-;;; lao-util.el --- utilities for Lao
+;;; lao-util.el --- utilities for Lao -*- coding: iso-2022-7bit; -*-
;; Copyright (C) 1997 Electrotechnical Laboratory, JAPAN.
;; Licensed to the Free Software Foundation.
+;; Copyright (C) 2001 Free Software Foundation, Inc.
;; Keywords: multilingual, Lao
;; Free Software Foundation, Inc., 59 Temple Place - Suite 330,
;; Boston, MA 02111-1307, USA.
+;;; Commentary:
+
;;; Code:
-;;;###autoload
-(defun setup-lao-environment ()
- "Setup multilingual environment (MULE) for Lao."
- (interactive)
- (set-language-environment "Lao"))
+;; Setting information of Thai characters.
+
+(defconst lao-category-table (make-category-table))
+(define-category ?c "Lao consonant" lao-category-table)
+(define-category ?s "Lao semi-vowel" lao-category-table)
+(define-category ?v "Lao upper/lower vowel" lao-category-table)
+(define-category ?t "Lao tone" lao-category-table)
(let ((l '((?\e(1!\e(B consonant "LETTER KOR KAI'" "CHICKEN")
(?\e(1"\e(B consonant "LETTER KHOR KHAI'" "EGG")
(?\e(1Q\e(B vowel-upper "VOWEL SIGN MAI KAN")
(?\e(1R\e(B vowel-base "VOWEL SIGN SARA AR")
(?\e(1S\e(B vowel-base "VOWEL SIGN SARA AM")
- (?\e(1T\e(B vowel-upper "VOWEL SIGN SARA I")
+ (?\e(1T\e(B vowel-upper "VOWEL SIGN SARA I")
(?\e(1U\e(B vowel-upper "VOWEL SIGN SARA II")
(?\e(1V\e(B vowel-upper "VOWEL SIGN SARA EU")
(?\e(1W\e(B vowel-upper "VOWEL SIGN SARA UR")
(?\e(1c\e(B vowel-base "VOWEL SIGN SARA EI MAI MUAN\"")
(?\e(1d\e(B vowel-base "VOWEL SIGN SARA AI MAI MAY")
(?\e(1e\e(B invalid nil)
- (?\e(1f\e(B special "KO LA (REPETITION)")
+ (?\e(1f\e(B special "KO LA (REPETITION)")
(?\e(1g\e(B invalid nil)
(?\e(1h\e(B tone "TONE MAI EK")
(?\e(1i\e(B tone "TONE MAI THO")
(?\e(1|\e(B consonant "LETTER NHOR NHUU" "MOUSE")
(?\e(1}\e(B consonant "LETTER MHOR MHAR" "DOG")
(?\e(1~\e(B invalid nil)
- ))
+ ;; Unicode equivalents
+ (?\e$,1D!\e(B consonant "LETTER KOR KAI'" "CHICKEN")
+ (?\e$,1D"\e(B consonant "LETTER KHOR KHAI'" "EGG")
+ (?\e$,1D$\e(B consonant "LETTER QHOR QHWARGN" "BUFFALO")
+ (?\e$,1D'\e(B consonant "LETTER NGOR NGUU" "SNAKE")
+ (?\e$,1D(\e(B consonant "LETTER JOR JUA" "BUDDHIST NOVICE")
+ (?\e$,1D*\e(B consonant "LETTER XOR X\"ARNG" "ELEPHANT")
+ (?\e$,1D-\e(B consonant "LETTER YOR YUNG" "MOSQUITO")
+ (?\e$,1D4\e(B consonant "LETTER DOR DANG" "NOSE")
+ (?\e$,1D5\e(B consonant "LETTER TOR TAR" "EYE")
+ (?\e$,1D6\e(B consonant "LETTER THOR THUNG" "TO ASK,QUESTION")
+ (?\e$,1D7\e(B consonant "LETTER DHOR DHARM" "FLAG")
+ (?\e$,1D9\e(B consonant "LETTER NOR NOK" "BIRD")
