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, 2009, 2010
   4 ;;   Free Software Foundation, Inc.
   5 ;; Copyright (C) 1997, 1998, 1999, 2000, 2001, 2002, 2003, 2004, 2005, 2006,
   6 ;;   2007, 2008, 2009, 2010
   7 ;;   National Institute of Advanced Industrial Science and Technology (AIST)
   8 ;;   Registration Number H14PRO021
   9 ;; Copyright (C) 2003
  10 ;;   National Institute of Advanced Industrial Science and Technology (AIST)
  11 ;;   Registration Number H13PRO009
  12
  13 ;; Keywords: multilingual, Lao, i18n
  14
  15 ;; This file is part of GNU Emacs.
  16
  17 ;; GNU Emacs is free software: you can redistribute it and/or modify
  18 ;; it under the terms of the GNU General Public License as published by
  19 ;; the Free Software Foundation, either version 3 of the License, or
  20 ;; (at your option) any later version.
  21
  22 ;; GNU Emacs is distributed in the hope that it will be useful,
  23 ;; but WITHOUT ANY WARRANTY; without even the implied warranty of
  24 ;; MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
  25 ;; GNU General Public License for more details.
  26
  27 ;; You should have received a copy of the GNU General Public License
  28 ;; along with GNU Emacs.  If not, see <http://www.gnu.org/licenses/>.
  29
  30 ;;; Commentary:
  31
  32 ;;; Code:
  33
  34 ;; Setting information of Thai characters.
  35
  36 (defconst lao-category-table (make-category-table))
  37 (define-category ?c "Lao consonant" lao-category-table)
  38 (define-category ?s "Lao semi-vowel" lao-category-table)
  39 (define-category ?v "Lao upper/lower vowel" lao-category-table)
  40 (define-category ?t "Lao tone" lao-category-table)
  41
  42 (let ((l '((?\e(1!\e(B consonant "LETTER KOR  KAI'" "CHICKEN")
  43            (?\e(1"\e(B consonant "LETTER KHOR KHAI'" "EGG")
  44            (?\e(1#\e(B invalid nil)
  45            (?\e(1$\e(B consonant "LETTER QHOR QHWARGN" "BUFFALO")
  46            (?\e(1%\e(B invalid nil)
  47            (?  invalid nil)
  48            (?\e(1'\e(B consonant "LETTER NGOR NGUU" "SNAKE")
  49            (?\e(1(\e(B consonant "LETTER JOR JUA" "BUDDHIST NOVICE")
  50            (?\e(1)\e(B invalid nil)
  51            (?\e(1*\e(B consonant "LETTER XOR X\"ARNG" "ELEPHANT")
  52            (?\e(1+\e(B invalid nil)
  53            (?\e(1,\e(B invalid nil)
  54            (?\e(1-\e(B consonant "LETTER YOR YUNG" "MOSQUITO")
  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(1.\e(B invalid nil)
  61            (?\e(14\e(B consonant "LETTER DOR DANG" "NOSE")
  62            (?\e(15\e(B consonant "LETTER TOR TAR" "EYE")
  63            (?\e(16\e(B consonant "LETTER THOR THUNG" "TO ASK,QUESTION")
  64            (?\e(17\e(B consonant "LETTER DHOR DHARM" "FLAG")
  65            (?\e(18\e(B invalid nil)
  66            (?\e(19\e(B consonant "LETTER NOR NOK" "BIRD")
  67            (?\e(1:\e(B consonant "LETTER BOR BED" "FISHHOOK")
