1
0
mirror of https://git.savannah.gnu.org/git/emacs.git synced 2024-12-15 09:47:20 +00:00
emacs/lisp/language/tibet-util.el
2009-01-05 03:18:22 +00:00

426 lines
15 KiB
EmacsLisp
Raw Blame History

This file contains invisible Unicode characters

This file contains invisible Unicode characters that are indistinguishable to humans but may be processed differently by a computer. If you think that this is intentional, you can safely ignore this warning. Use the Escape button to reveal them.

;;; tibet-util.el --- utilities for Tibetan -*- coding: iso-2022-7bit; -*-
;; Copyright (C) 1997, 2001, 2002, 2003, 2004, 2005, 2006, 2007, 2008, 2009
;; Free Software Foundation, Inc.
;; Copyright (C) 1995, 1996, 1997, 1998, 1999, 2000, 2001, 2002, 2003, 2004,
;; 2005, 2006, 2007, 2008, 2009
;; National Institute of Advanced Industrial Science and Technology (AIST)
;; Registration Number H14PRO021
;; Author: Toru TOMABECHI <Toru.Tomabechi@orient.unil.ch>
;; Keywords: multilingual, Tibetan
;; Created: Feb. 17. 1997
;; This file is part of GNU Emacs.
;; GNU Emacs is free software: you can redistribute it and/or modify
;; it under the terms of the GNU General Public License as published by
;; the Free Software Foundation, either version 3 of the License, or
;; (at your option) any later version.
;; GNU Emacs is distributed in the hope that it will be useful,
;; but WITHOUT ANY WARRANTY; without even the implied warranty of
;; MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
;; GNU General Public License for more details.
;; You should have received a copy of the GNU General Public License
;; along with GNU Emacs. If not, see <http://www.gnu.org/licenses/>.
;;; History:
;; 1997.03.13 Modification in treatment of text properties;
;; Support for some special signs and punctuations.
;; 1999.10.25 Modification for a new composition way by K.Handa.
;;; Commentary:
;;; Code:
(defconst tibetan-obsolete-glyphs
`(("$(7!=(B" . "$(7!=(B") ; 2 col <-> 1 col
("$(7!?(B" . "$(7!?(B")
("$(7!@(B" . "$(7!@(B")
("$(7!A(B" . "$(7!A(B")
("$(7"`(B" . "$(7"`(B")
("$(7!;(B" . "$(7!;(B")
("$(7!D(B" . "$(7!D(B")
;; Yes these are dirty. But ...
("$(7!>(B $(7!>(B" . ,(compose-string "$(7!>(B $(7!>(B" 0 3 [?$(7!>(B (Br . Bl) ? (Br . Bl) ?$(7!>(B]))
("$(7!4!5!5(B" . ,(compose-string
"$(7#R#S#S#S(B" 0 4
[?$(7#R(B (Br . Bl) ?$(7#S(B (Br . Bl) ?$(7#S(B (Br . Bl) ?$(7#S(B]))
("$(7!4!5(B" . ,(compose-string "$(7#R#S#S(B" 0 3 [?$(7#R(B (Br . Bl) ?$(7#S(B (Br . Bl) ?$(7#S(B]))
("$(7!6(B" . ,(compose-string "$(7#R#S!I(B" 0 3 [?$(7#R(B (Br . Bl) ?$(7#S(B (br . tr) ?$(7!I(B]))
("$(7!4(B" . ,(compose-string "$(7#R#S(B" 0 2 [?$(7#R(B (Br . Bl) ?$(7#S(B]))))
;;;###autoload
(defun tibetan-char-p (ch)
"Check if char CH is Tibetan character.
Returns non-nil if CH is Tibetan. Otherwise, returns nil."
(memq (char-charset ch) '(tibetan tibetan-1-column)))
;;; Functions for Tibetan <-> Tibetan-transcription.
;;;###autoload
(defun tibetan-tibetan-to-transcription (str)
"Transcribe Tibetan string STR and return the corresponding Roman string."
(let (;; Accumulate transcriptions here in reverse order.
(trans nil)
(len (length str))
(i 0)
ch this-trans)
(while (< i len)
(let ((idx (string-match tibetan-precomposition-rule-regexp str i)))
(if (eq idx i)
;; Ith character and the followings matches precomposable
;; Tibetan sequence.