+ (?\e$,1D:\e(B consonant "LETTER BOR BED" "FISHHOOK")
+ (?\e$,1D;\e(B consonant "LETTER POR PAR" "FISH")
+ (?\e$,1D<\e(B consonant "LETTER HPOR HPER\"" "BEE")
+ (?\e$,1D=\e(B consonant "LETTER FHOR FHAR" "WALL")
+ (?\e$,1D>\e(B consonant "LETTER PHOR PHUU" "MOUNTAIN")
+ (?\e$,1D?\e(B consonant "LETTER FOR FAI" "FIRE")
+ (?\e$,1DA\e(B consonant "LETTER MOR MAR\"" "HORSE")
+ (?\e$,1DB\e(B consonant "LETTER GNOR GNAR" "MEDICINE")
+ (?\e$,1DC\e(B consonant "LETTER ROR ROD" "CAR")
+ (?\e$,1DE\e(B consonant "LETTER LOR LIING" "MONKEY")
+ (?\e$,1DG\e(B consonant "LETTER WOR WII" "HAND FAN")
+ (?\e$,1DJ\e(B consonant "LETTER SOR SEA" "TIGER")
+ (?\e$,1DK\e(B consonant "LETTER HHOR HHAI" "JAR")
+ (?\e$,1DM\e(B consonant "LETTER OR OOW" "TAKE")
+ (?\e$,1DN\e(B consonant "LETTER HOR HEA" "BOAT")
+ (?\e$,1DO\e(B special "ELLIPSIS")
+ (?\e$,1DP\e(B vowel-base "VOWEL SIGN SARA A")
+ (?\e$,1DQ\e(B vowel-upper "VOWEL SIGN MAI KAN")
+ (?\e$,1DR\e(B vowel-base "VOWEL SIGN SARA AR")
+ (?\e$,1DS\e(B vowel-base "VOWEL SIGN SARA AM")
+ (?\e$,1DT\e(B vowel-upper "VOWEL SIGN SARA I")
+ (?\e$,1DU\e(B vowel-upper "VOWEL SIGN SARA II")
+ (?\e$,1DV\e(B vowel-upper "VOWEL SIGN SARA EU")
+ (?\e$,1DW\e(B vowel-upper "VOWEL SIGN SARA UR")
+ (?\e$,1DX\e(B vowel-lower "VOWEL SIGN SARA U")
+ (?\e$,1DY\e(B vowel-lower "VOWEL SIGN SARA UU")
+ (?\e$,1D[\e(B vowel-upper "VOWEL SIGN MAI KONG")
+ (?\e$,1D\\e(B semivowel-lower "SEMIVOWEL SIGN LO")
+ (?\e$,1D]\e(B vowel-base "SEMIVOWEL SIGN SARA IA")
+ (?\e$,1D`\e(B vowel-base "VOWEL SIGN SARA EE")
+ (?\e$,1Da\e(B vowel-base "VOWEL SIGN SARA AA")
+ (?\e$,1Db\e(B vowel-base "VOWEL SIGN SARA OO")
+ (?\e$,1Dc\e(B vowel-base "VOWEL SIGN SARA EI MAI MUAN\"")
+ (?\e$,1Dd\e(B vowel-base "VOWEL SIGN SARA AI MAI MAY")
+ (?\e$,1Df\e(B special "KO LA (REPETITION)")
+ (?\e$,1Dh\e(B tone "TONE MAI EK")
+ (?\e$,1Di\e(B tone "TONE MAI THO")
+ (?\e$,1Dj\e(B tone "TONE MAI TI")
+ (?\e$,1Dk\e(B tone "TONE MAI JADTAWAR")
+ (?\e$,1Dl\e(B tone "CANCELLATION MARK")
+ (?\e$,1Dm\e(B vowel-upper "VOWEL SIGN SARA OR")
+ (?\e$,1Dp\e(B special "DIGIT ZERO")
+ (?\e$,1Dq\e(B special "DIGIT ONE")
+ (?\e$,1Dr\e(B special "DIGIT TWO")
+ (?\e$,1Ds\e(B special "DIGIT THREE")
+ (?\e$,1Dt\e(B special "DIGIT FOUR")
+ (?\e$,1Du\e(B special "DIGIT FIVE")
+ (?\e$,1Dv\e(B special "DIGIT SIX")
+ (?\e$,1Dw\e(B special "DIGIT SEVEN")
+ (?\e$,1Dx\e(B special "DIGIT EIGHT")
+ (?\e$,1Dy\e(B special "DIGIT NINE")
+ (?\e$,1D|\e(B consonant "LETTER NHOR NHUU" "MOUSE")
+ (?\e$,1D}\e(B consonant "LETTER MHOR MHAR" "DOG")))