  68            (?\e(1;\e(B consonant "LETTER POR PAR" "FISH")
  69            (?\e(1<\e(B consonant "LETTER HPOR HPER\"" "BEE")
  70            (?\e(1=\e(B consonant "LETTER FHOR FHAR" "WALL")
  71            (?\e(1>\e(B consonant "LETTER PHOR PHUU" "MOUNTAIN")
  72            (?\e(1?\e(B consonant "LETTER FOR FAI" "FIRE")
  73            (?\e(1@\e(B invalid nil)
  74            (?\e(1A\e(B consonant "LETTER MOR MAR\"" "HORSE")
  75            (?\e(1B\e(B consonant "LETTER GNOR GNAR" "MEDICINE")
  76            (?\e(1C\e(B consonant "LETTER ROR ROD" "CAR")
  77            (?\e(1D\e(B invalid nil)
  78            (?\e(1E\e(B consonant "LETTER LOR LIING" "MONKEY")
  79            (?\e(1F\e(B invalid nil)
  80            (?\e(1G\e(B consonant "LETTER WOR WII" "HAND FAN")
  81            (?\e(1H\e(B invalid nil)
  82            (?\e(1I\e(B invalid nil)
  83            (?\e(1J\e(B consonant "LETTER SOR SEA" "TIGER")
  84            (?\e(1K\e(B consonant "LETTER HHOR HHAI" "JAR")
  85            (?\e(1L\e(B invalid nil)
  86            (?\e(1M\e(B consonant "LETTER OR OOW" "TAKE")
  87            (?\e(1N\e(B consonant "LETTER HOR HEA" "BOAT")
  88            (?\e(1O\e(B special "ELLIPSIS")
  89            (?\e(1P\e(B vowel-base "VOWEL SIGN SARA A")
  90            (?\e(1Q\e(B vowel-upper "VOWEL SIGN MAI KAN")
  91            (?\e(1R\e(B vowel-base "VOWEL SIGN SARA AR")
  92            (?\e(1S\e(B vowel-base "VOWEL SIGN SARA AM")
  93            (?\e(1T\e(B vowel-upper "VOWEL SIGN SARA I")
  94            (?\e(1U\e(B vowel-upper "VOWEL SIGN SARA II")
  95            (?\e(1V\e(B vowel-upper "VOWEL SIGN SARA EU")
  96            (?\e(1W\e(B vowel-upper "VOWEL SIGN SARA UR")
  97            (?\e(1X\e(B vowel-lower "VOWEL SIGN SARA U")
  98            (?\e(1Y\e(B vowel-lower "VOWEL SIGN SARA UU")
  99            (?\e(1Z\e(B invalid nil)
 100            (?\e(1[\e(B vowel-upper "VOWEL SIGN MAI KONG")
 101            (?\e(1\\e(B semivowel-lower "SEMIVOWEL SIGN LO")
 102            (?\e(1]\e(B vowel-base "SEMIVOWEL SIGN SARA IA")
 103            (?\e(1^\e(B invalid nil)
 104            (?\e(1_\e(B invalid nil)
 105            (?\e(1`\e(B vowel-base "VOWEL SIGN SARA EE")
 106            (?\e(1a\e(B vowel-base "VOWEL SIGN SARA AA")
 107            (?\e(1b\e(B vowel-base "VOWEL SIGN SARA OO")
 108            (?\e(1c\e(B vowel-base "VOWEL SIGN SARA EI MAI MUAN\"")
 109            (?\e(1d\e(B vowel-base "VOWEL SIGN SARA AI MAI MAY")
 110            (?\e(1e\e(B invalid nil)
 111            (?\e(1f\e(B special "KO LA (REPETITION)")
 112            (?\e(1g\e(B invalid nil)
 113            (?\e(1h\e(B tone "TONE MAI EK")
 114            (?\e(1i\e(B tone "TONE MAI THO")
 115            (?\e(1j\e(B tone "TONE MAI TI")
 116            (?\e(1k\e(B tone "TONE MAI JADTAWAR")
 117            (?\e(1l\e(B tone "CANCELLATION MARK")
 118            (?\e(1m\e(B vowel-upper "VOWEL SIGN SARA OR")