(setq i (match-end 0)
this-trans
(car (rassoc
(cdr (assoc (match-string 0 str)
tibetan-precomposition-rule-alist))
tibetan-precomposed-transcription-alist)))
(setq ch (substring str i (1+ i))
i (1+ i)
this-trans
(car (or (rassoc ch tibetan-consonant-transcription-alist)
(rassoc ch tibetan-vowel-transcription-alist)
(rassoc ch tibetan-subjoined-transcription-alist)))))
(setq trans (cons this-trans trans))))
(apply 'concat (nreverse trans))))
;;;###autoload
(defun tibetan-transcription-to-tibetan (str)
"Convert Tibetan Roman string STR to Tibetan character string.
The returned string has no composition information."
(let (;; Case is significant.
(case-fold-search nil)
(idx 0)
;; Accumulate Tibetan strings here in reverse order.
(t-str-list nil)
i subtrans)
(while (setq i (string-match tibetan-regexp str idx))
(if (< idx i)
;; STR contains a pattern that doesn't match Tibetan
;; transcription. Include the pattern as is.
(setq t-str-list (cons (substring str idx i) t-str-list)))
(setq subtrans (match-string 0 str)
idx (match-end 0))
(let ((t-char (cdr (assoc subtrans
tibetan-precomposed-transcription-alist))))
(if t-char
;; SUBTRANS corresponds to a transcription for
;; precomposable Tibetan sequence.
(setq t-char (car (rassoc t-char
tibetan-precomposition-rule-alist)))
(setq t-char
(cdr
(or (assoc subtrans tibetan-consonant-transcription-alist)
(assoc subtrans tibetan-vowel-transcription-alist)
(assoc subtrans tibetan-modifier-transcription-alist)
(assoc subtrans tibetan-subjoined-transcription-alist)))))
(setq t-str-list (cons t-char t-str-list))))
(if (< idx (length str))
(setq t-str-list (cons (substring str idx) t-str-list)))
(apply 'concat (nreverse t-str-list))))
;;;
;;; Functions for composing/decomposing Tibetan sequence.
;;;
;;; A Tibetan syllable is typically structured as follows:
;;;
;;; [Prefix] C [C+] V [M] [Suffix [Post suffix]]
;;;
;;; where C's are all vertically stacked, V appears below or above
;;; consonant cluster and M is always put above the C[C+]V combination.
;;; (Sanskrit visarga, though it is a vowel modifier, is considered
;;; to be a punctuation.)
;;;
;;; Here are examples of the words "bsgrubs" and "hfauM"
;;;
;;; $(7"7"G###C"U"7"G(B $(7"H"R"U"_(B
;;;
;;; M
;;; b s b s h
;;; g fa
;;; r u
;;; u
;;;
;;; Consonants `'' ($(7"A(B), `w' ($(7">(B), `y' ($(7"B(B), `r' ($(7"C(B) take special
;;; forms when they are used as subjoined consonant. Consonant `r'
;;; takes another special form when used as superjoined in such a case
;;; as "rka", while it does not change its form when conjoined with
;;; subjoined `'', `w' or `y' as in "rwa", "rya".
;; Append a proper composition rule and glyph to COMPONENTS to compose
;; CHAR with a composition that has COMPONENTS.
(defun tibetan-add-components (components char)
(let ((last (last components))
(stack-upper '(tc . bc))
(stack-under '(bc . tc))
rule comp-vowel tmp)
;; Special treatment for 'a chung.
;; If 'a follows a consonant, turn it into the subjoined form.
;; * Disabled by Tomabechi 2000/06/09 *
;; Because in Unicode, $(7"A(B may follow directly a consonant without
;; any intervening vowel, as in $(7"9"""Q"A!;(B=$(7"9(B $(7""(B $(7"A(B not $(7"9(B $(7""(B $(7"Q(B $(7"A(B
;;(if (and (= char ?$(7"A(B)
;; (aref (char-category-set (car last)) ?0))
;; (setq char ?$(7"R(B)) ;; modified for new font by Tomabechi 1999/12/10
;; Composite vowel signs are decomposed before being added
;; Added by Tomabechi 2000/06/08
(if (memq char '(?$(7"T(B ?$(7"V(B ?$(7"W(B ?$(7"X(B ?$(7"Y(B ?$(7"Z(B ?$(7"b(B))
(setq comp-vowel
(copy-sequence
(cddr (assoc (char-to-string char)
tibetan-composite-vowel-alist)))
char
(cadr (assoc (char-to-string char)
tibetan-composite-vowel-alist))))
(cond
;; Compose upper vowel sign vertically over.
