Mercurial > emacs
changeset 97840:8e246e0bdbba
Encoding changed to utf-8.
(thai-composition-function): Argument changed.
| author | Kenichi Handa <handa@m17n.org> |
|---|---|
| date | Fri, 29 Aug 2008 08:01:03 +0000 |
| parents | 6954172db216 |
| children | 10312e057a97 |
| files | lisp/language/thai-util.el |
| diffstat | 1 files changed, 113 insertions(+), 112 deletions(-) [+] |
line wrap: on
line diff
--- a/lisp/language/thai-util.el Fri Aug 29 07:59:57 2008 +0000 +++ b/lisp/language/thai-util.el Fri Aug 29 08:01:03 2008 +0000 @@ -1,4 +1,4 @@ -;;; thai-util.el --- utilities for Thai -*- coding: iso-2022-7bit; -*- +;;; thai-util.el --- utilities for Thai -*- coding: utf-8; -*- ;; Copyright (C) 1995, 1996, 1997, 1998, 1999, 2000, 2001, 2002, 2003, 2004, ;; 2005, 2006, 2007, 2008 @@ -49,107 +49,107 @@ ;; ;; where C: consonant, V: vowel upper, v: vowel lower, ;; T: tone mark, U: tone mark and upper sign. -;; Special rule: The sign `,Tl(B' can be put on the vowel `,TT(B'. +;; Special rule: The sign `์' can be put on the vowel `ิ'. (defvar thai-composition-pattern "\\cc\\(\\cu\\|\\cI\\cU\\|\\cv\\ct?\\)\\|\\cv\\ct\\|\\cI\\cU" "Regular expression matching a Thai composite sequence.") -(let ((l '((?,T!(B consonant "LETTER KO KAI") ; 0xA1 - (?,T"(B consonant "LETTER KHO KHAI") ; 0xA2 - (?,T#(B consonant "LETTER KHO KHUAT") ; 0xA3 - (?,T$(B consonant "LETTER KHO KHWAI") ; 0xA4 - (?,T%(B consonant "LETTER KHO KHON") ; 0xA5 - (?,T&(B consonant "LETTER KHO RAKHANG") ; 0xA6 - (?,T'(B consonant "LETTER NGO NGU") ; 0xA7 - (?,T((B consonant "LETTER CHO CHAN") ; 0xA8 - (?,T)(B consonant "LETTER CHO CHING") ; 0xA9 - (?,T*(B consonant "LETTER CHO CHANG") ; 0xAA - (?,T+(B consonant "LETTER SO SO") ; 0xAB - (?,T,(B consonant "LETTER CHO CHOE") ; 0xAC - (?,T-(B consonant "LETTER YO YING") ; 0xAD - (?,T.(B consonant "LETTER DO CHADA") ; 0xAE - (?,T/(B consonant "LETTER TO PATAK") ; 0xAF - (?,T0(B consonant "LETTER THO THAN") ; 0xB0 - (?,T1(B consonant "LETTER THO NANGMONTHO") ; 0xB1 - (?,T2(B consonant "LETTER THO PHUTHAO") ; 0xB2 - (?,T3(B consonant "LETTER NO NEN") ; 0xB3 - (?,T4(B consonant "LETTER DO DEK") ; 0xB4 - (?,T5(B consonant "LETTER TO TAO") ; 0xB5 - (?,T6(B consonant "LETTER THO THUNG") ; 0xB6 - (?,T7(B consonant "LETTER THO THAHAN") ; 0xB7 - (?,T8(B consonant "LETTER THO THONG") ; 0xB8 - (?,T9(B consonant "LETTER NO NU") ; 0xB9 - (?,T:(B consonant "LETTER BO BAIMAI") ; 0xBA - (?,T;(B consonant "LETTER PO PLA") ; 0xBB - (?,T<(B consonant "LETTER PHO PHUNG") ; 0xBC - (?,T=(B consonant "LETTER FO FA") ; 0xBD - (?,T>(B consonant "LETTER PHO PHAN") ; 0xBE - (?,T?