 119            (?\e(1n\e(B invalid nil)
 120            (?\e(1o\e(B invalid nil)
 121            (?\e(1p\e(B special "DIGIT ZERO")
 122            (?\e(1q\e(B special "DIGIT ONE")
 123            (?\e(1r\e(B special "DIGIT TWO")
 124            (?\e(1s\e(B special "DIGIT THREE")
 125            (?\e(1t\e(B special "DIGIT FOUR")
 126            (?\e(1u\e(B special "DIGIT FIVE")
 127            (?\e(1v\e(B special "DIGIT SIX")
 128            (?\e(1w\e(B special "DIGIT SEVEN")
 129            (?\e(1x\e(B special "DIGIT EIGHT")
 130            (?\e(1y\e(B special "DIGIT NINE")
 131            (?\e(1z\e(B invalid nil)
 132            (?\e(1{\e(B invalid nil)
 133            (?\e(1|\e(B consonant "LETTER NHOR NHUU" "MOUSE")
 134            (?\e(1}\e(B consonant "LETTER MHOR MHAR" "DOG")
 135            (?\e(1~\e(B invalid nil)))
 136       elm)
 137   (while l
 138     (setq elm (car l) l (cdr l))
 139     (let ((char (car elm))
 140           (ptype (nth 1 elm)))
 141       (cond ((eq ptype 'consonant)
 142              (modify-category-entry char ?c lao-category-table))
 143             ((memq ptype '(vowel-upper vowel-lower))
 144              (modify-category-entry char ?v lao-category-table))
 145             ((eq ptype 'semivowel-lower)
 146              (modify-category-entry char ?s lao-category-table))
 147             ((eq ptype 'tone)
 148              (modify-category-entry char ?t lao-category-table)))
 149       (put-char-code-property char 'phonetic-type ptype)
 150       (put-char-code-property char 'name (nth 2 elm))
 151       (put-char-code-property char 'meaning (nth 3 elm)))))
 152
 153 ;; The general composing rules are as follows:
 154 ;;
 155 ;;                          T
 156 ;;       V        T         V                  T
 157 ;; CV -> C, CT -> C, CVT -> C, Cv -> C, CvT -> C
 158 ;;                                   v         v
 159 ;;                             T
 160 ;;        V         T          V                    T
 161 ;; CsV -> C, CsT -> C, CsVT -> C, Csv -> C, CsvT -> C
 162 ;;        s         s          s         s          s
 163 ;;                                       v          v
 164
 165
 166 ;; where C: consonant, V: vowel upper, v: vowel lower,
 167 ;;       T: tone mark, s: semivowel lower
 168
 169 (defvar lao-composition-pattern
 170   "\\cc\\(\\ct\\|\\cv\\ct?\\|\\cs\\(\\ct\\|\\cv\\ct?\\)?\\)"
 171   "Regular expression matching a Lao composite sequence.")
 172
 173 ;;;###autoload
 174 (defun lao-compose-string (str)
 175   (with-category-table lao-category-table
 176    (let ((idx 0))
 177      (while (setq idx (string-match lao-composition-pattern str idx))
 178        (compose-string str idx (match-end 0))
 179        (setq idx (match-end 0))))
 180    str))
 181
 182 ;;; LRT: Lao <-> Roman Transcription
 183
 184 ;; Upper vowels and tone-marks are put on the letter.
 185 ;; Semi-vowel-sign-lo and lower vowels are put under the letter.