((aref (char-category-set char) ?2)
(setq rule stack-upper))
;; Compose lower vowel sign vertically under.
((aref (char-category-set char) ?3)
(if (or (eq char ?$(7"Q(B) ;; `$(7"Q(B' and `$,1FP(B' should not visible when composed.
(eq char #xF70))
(setq rule nil)
(setq rule stack-under)))
;; Transform ra-mgo (superscribed r) if followed by a subjoined
;; consonant other than w, ', y, r.
((and (= (car last) ?$(7"C(B)
(not (memq char '(?$(7#>(B ?$(7"R(B ?$(7#B(B ?$(7#C(B))))
(setcar last ?$(7!"(B) ;; modified for newfont by Tomabechi 1999/12/10
(setq rule stack-under))
;; Transform initial base consonant if followed by a subjoined
;; consonant but 'a.
(t
(let ((laststr (char-to-string (car last))))
(if (and (/= char ?$(7"R(B) ;; modified for new font by Tomabechi
(string-match "[$(7"!(B-$(7"="?"@"D(B-$(7"J"K(B]" laststr))
(setcar last (string-to-char
(cdr (assoc (char-to-string (car last))
tibetan-base-to-subjoined-alist)))))
(setq rule stack-under))))
(if rule
(setcdr last (list rule char)))
;; Added by Tomabechi 2000/06/08
(if comp-vowel
(nconc last comp-vowel))
))
;;;###autoload
(defun tibetan-compose-string (str)
"Compose Tibetan string STR."
(let ((idx 0))
;; `$(7"A(B' is included in the pattern for subjoined consonants
;; because we treat it specially in tibetan-add-components.
;; (This feature is removed by Tomabechi 2000/06/08)
(while (setq idx (string-match tibetan-composable-pattern str idx))
(let ((from idx)
(to (match-end 0))
components)
(if (eq (string-match tibetan-precomposition-rule-regexp str idx) idx)
(setq idx (match-end 0)
components
(list (string-to-char
(cdr
(assoc (match-string 0 str)
tibetan-precomposition-rule-alist)))))
(setq components (list (aref str idx))
idx (1+ idx)))
(while (< idx to)
(tibetan-add-components components (aref str idx))
(setq idx (1+ idx)))
(compose-string str from to components))))
str)
;;;###autoload
(defun tibetan-compose-region (beg end)
"Compose Tibetan text the region BEG and END."
(interactive "r")
(let (str result chars)
(save-excursion
(save-restriction
(narrow-to-region beg end)
(goto-char (point-min))
;; `$(7"A(B' is included in the pattern for subjoined consonants
;; because we treat it specially in tibetan-add-components.
;; (This feature is removed by Tomabechi 2000/06/08)
(while (re-search-forward tibetan-composable-pattern nil t)
(let ((from (match-beginning 0))
(to (match-end 0))
components)
(goto-char from)
(if (looking-at tibetan-precomposition-rule-regexp)
(progn
(setq components
(list (string-to-char
(cdr
(assoc (match-string 0)
tibetan-precomposition-rule-alist)))))
(goto-char (match-end 0)))
(setq components (list (char-after from)))
(forward-char 1))
(while (< (point) to)
(tibetan-add-components components (following-char))
(forward-char 1))
(compose-region from to components)))))))
(defvar tibetan-decompose-precomposition-alist
(mapcar (function (lambda (x) (cons (string-to-char (cdr x)) (car x))))
tibetan-precomposition-rule-alist))
;;;###autoload
(defun tibetan-decompose-region (from to)
"Decompose Tibetan text in the region FROM and TO.
This is different from decompose-region because precomposed Tibetan characters
are decomposed into normal Tibetan character sequences."
(interactive "r")
(save-restriction
(narrow-to-region from to)
(decompose-region from to)
(goto-char from)
(while (not (eobp))
(let* ((char (following-char))
(slot (assq char tibetan-decompose-precomposition-alist)))
(if slot
(progn
(delete-char 1)
(insert (cdr slot)))
(forward-char 1))))))
;;;###autoload
(defun tibetan-decompose-string (str)
"Decompose Tibetan string STR.
This is different from decompose-string because precomposed Tibetan characters
are decomposed into normal Tibetan character sequences."