elm)
(while l
- (setq elm (car l))
- (put-char-code-property (car elm) 'phonetic-type (car (cdr elm)))
- (put-char-code-property (car elm) 'name (nth 2 elm))
- (put-char-code-property (car elm) 'meaning (nth 3 elm))
- (setq l (cdr l))))
+ (setq elm (car l) l (cdr l))
+ (let ((char (car elm))
+ (ptype (nth 1 elm)))
+ (cond ((eq ptype 'consonant)
+ (modify-category-entry char ?c lao-category-table))
+ ((memq ptype '(vowel-upper vowel-lower))
+ (modify-category-entry char ?v lao-category-table))
+ ((eq ptype 'semivowel-lower)
+ (modify-category-entry char ?s lao-category-table))
+ ((eq ptype 'tone)
+ (modify-category-entry char ?t lao-category-table)))
+ (put-char-code-property char 'phonetic-type ptype)
+ (put-char-code-property char 'name (nth 2 elm))
+ (put-char-code-property char 'meaning (nth 3 elm)))))
+
+;; The general composing rules are as follows:
+;;
+;; T
+;; V T V T
+;; CV -> C, CT -> C, CVT -> C, Cv -> C, CvT -> C
+;; v v
+;; T
+;; V T V T
+;; CsV -> C, CsT -> C, CsVT -> C, Csv -> C, CvT -> C
+;; s s s s s
+;; v v
+
+
+;; where C: consonant, V: vowel upper, v: vowel lower,
+;; T: tone mark, s: semivowel lower
+
+(defvar lao-composition-pattern
+ "\\cc\\(\\ct\\|\\cv\\ct?\\|\\cs\\(\\ct\\|\\cv\\ct?\\)?\\)"
+ "Regular expression matching a Lao composite sequence.")
+
+;;;###autoload
+(defun lao-compose-string (str)
+ (with-category-table lao-category-table
+ (let ((idx 0))
+ (while (setq idx (string-match lao-composition-pattern str idx))
+ (compose-string str idx (match-end 0))
+ (setq idx (match-end 0))))
+ str))
+
+;;; LRT: Lao <-> Roman Transcription
+
+;; Upper vowels and tone-marks are put on the letter.
+;; Semi-vowel-sign-lo and lower vowels are put under the letter.
+
+(defconst lao-transcription-consonant-alist
+ (sort '(;; single consonants
+ ("k" . "\e(1!\e(B")
+ ("kh" . "\e(1"\e(B")
+ ("qh" . "\e(1$\e(B")
+ ("ng" . "\e(1'\e(B")
+ ("j" . "\e(1(\e(B")
+ ("s" . "\e(1J\e(B")
+ ("x" . "\e(1*\e(B")
+ ("y" . "\e(1-\e(B")
+ ("d" . "\e(14\e(B")
+ ("t" . "\e(15\e(B")
+ ("th" . "\e(16\e(B")
+ ("dh" . "\e(17\e(B")
+ ("n" . "\e(19\e(B")
+ ("b" . "\e(1:\e(B")
+ ("p" . "\e(1;\e(B")
+ ("hp" . "\e(1<\e(B")
+ ("fh" . "\e(1=\e(B")
+ ("ph" . "\e(1>\e(B")
+ ("f" . "\e(1?\e(B")
+ ("m" . "\e(1A\e(B")
+ ("gn" . "\e(1B\e(B")
+ ("l" . "\e(1E\e(B")
+ ("r" . "\e(1C\e(B")
+ ("v" . "\e(1G\e(B")
+ ("w" . "\e(1G\e(B")
+ ("hh" . "\e(1K\e(B")
+ ("O" . "\e(1M\e(B")
+ ("h" . "\e(1N\e(B")
+ ("nh" . "\e(1|\e(B")
+ ("mh" . "\e(1}\e(B")
+ ("lh" . ["\e(1K\\e(B"])
+ ;; double consonants