 186
 187 (defconst lao-transcription-consonant-alist
 188   (sort '(;; single consonants
 189           ("k" . "\e(1!\e(B")
 190           ("kh" . "\e(1"\e(B")
 191           ("qh" . "\e(1$\e(B")
 192           ("ng" . "\e(1'\e(B")
 193           ("j" . "\e(1(\e(B")
 194           ("s" . "\e(1J\e(B")
 195           ("x" . "\e(1*\e(B")
 196           ("y" . "\e(1-\e(B")
 197           ("d" . "\e(14\e(B")
 198           ("t" . "\e(15\e(B")
 199           ("th" . "\e(16\e(B")
 200           ("dh" . "\e(17\e(B")
 201           ("n" . "\e(19\e(B")
 202           ("b" . "\e(1:\e(B")
 203           ("p" . "\e(1;\e(B")
 204           ("hp" . "\e(1<\e(B")
 205           ("fh" . "\e(1=\e(B")
 206           ("ph" . "\e(1>\e(B")
 207           ("f" . "\e(1?\e(B")
 208           ("m" . "\e(1A\e(B")
 209           ("gn" . "\e(1B\e(B")
 210           ("l" . "\e(1E\e(B")
 211           ("r" . "\e(1C\e(B")
 212           ("v" . "\e(1G\e(B")
 213           ("w" . "\e(1G\e(B")
 214           ("hh" . "\e(1K\e(B")
 215           ("O" . "\e(1M\e(B")
 216           ("h" . "\e(1N\e(B")
 217           ("nh" . "\e(1|\e(B")
 218           ("mh" . "\e(1}\e(B")
 219           ("lh" . ["\e(1K\\e(B"])
 220           ;; double consonants
 221           ("ngh" . ["\e(1K'\e(B"])
 222           ("yh" . ["\e(1K]\e(B"])
 223           ("wh" . ["\e(1KG\e(B"])
 224           ("hl" . ["\e(1KE\e(B"])
 225           ("hy" . ["\e(1K-\e(B"])
 226           ("hn" . ["\e(1K9\e(B"])
 227           ("hm" . ["\e(1KA\e(B"])
 228           )
 229         (function (lambda (x y) (> (length (car x)) (length (car y)))))))
 230
 231 (defconst lao-transcription-semi-vowel-alist
 232   '(("r" . "\e(1\\e(B")))
 233
 234 (defconst lao-transcription-vowel-alist
 235   (sort '(("a" . "\e(1P\e(B")
 236           ("ar" . "\e(1R\e(B")
 237           ("i" . "\e(1T\e(B")
 238           ("ii" . "\e(1U\e(B")
 239           ("eu" . "\e(1V\e(B")
 240           ("ur" . "\e(1W\e(B")
 241           ("u" . "\e(1X\e(B")
 242           ("uu" . "\e(1Y\e(B")
 243           ("e" . ["\e(1`P\e(B"])
 244           ("ee" . "\e(1`\e(B")
 245           ("ae" . ["\e(1aP\e(B"])
 246           ("aa" . "\e(1a\e(B")
 247           ("o" . ["\e(1bP\e(B"])
 248           ("oo" . "\e(1b\e(B")
 249           ("oe" . ["\e(1`RP\e(B"])
 250           ("or" . "\e(1m\e(B")
 251           ("er" . ["\e(1`T\e(B"])
 252           ("ir" . ["\e(1`U\e(B"])
 253           ("ua" . ["\e(1[GP\e(B"])
 254           ("uaa" . ["\e(1[G\e(B"])
 255           ("ie" . ["\e(1`Q]P\e(B"])
 256           ("ia" . ["\e(1`Q]\e(B"])
 257           ("ea" . ["\e(1`VM\e(B"])
 258           ("eaa" . ["\e(1`WM\e(B"])
 259           ("ai" . "\e(1d\e(B")
 260           ("ei" . "\e(1c\e(B")
 261           ("ao" . ["\e(1`[R\e(B"])
 262           ("aM" . "\e(1S\e(B"))
 263         (function (lambda (x y) (> (length (car x)) (length (car y)))))))
 264
 265 ;; Maa-sakod is put at the tail.