(B consonant "LETTER FO FAN") ; 0xBF - (?,T@(B consonant "LETTER PHO SAMPHAO") ; 0xC0 - (?,TA(B consonant "LETTER MO MA") ; 0xC1 - (?,TB(B consonant "LETTER YO YAK") ; 0xC2 - (?,TC(B consonant "LETTER RO RUA") ; 0xC3 - (?,TD(B vowel-base "LETTER RU (Pali vowel letter)") ; 0xC4 - (?,TE(B consonant "LETTER LO LING") ; 0xC5 - (?,TF(B vowel-base "LETTER LU (Pali vowel letter)") ; 0xC6 - (?,TG(B consonant "LETTER WO WAEN") ; 0xC7 - (?,TH(B consonant "LETTER SO SALA") ; 0xC8 - (?,TI(B consonant "LETTER SO RUSI") ; 0xC9 - (?,TJ(B consonant "LETTER SO SUA") ; 0xCA - (?,TK(B consonant "LETTER HO HIP") ; 0xCB - (?,TL(B consonant "LETTER LO CHULA") ; 0xCC - (?,TM(B consonant "LETTER O ANG") ; 0xCD - (?,TN(B consonant "LETTER HO NOK HUK") ; 0xCE - (?,TO(B special "PAI YAN NOI (abbreviation)") ; 0xCF - (?,TP(B vowel-base "VOWEL SIGN SARA A") ; 0xD0 - (?,TQ(B vowel-upper "VOWEL SIGN MAI HAN-AKAT N/S-T") ; 0xD1 - (?,TR(B vowel-base "VOWEL SIGN SARA AA") ; 0xD2 - (?,TS(B vowel-base "VOWEL SIGN SARA AM") ; 0xD3 - (?,TT(B vowel-upper "VOWEL SIGN SARA I N/S-T") ; 0xD4 - (?,TU(B vowel-upper "VOWEL SIGN SARA II N/S-T") ; 0xD5 - (?,TV(B vowel-upper "VOWEL SIGN SARA UE N/S-T") ; 0xD6 - (?,TW(B vowel-upper "VOWEL SIGN SARA UEE N/S-T") ; 0xD7 - (?,TX(B vowel-lower "VOWEL SIGN SARA U N/S-B") ; 0xD8 - (?,TY(B vowel-lower "VOWEL SIGN SARA UU N/S-B") ; 0xD9 - (?,TZ(B vowel-lower "VOWEL SIGN PHINTHU N/S-B (Pali virama)") ; 0xDA - (?,T[(B invalid nil) ; 0xDA - (?,T\(B invalid nil) ; 0xDC - (?,T](B invalid nil) ; 0xDC - (?,T^(B invalid nil) ; 0xDC - (?,T_(B special "BAHT SIGN (currency symbol)") ; 0xDF - (?,T`(B vowel-base "VOWEL SIGN SARA E") ; 0xE0 - (?,Ta(B vowel-base "VOWEL SIGN SARA AE") ; 0xE1 - (?,Tb(B vowel-base "VOWEL SIGN SARA O") ; 0xE2 - (?,Tc(B vowel-base "VOWEL SIGN SARA MAI MUAN") ; 0xE3 - (?,Td(B vowel-base "VOWEL SIGN SARA MAI MALAI") ; 0xE4 - (?,Te(B vowel-base "LAK KHANG YAO") ; 0xE5 - (?,Tf(B special "MAI YAMOK (repetion)") ; 0xE6 - (?,Tg(B sign-upper "VOWEL SIGN MAI TAI KHU N/S-T") ; 0xE7 - (?,Th(B tone "TONE MAI EK N/S-T") ; 0xE8 - (?,Ti(B tone "TONE MAI THO N/S-T") ; 0xE9 - (?,Tj(B tone "TONE MAI TRI N/S-T") ; 0xEA - (?,Tk(B tone "TONE MAI CHATTAWA N/S-T") ; 0xEB - (?,Tl(B sign-upper "THANTHAKHAT N/S-T (cancellation mark)") ; 0xEC - (?,Tm(B sign-upper "NIKKHAHIT N/S-T (final nasal)") ; 0xED - (?,Tn(B sign-upper "YAMAKKAN N/S-T") ; 0xEE - (?,To(B special "FONRMAN") ; 0xEF - (?,Tp(B special "DIGIT ZERO") ; 0xF0 - (?,Tq(B special "DIGIT ONE") ; 0xF1 - (?,Tr(B special "DIGIT TWO") ; 0xF2 - (?,Ts(B special "DIGIT THREE") ; 0xF3 - (?,Tt(B special "DIGIT FOUR") ; 0xF4 - (?,Tu(B special "DIGIT FIVE") ; 0xF5 - (?,Tv(B special "DIGIT SIX") ; 0xF6 - (?,Tw(B special "DIGIT SEVEN") ; 0xF7 - (?,Tx(B special "DIGIT EIGHT") ; 0xF8 - (?,Ty(B special "DIGIT NINE") ; 0xF9 - (?,Tz(B special "ANGKHANKHU (ellipsis)") ; 0xFA - (?,T{(B special "KHOMUT (beginning of religious texts)") ; 0xFB - (?,T|(B invalid nil) ; 0xFC - (?,T}(B invalid nil) ; 0xFD - (?,T~(B invalid nil) ; 0xFE +(let ((l '((?ก consonant "LETTER KO KAI") ; 0xA1 + (?ข consonant "LETTER KHO KHAI") ; 0xA2 + (?