+ ("ngh" . ["\e(1K'\e(B"])
+ ("yh" . ["\e(1K]\e(B"])
+ ("wh" . ["\e(1KG\e(B"])
+ ("hl" . ["\e(1KE\e(B"])
+ ("hy" . ["\e(1K-\e(B"])
+ ("hn" . ["\e(1K9\e(B"])
+ ("hm" . ["\e(1KA\e(B"])
+ )
+ (function (lambda (x y) (> (length (car x)) (length (car y)))))))
+
+(defconst lao-transcription-semi-vowel-alist
+ '(("r" . "\e(1\\e(B")))
+
+(defconst lao-transcription-vowel-alist
+ (sort '(("a" . "\e(1P\e(B")
+ ("ar" . "\e(1R\e(B")
+ ("i" . "\e(1T\e(B")
+ ("ii" . "\e(1U\e(B")
+ ("eu" . "\e(1V\e(B")
+ ("ur" . "\e(1W\e(B")
+ ("u" . "\e(1X\e(B")
+ ("uu" . "\e(1Y\e(B")
+ ("e" . ["\e(1`P\e(B"])
+ ("ee" . "\e(1`\e(B")
+ ("ae" . ["\e(1aP\e(B"])
+ ("aa" . "\e(1a\e(B")
+ ("o" . ["\e(1bP\e(B"])
+ ("oo" . "\e(1b\e(B")
+ ("oe" . ["\e(1`RP\e(B"])
+ ("or" . "\e(1m\e(B")
+ ("er" . ["\e(1`T\e(B"])
+ ("ir" . ["\e(1`U\e(B"])
+ ("ua" . ["\e(1[GP\e(B"])
+ ("uaa" . ["\e(1[G\e(B"])
+ ("ie" . ["\e(1`Q]P\e(B"])
+ ("ia" . ["\e(1`Q]\e(B"])
+ ("ea" . ["\e(1`VM\e(B"])
+ ("eaa" . ["\e(1`WM\e(B"])
+ ("ai" . "\e(1d\e(B")
+ ("ei" . "\e(1c\e(B")
+ ("ao" . ["\e(1`[R\e(B"])
+ ("aM" . "\e(1S\e(B"))
+ (function (lambda (x y) (> (length (car x)) (length (car y)))))))
+
+;; Maa-sakod is put at the tail.
+(defconst lao-transcription-maa-sakod-alist
+ '(("k" . "\e(1!\e(B")
+ ("g" . "\e(1'\e(B")
+ ("y" . "\e(1-\e(B")
+ ("d" . "\e(14\e(B")
+ ("n" . "\e(19\e(B")
+ ("b" . "\e(1:\e(B")
+ ("m" . "\e(1A\e(B")
+ ("v" . "\e(1G\e(B")
+ ("w" . "\e(1G\e(B")
+ ))
+
+(defconst lao-transcription-tone-alist
+ '(("'" . "\e(1h\e(B")
+ ("\"" . "\e(1i\e(B")
+ ("^" . "\e(1j\e(B")
+ ("+" . "\e(1k\e(B")
+ ("~" . "\e(1l\e(B")))
+
+(defconst lao-transcription-punctuation-alist
+ '(("\\0" . "\e(1p\e(B")
+ ("\\1" . "\e(1q\e(B")
+ ("\\2" . "\e(1r\e(B")
+ ("\\3" . "\e(1s\e(B")
+ ("\\4" . "\e(1t\e(B")
+ ("\\5" . "\e(1u\e(B")
+ ("\\6" . "\e(1v\e(B")
+ ("\\7" . "\e(1w\e(B")
+ ("\\8" . "\e(1x\e(B")
+ ("\\9" . "\e(1y\e(B")
+ ("\\\\" . "\e(1f\e(B")
+ ("\\$" . "\e(1O\e(B")))
+
+(defconst lao-transcription-pattern
+ (concat
+ "\\("
+ (mapconcat 'car lao-transcription-consonant-alist "\\|")
+ "\\)\\("
+ (mapconcat 'car lao-transcription-semi-vowel-alist "\\|")
+ "\\)?\\(\\("
+ (mapconcat 'car lao-transcription-vowel-alist "\\|")
+ "\\)\\("
+ (mapconcat 'car lao-transcription-maa-sakod-alist "\\|")
+ "\\)?\\("
+ (mapconcat (lambda (x) (regexp-quote (car x)))
+ lao-transcription-tone-alist "\\|")
+ "\\)?\\)?\\|"
+ (mapconcat (lambda (x) (regexp-quote (car x)))
+ lao-transcription-punctuation-alist "\\|")
+ )
+ "Regexp of Roman transcription pattern for one Lao syllable.")