 266 (defconst lao-transcription-maa-sakod-alist
 267   '(("k" . "\e(1!\e(B")
 268     ("g" . "\e(1'\e(B")
 269     ("y" . "\e(1-\e(B")
 270     ("d" . "\e(14\e(B")
 271     ("n" . "\e(19\e(B")
 272     ("b" . "\e(1:\e(B")
 273     ("m" . "\e(1A\e(B")
 274     ("v" . "\e(1G\e(B")
 275     ("w" . "\e(1G\e(B")
 276     ))
 277
 278 (defconst lao-transcription-tone-alist
 279   '(("'" . "\e(1h\e(B")
 280     ("\"" . "\e(1i\e(B")
 281     ("^" . "\e(1j\e(B")
 282     ("+" . "\e(1k\e(B")
 283     ("~" . "\e(1l\e(B")))
 284
 285 (defconst lao-transcription-punctuation-alist
 286   '(("\\0" . "\e(1p\e(B")
 287     ("\\1" . "\e(1q\e(B")
 288     ("\\2" . "\e(1r\e(B")
 289     ("\\3" . "\e(1s\e(B")
 290     ("\\4" . "\e(1t\e(B")
 291     ("\\5" . "\e(1u\e(B")
 292     ("\\6" . "\e(1v\e(B")
 293     ("\\7" . "\e(1w\e(B")
 294     ("\\8" . "\e(1x\e(B")
 295     ("\\9" . "\e(1y\e(B")
 296     ("\\\\" . "\e(1f\e(B")
 297     ("\\$" . "\e(1O\e(B")))
 298
 299 (defconst lao-transcription-pattern
 300   (concat
 301    "\\("
 302    (mapconcat 'car lao-transcription-consonant-alist "\\|")
 303    "\\)\\("
 304    (mapconcat 'car lao-transcription-semi-vowel-alist "\\|")
 305    "\\)?\\(\\("
 306    (mapconcat 'car lao-transcription-vowel-alist "\\|")
 307    "\\)\\("
 308    (mapconcat 'car lao-transcription-maa-sakod-alist "\\|")
 309    "\\)?\\("
 310    (mapconcat (lambda (x) (regexp-quote (car x)))
 311               lao-transcription-tone-alist "\\|")
 312    "\\)?\\)?\\|"
 313    (mapconcat (lambda (x) (regexp-quote (car x)))
 314               lao-transcription-punctuation-alist "\\|")
 315    )
 316   "Regexp of Roman transcription pattern for one Lao syllable.")
 317
 318 (defconst lao-transcription-pattern
 319   (concat
 320    "\\("
 321    (regexp-opt (mapcar 'car lao-transcription-consonant-alist))
 322    "\\)\\("
 323    (regexp-opt (mapcar 'car lao-transcription-semi-vowel-alist))
 324    "\\)?\\(\\("
 325    (regexp-opt (mapcar 'car lao-transcription-vowel-alist))
 326    "\\)\\("
 327    (regexp-opt (mapcar 'car lao-transcription-maa-sakod-alist))
 328    "\\)?\\("
 329    (regexp-opt (mapcar 'car lao-transcription-tone-alist))
 330    "\\)?\\)?\\|"
 331    (regexp-opt (mapcar 'car lao-transcription-punctuation-alist))
 332    )
 333   "Regexp of Roman transcription pattern for one Lao syllable.")
 334
 335 (defconst lao-vowel-reordering-rule
 336   '(("\e(1P\e(B" (0 ?\e(1P\e(B) (0 ?\e(1Q\e(B))
 337     ("\e(1R\e(B" (0 ?\e(1R\e(B))
 338     ("\e(1T\e(B" (0 ?\e(1U\e(B))
 339     ("\e(1U\e(B" (0 ?\e(1U\e(B))
 340     ("\e(1V\e(B" (0 ?\e(1V\e(B))
 341     ("\e(1W\e(B" (0 ?\e(1W\e(B))
 342     ("\e(1X\e(B" (0 ?\e(1X\e(B))
 343     ("\e(1Y\e(B" (0 ?\e(1Y\e(B))
 344     ("\e(1`P\e(B" (?\e(1`\e(B 0 ?\e(1P\e(B) (?\e(1`\e(B 0 ?\e(1Q\e(B))
 345     ("\e(1`\e(B" (?\e(1`\e(B 0))
 346     ("\e(1aP\e(B" (?\e(1a\e(B 0 ?\e(1P\e(B) (?\e(1a\e(B 0 ?\e(1Q\e(B))
 347     ("\e(1a\e(B" (?\e(1a\e(B 0))
 348     ("\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))
 349     ("\e(1b\e(B" (?\e(1b\e(B 0))
 350     ("\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))
 351     ("\e(1m\e(B" (0 ?\e(1m\e(B) (0 ?\e(1M\e(B))
 352     ("\e(1`T\e(B" (?\e(1`\e(B 0 ?\e(1T\e(B))
 353     ("\e(1`U\e(B" (?\e(1`\e(B 0 ?\e(1U\e(B))
 354     ("\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))
 355     ("\e(1[G\e(B" (0 ?\e(1[\e(B ?\e(1G\e(B) (0 ?\e(1G\e(B))
 356     ("\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))
 357     ("\e(1`Q]\e(B" (?\e(1`\e(B 0 ?\e(1Q\e(B ?\e(1]\e(B) (0 ?\e(1]\e(B))
 358     ("\e(1`VM\e(B" (?\e(1`\e(B 0 ?\e(1V\e(B ?\e(1M\e(B))
 359     ("\e(1`WM\e(B" (?\e(1`\e(B 0 ?\e(1W\e(B ?\e(1M\e(B))
 360     ("\e(1d\e(B" (?\e(1d\e(B 0))
 361     ("\e(1c\e(B" (?\e(1c\e(B 0))
 362     ("\e(1`[R\e(B" (?\e(1`\e(B 0 ?\e(1[\e(B ?\e(1R\e(B))
 363     ("\e(1S\e(B" (0 ?\e(1S\e(B)))
 364   "Alist of Lao vowel string vs the corresponding re-ordering rule.
