
 481 (defun lao-transcribe-single-roman-syllable-to-lao (from to &optional str)
 482   "Transcribe a Romanized Lao syllable in the region FROM and TO to Lao string.
 483 Only the first syllable is transcribed.
 484 The value has the form: (START END LAO-STRING), where
 485 START and END are the beggining and end positions of the Roman Lao syllable,
 486 LAO-STRING is the Lao character transcription of it.
 487
 488 Optional 3rd arg STR, if non-nil, is a string to search for Roman Lao
 489 syllable.  In that case, FROM and TO are indexes to STR."
 490   (if str
 491       (if (setq from (string-match lao-transcription-pattern str from))
 492           (progn
 493             (if (>= from to)
 494                 (setq from nil)
 495               (setq to (match-end 0)))))
 496     (save-excursion
 497       (goto-char from)
 498       (if (setq to (re-search-forward lao-transcription-pattern to t))
 499           (setq from (match-beginning 0))
 500         (setq from nil))))
 501   (if from
 502       (let* ((consonant (match-string 1 str))
 503              (semivowel (match-string 3 str))
 504              (vowel (match-string 5 str))
 505              (maa-sakod (match-string 8 str))
 506              (tone (match-string 9 str))
 507              lao-consonant lao-semivowel lao-vowel lao-maa-sakod lao-tone
 508              clen cidx)
 509         (setq to (match-end 0))
 510         (if (not consonant)
 511             (setq str (cdr (assoc (match-string 0 str)
 512                                   lao-transcription-punctuation-alist)))
 513           (setq lao-consonant
 514                 (cdr (assoc consonant lao-transcription-consonant-alist)))
 515           (if (vectorp lao-consonant)
 516               (setq lao-consonant (aref lao-consonant 0)))
 517           (setq clen (length lao-consonant))
 518           (if semivowel
 519               ;; Include semivowel in STR.
 520               (setq lao-semivowel
 521                     (cdr (assoc semivowel lao-transcription-semi-vowel-alist))
 522                     str (if (= clen 1)
 523                             (concat lao-consonant lao-semivowel)
 524                           (concat (substring lao-consonant 0 1) lao-semivowel
 525                                   (substring lao-consonant 1))))
 526             (setq str lao-consonant))
 527           (if vowel
 528               (let (rule)
 529                 (setq lao-vowel
 530                       (cdr (assoc vowel lao-transcription-vowel-alist)))
 531                 (if (vectorp lao-vowel)
 532                     (setq lao-vowel (aref lao-vowel 0)))
 533                 (setq rule (assoc lao-vowel lao-vowel-reordering-rule))
 534                 (if (null maa-sakod)
 535                     (setq rule (nth 1 rule))
 536                   (setq lao-maa-sakod
 537                         (cdr (assoc maa-sakod lao-transcription-maa-sakod-alist))
 538                         rule
 539                         (or (cdr (assq (aref lao-maa-sakod 0) (nthcdr 2 rule)))
 540                             (nth 2 rule)
 541                             (nth 1 rule))))
 542                 (or rule
 543                     (error "Lao vowel %S has no re-ordering rule" lao-vowel))
 544                 (setq lao-consonant str str "")
 545                 (while rule
 546                   (if (= (car rule) 0)
 547                       (setq str (concat str lao-consonant)
 548                             cidx (length str))
 549                     (setq str (concat str (list (car rule)))))
 550                   (setq rule (cdr rule)))
 551                 (or cidx
 552                     (error "Lao vowel %S has malformed re-ordering rule" vowel))
 553                 ;; Set CIDX to after upper or lower vowel if any.
 554                 (let ((len (length str)))
 555                   (while (and (< cidx len)
 556                               (memq (get-char-code-property (aref str cidx)
 557                                                             'phonetic-type)
 558                                     '(vowel-lower vowel-upper)))
 559                     (setq cidx (1+ cidx))))
 560                 (if lao-maa-sakod
 561                     (setq str (concat str lao-maa-sakod)))
 562                 (if tone
 563                     (setq lao-tone
 564                           (cdr (assoc tone lao-transcription-tone-alist))
 565                           str (concat (substring str 0 cidx) lao-tone
 566                                       (substring str cidx)))))))
 567         (list from to (lao-compose-string str)))))
 568
 569 ;;;###autoload
 570 (defun lao-transcribe-roman-to-lao-string (str)
 571   "Transcribe Romanized Lao string STR to Lao character string."
 572   (let ((from 0)
 573         (to (length str))
 574         (lao-str "")
 575         val)
 576     (while (setq val (lao-transcribe-single-roman-syllable-to-lao from to str))
 577       (let ((start (car val))
 578             (end (nth 1 val))
 579             (lao (nth 2 val)))
 580         (if (> start from)
 581             (setq lao-str (concat lao-str (substring str from start) lao))
 582           (setq lao-str (concat lao-str lao)))
 583         (setq from end)))
 584     (if (< from to)
 585         (concat lao-str (substring str from to))
 586       lao-str)))
 587
 588 ;;;###autoload
 589 (defun lao-post-read-conversion (len)
 590   (lao-compose-region (point) (+ (point) len))
 591   len)
 592
 593 ;;;###autoload
 594 (defun lao-composition-function (from to pattern &optional string)
 595   "Compose Lao text in the region FROM and TO.
 596 The text matches the regular expression PATTERN.
 597 Optional 4th argument STRING, if non-nil, is a string containing text
 598 to compose.
 599
 600 The return value is number of composed characters."
 601   (if (< (1+ from) to)
 602       (progn
 603         (if string
 604             (compose-string string from to)
 605           (compose-region from to))
 606         (- to from))))
 607
 608 ;;;###autoload
 609 (defun lao-compose-region (from to)
 610   (interactive "r")
 611   (save-restriction
 612     (narrow-to-region from to)
 613     (goto-char (point-min))
 614     (with-category-table lao-category-table
 615       (while (re-search-forward lao-composition-pattern nil t)
 616         (compose-region (match-beginning 0) (point))))))
 617
 618 ;;
 619 (provide 'lao-util)
 620
 621 ;;; arch-tag: 1f828781-3cb8-4695-88af-8f33222338ce
 622 ;;; lao-util.el ends here