+
+(defconst lao-transcription-pattern
+ (concat
+ "\\("
+ (regexp-opt (mapcar 'car lao-transcription-consonant-alist))
+ "\\)\\("
+ (regexp-opt (mapcar 'car lao-transcription-semi-vowel-alist))
+ "\\)?\\(\\("
+ (regexp-opt (mapcar 'car lao-transcription-vowel-alist))
+ "\\)\\("
+ (regexp-opt (mapcar 'car lao-transcription-maa-sakod-alist))
+ "\\)?\\("
+ (regexp-opt (mapcar 'car lao-transcription-tone-alist))
+ "\\)?\\)?\\|"
+ (regexp-opt (mapcar 'car lao-transcription-punctuation-alist))
+ )
+ "Regexp of Roman transcription pattern for one Lao syllable.")
+
+(defconst lao-vowel-reordering-rule
+ '(("\e(1P\e(B" (0 ?\e(1P\e(B) (0 ?\e(1Q\e(B))
+ ("\e(1R\e(B" (0 ?\e(1R\e(B))
+ ("\e(1T\e(B" (0 ?\e(1U\e(B))
+ ("\e(1U\e(B" (0 ?\e(1U\e(B))
+ ("\e(1V\e(B" (0 ?\e(1V\e(B))
+ ("\e(1W\e(B" (0 ?\e(1W\e(B))
+ ("\e(1X\e(B" (0 ?\e(1X\e(B))
+ ("\e(1Y\e(B" (0 ?\e(1Y\e(B))
+ ("\e(1`P\e(B" (?\e(1`\e(B 0 ?\e(1P\e(B) (?\e(1`\e(B 0 ?\e(1Q\e(B))
+ ("\e(1`\e(B" (?\e(1`\e(B 0))
+ ("\e(1aP\e(B" (?\e(1a\e(B 0 ?\e(1P\e(B) (?\e(1a\e(B 0 ?\e(1Q\e(B))
+ ("\e(1a\e(B" (?\e(1a\e(B 0))
+ ("\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))
+ ("\e(1b\e(B" (?\e(1b\e(B 0))
+ ("\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))
+ ("\e(1m\e(B" (0 ?\e(1m\e(B) (0 ?\e(1M\e(B))
+ ("\e(1`T\e(B" (?\e(1`\e(B 0 ?\e(1T\e(B))
+ ("\e(1`U\e(B" (?\e(1`\e(B 0 ?\e(1U\e(B))
+ ("\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))
+ ("\e(1[G\e(B" (0 ?\e(1[\e(B ?\e(1G\e(B) (0 ?\e(1G\e(B))
+ ("\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))
+ ("\e(1`Q]\e(B" (?\e(1`\e(B 0 ?\e(1Q\e(B ?\e(1]\e(B) (0 ?\e(1]\e(B))
+ ("\e(1`VM\e(B" (?\e(1`\e(B 0 ?\e(1V\e(B ?\e(1M\e(B))
+ ("\e(1`WM\e(B" (?\e(1`\e(B 0 ?\e(1W\e(B ?\e(1M\e(B))
+ ("\e(1d\e(B" (?\e(1d\e(B 0))
+ ("\e(1c\e(B" (?\e(1c\e(B 0))
+ ("\e(1`[R\e(B" (?\e(1`\e(B 0 ?\e(1[\e(B ?\e(1R\e(B))
+ ("\e(1S\e(B" (0 ?\e(1S\e(B))
+
+ ;; Unicode equivalents
+ ("\e$,1DP\e(B" (0 ?\e$,1DP\e(B) (0 ?\e$,1DQ\e(B))
+ ("\e$,1DR\e(B" (0 ?\e$,1DR\e(B))
+ ("\e$,1DT\e(B" (0 ?\e$,1DU\e(B))
+ ("\e$,1DU\e(B" (0 ?\e$,1DU\e(B))
+ ("\e$,1DV\e(B" (0 ?\e$,1DV\e(B))
+ ("\e$,1DW\e(B" (0 ?\e$,1DW\e(B))
+ ("\e$,1DX\e(B" (0 ?\e$,1DX\e(B))
+ ("\e$,1DY\e(B" (0 ?\e$,1DY\e(B))
+ ("\e$,1D`DP\e(B" (?\e$,1D`\e(B 0 ?\e$,1DP\e(B) (?\e$,1D`\e(B 0 ?\e$,1DQ\e(B))
+ ("\e$,1D`\e(B" (?\e$,1D`\e(B 0))
+ ("\e$,1DaDP\e(B" (?\e$,1Da\e(B 0 ?\e$,1DP\e(B) (?\e$,1Da\e(B 0 ?\e$,1DQ\e(B))
+ ("\e$,1Da\e(B" (?\e$,1Da\e(B 0))
+ ("\e$,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$,1DQ\e(B))
+ ("\e$,1Db\e(B" (?\e$,1Db\e(B 0))
+ ("\e$,1D`DRDP\e(B" (?\e$,1D`\e(B 0 ?\e$,1DR\e(B ?\e$,1DP\e(B) (0 ?\e$,1DQ\e(B ?\e$,1DM\e(B))
+ ("\e$,1Dm\e(B" (0 ?\e$,1Dm\e(B) (0 ?\e$,1DM\e(B))
+ ("\e$,1D`DT\e(B" (?\e$,1D`\e(B 0 ?\e$,1DT\e(B))
+ ("\e$,1D`DU\e(B" (?\e$,1D`\e(B 0 ?\e$,1DU\e(B))
+ ("\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))