 365 Each element has this form:
 366         (VOWEL NO-MAA-SAKOD-RULE WITH-MAA-SAKOD-RULE (MAA-SAKOD-0 RULE-0) ...)
 367
 368 VOWEL is a vowel string (e.g. \"\e(1`Q]P\e(B\").
 369
 370 NO-MAA-SAKOD-RULE is a rule to re-order and modify VOWEL following a
 371 consonant.  It is a list vowel characters or 0.  The element 0
 372 indicate the place to embed a consonant.
 373
 374 Optional WITH-MAA-SAKOD-RULE is a rule to re-order and modify VOWEL
 375 follwoing a consonant and preceding a maa-sakod character.  If it is
 376 nil, NO-MAA-SAKOD-RULE is used.  The maa-sakod character is alwasy
 377 appended at the tail.
 378
 379 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
 380 string following a consonant `\e(1!\e(B' should be re-ordered as \"\e(1`!WM\e(B\".
 381
 382 Optional (MAA-SAKOD-n RULE-n) are rules specially applied to maa-sakod
 383 character MAA-SAKOD-n.")
 384
 385 ;;;###autoload
 386 (defun lao-transcribe-single-roman-syllable-to-lao (from to &optional str)
 387   "Transcribe a Romanized Lao syllable in the region FROM and TO to Lao string.
 388 Only the first syllable is transcribed.
 389 The value has the form: (START END LAO-STRING), where
 390 START and END are the beggining and end positions of the Roman Lao syllable,
 391 LAO-STRING is the Lao character transcription of it.
 392
 393 Optional 3rd arg STR, if non-nil, is a string to search for Roman Lao
 394 syllable.  In that case, FROM and TO are indexes to STR."
 395   (if str
 396       (if (setq from (string-match lao-transcription-pattern str from))
 397           (progn
 398             (if (>= from to)
 399                 (setq from nil)
 400               (setq to (match-end 0)))))
 401     (save-excursion
 402       (goto-char from)
 403       (if (setq to (re-search-forward lao-transcription-pattern to t))
 404           (setq from (match-beginning 0))
 405         (setq from nil))))
 406   (if from
 407       (let* ((consonant (match-string 1 str))
 408              (semivowel (match-string 3 str))
 409              (vowel (match-string 5 str))
 410              (maa-sakod (match-string 8 str))
 411              (tone (match-string 9 str))
 412              lao-consonant lao-semivowel lao-vowel lao-maa-sakod lao-tone
 413              clen cidx)
 414         (setq to (match-end 0))
 415         (if (not consonant)
 416             (setq str (cdr (assoc (match-string 0 str)
 417                                   lao-transcription-punctuation-alist)))
 418           (setq lao-consonant
 419                 (cdr (assoc consonant lao-transcription-consonant-alist)))
 420           (if (vectorp lao-consonant)
 421               (setq lao-consonant (aref lao-consonant 0)))
 422           (setq clen (length lao-consonant))
 423           (if semivowel
 424               ;; Include semivowel in STR.