(let ((new "")
(len (length str))
(idx 0)
char slot)
(while (< idx len)
(setq char (aref str idx)
slot (assq (aref str idx) tibetan-decompose-precomposition-alist)
new (concat new (if slot (cdr slot) (char-to-string char)))
idx (1+ idx)))
new))
;;;
;;; This variable is used to avoid repeated decomposition.
;;;
(setq-default tibetan-decomposed nil)
;;;###autoload
(defun tibetan-decompose-buffer ()
"Decomposes Tibetan characters in the buffer into their components.
See also the documentation of the function `tibetan-decompose-region'."
(interactive)
(make-local-variable 'tibetan-decomposed)
(cond ((not tibetan-decomposed)
(tibetan-decompose-region (point-min) (point-max))
(setq tibetan-decomposed t))))
;;;###autoload
(defun tibetan-compose-buffer ()
"Composes Tibetan character components in the buffer.
See also docstring of the function tibetan-compose-region."
(interactive)
(make-local-variable 'tibetan-decomposed)
(tibetan-compose-region (point-min) (point-max))
(setq tibetan-decomposed nil))
;;;###autoload
(defun tibetan-post-read-conversion (len)
(save-excursion
(save-restriction
(let ((buffer-modified-p (buffer-modified-p)))
(narrow-to-region (point) (+ (point) len))
(tibetan-compose-region (point-min) (point-max))
(set-buffer-modified-p buffer-modified-p)
(make-local-variable 'tibetan-decomposed)
(setq tibetan-decomposed nil)
(- (point-max) (point-min))))))
;;;###autoload
(defun tibetan-pre-write-conversion (from to)
(setq tibetan-decomposed-temp tibetan-decomposed)
(let ((old-buf (current-buffer)))
(set-buffer (generate-new-buffer " *temp*"))
(if (stringp from)
(insert from)
(insert-buffer-substring old-buf from to))
(if (not tibetan-decomposed-temp)
(tibetan-decompose-region (point-min) (point-max)))
;; Should return nil as annotations.
nil))
;;;
;;; Unicode-related definitions.
;;;
(defvar tibetan-canonicalize-for-unicode-alist
'(("$(7"Q(B" . "") ;; remove vowel a
("$(7"T(B" . "$(7"R"S(B") ;; decompose vowels whose use is ``discouraged'' in Unicode 3.0
("$(7"V(B" . "$(7"R"U(B")
("$(7"W(B" . "$(7#C"a(B")
("$(7"X(B" . "$(7#C"R"a(B")
("$(7"Y(B" . "$(7#D"a(B")
("$(7"Z(B" . "$(7#D"R"a(B")
("$(7"b(B" . "$(7"R"a(B"))
"Rules for canonicalizing Tibetan vowels for Unicode.")
(defvar tibetan-canonicalize-for-unicode-regexp
"[$(7"Q"T"V"W"X"Y"Z"b(B]"
"Regexp for Tibetan vowels to be canonicalized in Unicode.")
(defun tibetan-canonicalize-for-unicode-region (from to)
(save-restriction
(narrow-to-region from to)
(goto-char from)
(while (re-search-forward tibetan-canonicalize-for-unicode-regexp nil t)
(let (
;;(from (match-beginning 0))
;;(to (match-end 0))
(canonical-form
(cdr (assoc (match-string 0)
tibetan-canonicalize-for-unicode-alist))))
;;(goto-char from)
;;(delete-region from to)
;;(insert canonical-form)
(replace-match canonical-form)
))))
(defvar tibetan-strict-unicode t
"*Flag to control Tibetan canonicalizing for Unicode.
If non-nil, the vowel a is removed and composite vowels are decomposed
before writing buffer in Unicode. See also
`tibetan-canonicalize-for-unicode-regexp' and
`tibetan-canonicalize-for-unicode-alist'.")
;;;###autoload
(defun tibetan-pre-write-canonicalize-for-unicode (from to)
(let ((old-buf (current-buffer))
(strict-unicode tibetan-strict-unicode))
(set-buffer (generate-new-buffer " *temp*"))
(if (stringp from)
(insert from)
(insert-buffer-substring old-buf from to))
(if strict-unicode
(tibetan-canonicalize-for-unicode-region (point-min) (point-max)))
;; Should return nil as annotations.
nil))
(provide 'tibet-util)
;; arch-tag: 7a7333e8-1584-446c-b39c-a02b9def265d
;;; tibet-util.el ends here