ฃ consonant "LETTER KHO KHUAT") ; 0xA3 + (?ค consonant "LETTER KHO KHWAI") ; 0xA4 + (?ฅ consonant "LETTER KHO KHON") ; 0xA5 + (?ฆ consonant "LETTER KHO RAKHANG") ; 0xA6 + (?ง consonant "LETTER NGO NGU") ; 0xA7 + (?จ consonant "LETTER CHO CHAN") ; 0xA8 + (?ฉ consonant "LETTER CHO CHING") ; 0xA9 + (?ช consonant "LETTER CHO CHANG") ; 0xAA + (?ซ consonant "LETTER SO SO") ; 0xAB + (?ฌ consonant "LETTER CHO CHOE") ; 0xAC + (?ญ consonant "LETTER YO YING") ; 0xAD + (?ฎ consonant "LETTER DO CHADA") ; 0xAE + (?ฏ consonant "LETTER TO PATAK") ; 0xAF + (?ฐ consonant "LETTER THO THAN") ; 0xB0 + (?ฑ consonant "LETTER THO NANGMONTHO") ; 0xB1 + (?ฒ consonant "LETTER THO PHUTHAO") ; 0xB2 + (?ณ consonant "LETTER NO NEN") ; 0xB3 + (?ด consonant "LETTER DO DEK") ; 0xB4 + (?ต consonant "LETTER TO TAO") ; 0xB5 + (?ถ consonant "LETTER THO THUNG") ; 0xB6 + (?ท consonant "LETTER THO THAHAN") ; 0xB7 + (?ธ consonant "LETTER THO THONG") ; 0xB8 + (?น consonant "LETTER NO NU") ; 0xB9 + (?บ consonant "LETTER BO BAIMAI") ; 0xBA + (?ป consonant "LETTER PO PLA") ; 0xBB + (?ผ consonant "LETTER PHO PHUNG") ; 0xBC + (?ฝ consonant "LETTER FO FA") ; 0xBD + (?พ consonant "LETTER PHO PHAN") ; 0xBE + (?ฟ consonant "LETTER FO FAN") ; 0xBF + (?ภ consonant "LETTER PHO SAMPHAO") ; 0xC0 + (?ม consonant "LETTER MO MA") ; 0xC1 + (?ย consonant "LETTER YO YAK") ; 0xC2 + (?ร consonant "LETTER RO RUA") ; 0xC3 + (?ฤ vowel-base "LETTER RU (Pali vowel letter)") ; 0xC4 + (?ล consonant "LETTER LO LING") ; 0xC5 + (?ฦ vowel-base "LETTER LU (Pali vowel letter)") ; 0xC6 + (?ว consonant "LETTER WO WAEN") ; 0xC7 + (?ศ consonant "LETTER SO SALA") ; 0xC8 + (?ษ consonant "LETTER SO RUSI") ; 0xC9 + (?ส consonant "LETTER SO SUA") ; 0xCA + (?ห consonant "LETTER HO HIP") ; 0xCB + (?ฬ consonant "LETTER LO CHULA") ; 0xCC + (?อ consonant "LETTER O ANG") ; 0xCD + (?ฮ consonant "LETTER HO NOK HUK") ; 0xCE + (?ฯ special "PAI YAN NOI (abbreviation)") ; 0xCF + (?ะ vowel-base "VOWEL SIGN SARA A") ; 0xD0 + (?ั vowel-upper "VOWEL SIGN MAI HAN-AKAT N/S-T") ; 0xD1 + (?า vowel-base "VOWEL SIGN SARA AA") ; 0xD2 + (?ำ vowel-base "VOWEL SIGN SARA AM") ; 0xD3 + (?ิ vowel-upper "VOWEL SIGN SARA I N/S-T") ; 0xD4 + (?ี vowel-upper "VOWEL SIGN SARA II N/S-T") ; 0xD5 + (?ึ vowel-upper "VOWEL SIGN SARA UE N/S-T") ; 0xD6 + (?ื vowel-upper "VOWEL SIGN SARA UEE N/S-T") ; 0xD7 + (?ุ vowel-lower "VOWEL SIGN SARA U N/S-B") ; 0xD8 + (?ู vowel-lower "VOWEL SIGN SARA UU N/S-B") ; 0xD9 + (?ฺ vowel-lower "VOWEL SIGN PHINTHU N/S-B (Pali virama)") ; 0xDA + (? invalid nil) ; 0xDA + (? invalid nil) ; 0xDC + (? invalid nil) ; 0xDC + (? invalid nil) ; 0xDC + (?฿ special "BAHT SIGN (currency symbol)") ; 0xDF + (?เ vowel-base "VOWEL SIGN SARA E") ; 0xE0 + (?แ vowel-base "VOWEL SIGN SARA AE") ; 0xE1 + (?โ vowel-base "VOWEL SIGN SARA O") ; 0xE2 + (?ใ vowel-base "VOWEL SIGN SARA MAI MUAN") ; 0xE3 + (?ไ vowel-base "VOWEL SIGN SARA MAI MALAI") ; 0xE4 + (?