 425               (setq lao-semivowel
 426                     (cdr (assoc semivowel lao-transcription-semi-vowel-alist))
 427                     str (if (= clen 1)
 428                             (concat lao-consonant lao-semivowel)
 429                           (concat (substring lao-consonant 0 1) lao-semivowel
 430                                   (substring lao-consonant 1))))
 431             (setq str lao-consonant))
 432           (if vowel
 433               (let (rule)
 434                 (setq lao-vowel
 435                       (cdr (assoc vowel lao-transcription-vowel-alist)))
 436                 (if (vectorp lao-vowel)
 437                     (setq lao-vowel (aref lao-vowel 0)))
 438                 (setq rule (assoc lao-vowel lao-vowel-reordering-rule))
 439                 (if (null maa-sakod)
 440                     (setq rule (nth 1 rule))
 441                   (setq lao-maa-sakod
 442                         (cdr (assoc maa-sakod lao-transcription-maa-sakod-alist))
 443                         rule
 444                         (or (cdr (assq (aref lao-maa-sakod 0) (nthcdr 2 rule)))
 445                             (nth 2 rule)
 446                             (nth 1 rule))))
 447                 (or rule
 448                     (error "Lao vowel %S has no re-ordering rule" lao-vowel))
 449                 (setq lao-consonant str str "")
 450                 (while rule
 451                   (if (= (car rule) 0)
 452                       (setq str (concat str lao-consonant)
 453                             cidx (length str))
 454                     (setq str (concat str (list (car rule)))))
 455                   (setq rule (cdr rule)))
 456                 (or cidx
 457                     (error "Lao vowel %S has malformed re-ordering rule" vowel))
 458                 ;; Set CIDX to after upper or lower vowel if any.
 459                 (let ((len (length str)))
 460                   (while (and (< cidx len)
 461                               (memq (get-char-code-property (aref str cidx)
 462                                                             'phonetic-type)
 463                                     '(vowel-lower vowel-upper)))
 464                     (setq cidx (1+ cidx))))
 465                 (if lao-maa-sakod
 466                     (setq str (concat str lao-maa-sakod)))
 467                 (if tone
 468                     (setq lao-tone
 469                           (cdr (assoc tone lao-transcription-tone-alist))
 470                           str (concat (substring str 0 cidx) lao-tone
 471                                       (substring str cidx)))))))
 472         (list from to (lao-compose-string str)))))
 473
 474 ;;;###autoload
 475 (defun lao-transcribe-roman-to-lao-string (str)
 476   "Transcribe Romanized Lao string STR to Lao character string."
 477   (let ((from 0)
 478         (to (length str))
 479         (lao-str "")
 480         val)
 481     (while (setq val (lao-transcribe-single-roman-syllable-to-lao from to str))
 482       (let ((start (car val))
 483             (end (nth 1 val))
 484             (lao (nth 2 val)))
 485         (if (> start from)
 486             (setq lao-str (concat lao-str (substring str from start) lao))
 487           (setq lao-str (concat lao-str lao)))
 488         (setq from end)))
 489     (if (< from to)
 490         (concat lao-str (substring str from to))
 491       lao-str)))
 492
 493 ;;;###autoload
 494 (defun lao-composition-function (gstring)
 495   (if (= (lgstring-char-len gstring) 1)
 496       (compose-gstring-for-graphic gstring)
 497     (or (font-shape-gstring gstring)
 498         (let ((glyph-len (lgstring-glyph-len gstring))
 499               (i 0)
 500               glyph)
 501           (while (and (< i glyph-len)
 502                       (setq glyph (lgstring-glyph gstring i)))
 503             (setq i (1+ i)))
 504           (compose-glyph-string-relative gstring 0 i 0.1)))))
 505
 506 ;;;###autoload
 507 (defun lao-compose-region (from to)
 508   (interactive "r")
 509   (save-restriction
 510     (narrow-to-region from to)
 511     (goto-char (point-min))
 512     (with-category-table lao-category-table
 513       (while (re-search-forward lao-composition-pattern nil t)
 514         (compose-region (match-beginning 0) (point))))))
 515
 516 ;;
 517 (provide 'lao-util)
 518
 519 ;; arch-tag: 1f828781-3cb8-4695-88af-8f33222338ce
 520 ;;; lao-util.el ends here