+ ("\e$,1D[DG\e(B" (0 ?\e$,1D[\e(B ?\e$,1DG\e(B) (0 ?\e$,1DG\e(B))
+ ("\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$,1D]\e(B))
+ ("\e$,1D`DQD]\e(B" (?\e$,1D`\e(B 0 ?\e$,1DQ\e(B ?\e$,1D]\e(B) (0 ?\e$,1D]\e(B))
+ ("\e$,1D`DVDM\e(B" (?\e$,1D`\e(B 0 ?\e$,1DV\e(B ?\e$,1DM\e(B))
+ ("\e$,1D`DWDM\e(B" (?\e$,1D`\e(B 0 ?\e$,1DW\e(B ?\e$,1DM\e(B))
+ ("\e$,1Dd\e(B" (?\e$,1Dd\e(B 0))
+ ("\e$,1Dc\e(B" (?\e$,1Dc\e(B 0))
+ ("\e$,1D`D[DR\e(B" (?\e$,1D`\e(B 0 ?\e$,1D[\e(B ?\e$,1DR\e(B))
+ ("\e$,1DS\e(B" (0 ?\e$,1DS\e(B)))
+ "Alist of Lao vowel string vs the corresponding re-ordering rule.
+Each element has this form:
+ (VOWEL NO-MAA-SAKOD-RULE WITH-MAA-SAKOD-RULE (MAA-SAKOD-0 RULE-0) ...)
+
+VOWEL is a vowel string (e.g. \"\e(1`Q]P\e(B\").
+
+NO-MAA-SAKOD-RULE is a rule to re-order and modify VOWEL following a
+consonant. It is a list vowel characters or 0. The element 0
+indicate the place to embed a consonant.
+
+Optional WITH-MAA-SAKOD-RULE is a rule to re-order and modify VOWEL
+follwoing a consonant and preceding a maa-sakod character. If it is
+nil, NO-MAA-SAKOD-RULE is used. The maa-sakod character is alwasy
+appended at the tail.
+
+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
+string following a consonant `\e(1!\e(B' should be re-ordered as \"\e(1`!WM\e(B\".
+
+Optional (MAA-SAKOD-n RULE-n) are rules specially applied to maa-sakod
+character MAA-SAKOD-n.")
+
+;;;###autoload
+(defun lao-transcribe-single-roman-syllable-to-lao (from to &optional str)
+ "Transcribe a Romanized Lao syllable in the region FROM and TO to Lao string.
+Only the first syllable is transcribed.
+The value has the form: (START END LAO-STRING), where
+START and END are the beggining and end positions of the Roman Lao syllable,
+LAO-STRING is the Lao character transcription of it.
+
+Optional 3rd arg STR, if non-nil, is a string to search for Roman Lao
+syllable. In that case, FROM and TO are indexes to STR."
+ (if str
+ (if (setq from (string-match lao-transcription-pattern str from))
+ (progn
+ (if (>= from to)
+ (setq from nil)
+ (setq to (match-end 0)))))
+ (save-excursion
+ (goto-char from)
+ (if (setq to (re-search-forward lao-transcription-pattern to t))
+ (setq from (match-beginning 0))
+ (setq from nil))))
+ (if from
+ (let* ((consonant (match-string 1 str))
+ (semivowel (match-string 3 str))
+ (vowel (match-string 5 str))
+ (maa-sakod (match-string 8 str))
+ (tone (match-string 9 str))
+ lao-consonant lao-semivowel lao-vowel lao-maa-sakod lao-tone
+ clen cidx)
+ (setq to (match-end 0))
+ (if (not consonant)
+ (setq str (cdr (assoc (match-string 0 str)
+ lao-transcription-punctuation-alist)))
+ (setq lao-consonant
+ (cdr (assoc consonant lao-transcription-consonant-alist)))
+ (if (vectorp lao-consonant)
+ (setq lao-consonant (aref lao-consonant 0)))
+ (setq clen (length lao-consonant))
+ (if semivowel
+ ;; Include semivowel in STR.