ๅ vowel-base "LAK KHANG YAO") ; 0xE5 + (?ๆ special "MAI YAMOK (repetion)") ; 0xE6 + (?็ sign-upper "VOWEL SIGN MAI TAI KHU N/S-T") ; 0xE7 + (?่ tone "TONE MAI EK N/S-T") ; 0xE8 + (?้ tone "TONE MAI THO N/S-T") ; 0xE9 + (?๊ tone "TONE MAI TRI N/S-T") ; 0xEA + (?๋ tone "TONE MAI CHATTAWA N/S-T") ; 0xEB + (?์ sign-upper "THANTHAKHAT N/S-T (cancellation mark)") ; 0xEC + (?ํ sign-upper "NIKKHAHIT N/S-T (final nasal)") ; 0xED + (?๎ sign-upper "YAMAKKAN N/S-T") ; 0xEE + (?๏ special "FONRMAN") ; 0xEF + (?๐ special "DIGIT ZERO") ; 0xF0 + (?๑ special "DIGIT ONE") ; 0xF1 + (?๒ special "DIGIT TWO") ; 0xF2 + (?๓ special "DIGIT THREE") ; 0xF3 + (?๔ special "DIGIT FOUR") ; 0xF4 + (?๕ special "DIGIT FIVE") ; 0xF5 + (?๖ special "DIGIT SIX") ; 0xF6 + (?๗ special "DIGIT SEVEN") ; 0xF7 + (?๘ special "DIGIT EIGHT") ; 0xF8 + (?๙ special "DIGIT NINE") ; 0xF9 + (?๚ special "ANGKHANKHU (ellipsis)") ; 0xFA + (?๛ special "KHOMUT (beginning of religious texts)") ; 0xFB + (? invalid nil) ; 0xFC + (? invalid nil) ; 0xFD + (? invalid nil) ; 0xFE )) elm) (while l @@ -161,7 +161,7 @@ (modify-category-entry char ?c thai-category-table)) ((memq ptype '(vowel-upper vowel-lower)) (modify-category-entry char ?v thai-category-table) - (if (= char ?,TT(B) + (if (= char ?ิ) ;; Give category `I' to "SARA I". (modify-category-entry char ?I thai-category-table))) ((eq ptype 'tone) @@ -169,7 +169,7 @@ (modify-category-entry char ?u thai-category-table)) ((eq ptype 'sign-upper) (modify-category-entry char ?u thai-category-table) - (if (= char ?,Tl(B) + (if (= char ?์) ;; Give category `U' to "THANTHAKHAT". (modify-category-entry char ?U thai-category-table)))) (put-char-code-property char 'name (nth 2 elm))))) @@ -227,20 +227,21 @@ (thai-compose-region (point-min) (point-max))) ;;;###autoload -(defun thai-composition-function (pos to font-object string) - (setq pos (1- pos)) - (with-category-table thai-category-table - (if string - (if (and (>= pos 0) - (eq (string-match thai-composition-pattern string pos) pos)) - (prog1 (match-end 0) - (thai-compose-syllable pos (match-end 0) nil string))) - (if (>= pos (point-min)) - (progn - (goto-char pos) - (if (looking-at thai-composition-pattern) - (prog1 (match-end 0) - (thai-compose-syllable pos (match-end 0))))))))) +(defun thai-composition-function (gstring) + (if (= (lgstring-char-len gstring) 1) + (compose-gstring-for-graphic gstring) + (or (font-shape-gstring gstring) + (let ((glyph-len (lgstring-glyph-len gstring)) + (last-char (lgstring-char gstring + (1- (lgstring-char-len gstring)))) + (i 0) + glyph) + (while (and (< i glyph-len) + (setq glyph (lgstring-glyph gstring i))) + (setq i (1+ i))) + (if (= last-char ?ำ) + (setq i (1- i))) + (compose-glyph-string-relative gstring 0 i 0.1))))) ;; Thai-word-mode requires functions in the feature `thai-word'. (require 'thai-word)