+ (setq lao-semivowel
+ (cdr (assoc semivowel lao-transcription-semi-vowel-alist))
+ str (if (= clen 1)
+ (concat lao-consonant lao-semivowel)
+ (concat (substring lao-consonant 0 1) lao-semivowel
+ (substring lao-consonant 1))))
+ (setq str lao-consonant))
+ (if vowel
+ (let (rule)
+ (setq lao-vowel
+ (cdr (assoc vowel lao-transcription-vowel-alist)))
+ (if (vectorp lao-vowel)
+ (setq lao-vowel (aref lao-vowel 0)))
+ (setq rule (assoc lao-vowel lao-vowel-reordering-rule))
+ (if (null maa-sakod)
+ (setq rule (nth 1 rule))
+ (setq lao-maa-sakod
+ (cdr (assoc maa-sakod lao-transcription-maa-sakod-alist))
+ rule
+ (or (cdr (assq (aref lao-maa-sakod 0) (nthcdr 2 rule)))
+ (nth 2 rule)
+ (nth 1 rule))))
+ (or rule
+ (error "Lao vowel %S has no re-ordering rule" lao-vowel))
+ (setq lao-consonant str str "")
+ (while rule
+ (if (= (car rule) 0)
+ (setq str (concat str lao-consonant)
+ cidx (length str))
+ (setq str (concat str (list (car rule)))))
+ (setq rule (cdr rule)))
+ (or cidx
+ (error "Lao vowel %S has malformed re-ordering rule" vowel))
+ ;; Set CIDX to after upper or lower vowel if any.
+ (let ((len (length str)))
+ (while (and (< cidx len)
+ (memq (get-char-code-property (aref str cidx)
+ 'phonetic-type)
+ '(vowel-lower vowel-upper)))
+ (setq cidx (1+ cidx))))
+ (if lao-maa-sakod
+ (setq str (concat str lao-maa-sakod)))
+ (if tone
+ (setq lao-tone
+ (cdr (assoc tone lao-transcription-tone-alist))
+ str (concat (substring str 0 cidx) lao-tone
+ (substring str cidx)))))))
+ (list from to (lao-compose-string str)))))
+
+;;;###autoload
+(defun lao-transcribe-roman-to-lao-string (str)
+ "Transcribe Romanized Lao string STR to Lao character string."
+ (let ((from 0)
+ (to (length str))
+ (lao-str "")
+ val)
+ (while (setq val (lao-transcribe-single-roman-syllable-to-lao from to str))
+ (let ((start (car val))
+ (end (nth 1 val))
+ (lao (nth 2 val)))
+ (if (> start from)
+ (setq lao-str (concat lao-str (substring str from start) lao))
+ (setq lao-str (concat lao-str lao)))
+ (setq from end)))
+ (if (< from to)
+ (concat lao-str (substring str from to))
+ lao-str)))
+
+;;;###autoload
+(defun lao-post-read-conversion (len)
+ (lao-compose-region (point) (+ (point) len))
+ len)
+
+;;;###autoload
+(defun lao-composition-function (from to pattern &optional string)
+ "Compose Lao text in the region FROM and TO.
+The text matches the regular expression PATTERN.
+Optional 4th argument STRING, if non-nil, is a string containing text
+to compose.
+
+The return value is number of composed characters."
+ (if (< (1+ from) to)
+ (progn
+ (if string
+ (compose-string string from to)
+ (compose-region from to))
+ (- to from))))
+
+;;;###autoload
+(defun lao-compose-region (from to)
+ (interactive "r")
+ (save-restriction
+ (narrow-to-region from to)
+ (goto-char (point-min))
+ (with-category-table lao-category-table
+ (while (re-search-forward lao-composition-pattern nil t)
+ (compose-region (match-beginning 0) (point))))))
;;
(provide 'lao-util)
HCoop / bpt / emacs.git / blobdiff