1 ;;; thai-word.el -- find Thai word boundaries
3 ;; Copyright (C) 2000, 2001, 2002, 2003, 2004
4 ;; Electrotechnical Laboratory, JAPAN.
6 ;; Author: Kenichi HANDA <handa@etl.go.jp>
8 ;; Keywords: thai, word break, emacs
10 ;; This program is free software; you can redistribute it and/or modify
11 ;; it under the terms of the GNU General Public License as published by
12 ;; the Free Software Foundation; either version 2, or (at your option)
15 ;; This program is distributed in the hope that it will be useful,
16 ;; but WITHOUT ANY WARRANTY; without even the implied warranty of
17 ;; MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
18 ;; GNU General Public License for more details.
20 ;; You should have received a copy of the GNU General Public License
21 ;; along with GNU Emacs; see the file COPYING. If not, write to
22 ;; the Free Software Foundation, 675 Mass Ave, Cambridge, MA 02139, USA.
24 ;; The used Thai word list has been taken from IBM's ICU4J project
25 ;; (file `thai6.ucs', version 1.4, converted to TIS encoding, with
26 ;; removal of three incorrect entries) to which the following license
29 ;; COPYRIGHT AND PERMISSION NOTICE
32 ;; Copyright (c) 1995-2001 International Business Machines
33 ;; Corporation and others
35 ;; All rights reserved.
38 ;; Permission is hereby granted, free of charge, to any person
39 ;; obtaining a copy of this software and associated documentation
40 ;; files (the "Software"), to deal in the Software without
41 ;; restriction, including without limitation the rights to use,
42 ;; copy, modify, merge, publish, distribute, and/or sell copies of
43 ;; the Software, and to permit persons to whom the Software is
44 ;; furnished to do so, provided that the above copyright notice(s)
45 ;; and this permission notice appear in all copies of the Software
46 ;; and that both the above copyright notice(s) and this permission
47 ;; notice appear in supporting documentation.
49 ;; THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
50 ;; EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES
51 ;; OF MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
52 ;; NONINFRINGEMENT OF THIRD PARTY RIGHTS. IN NO EVENT SHALL THE
53 ;; COPYRIGHT HOLDER OR HOLDERS INCLUDED IN THIS NOTICE BE LIABLE
54 ;; FOR ANY CLAIM, OR ANY SPECIAL INDIRECT OR CONSEQUENTIAL DAMAGES,
55 ;; OR ANY DAMAGES WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR
56 ;; PROFITS, WHETHER IN AN ACTION OF CONTRACT, NEGLIGENCE OR OTHER
57 ;; TORTIOUS ACTION, ARISING OUT OF OR IN CONNECTION WITH THE USE OR
58 ;; PERFORMANCE OF THIS SOFTWARE.
60 ;; Except as contained in this notice, the name of a copyright
61 ;; holder shall not be used in advertising or otherwise to promote
62 ;; the sale, use or other dealings in this Software without prior
63 ;; written authorization of the copyright holder.
66 ;; This file implements an algorithm to find Thai word breaks using a
67 ;; dictionary. It is based on the C program `cttex' written by
68 ;; Vuthichai Ampornaramveth <vuthi@nii.ac.jp>.
71 ;; Table of Thai words. It is a nested alist (see `nested-alist-p'),
72 ;; which means that you can easily index the list character by
75 (defconst thai-word-table nil
)
78 ;; Set up `thai-word-table'.
6226 "ÊÁºÙóÒÒÊÔ·¸ÔÃÒªÂì"
10736 (setq thai-word-table
(list 'thai-words
))
10738 (set-nested-alist (car l
) 1 thai-word-table
)
10742 (defun thai-update-word-table (file &optional append
)
10743 "Update Thai word table by replacing the current word list with
10744 FILE. If called with a prefix argument, FILE is appended instead to
10745 the current word list."
10746 (interactive "FThai word table file: \nP")
10747 (let ((buf (generate-new-buffer "*thai-work*"))
10748 (coding-system-for-read 'thai-tis620
)
10749 (table (if append thai-word-table
(list 'thai-words
))))
10753 (insert-file-contents file
)
10754 (goto-char (point-min))
10755 (while (re-search-forward "\\ct+" nil t
)
10756 (set-nested-alist (match-string 0) 1 table
)))
10758 (setq thai-word-table table
)))
10761 ;; Two special Thai characters regarded as suffix of words.
10763 (defconst thai-MaiYaMok
(make-char 'thai-tis620 ?
\xE6))
10764 (defconst thai-PaiYanNoi
(make-char 'thai-tis620 ?
\xCF))
10767 ;; Find Thai words starting at POS and return a list of positions of
10768 ;; the Thai word ends. It doesn't move point. LIMIT limits the
10769 ;; maximum position. IGNORE is a list of positions to ignore. It is
10770 ;; assumed that all following characters to LIMIT are Thai. If the
10771 ;; following char is not Thai (i.e., POS is equal to LIMIT), return t.
10773 ;; Note that the longest word position comes first.
10775 (defun thai-find-word-ends (pos limit
&optional ignore
)
10778 (let* ((char (char-after pos
))
10779 (this (cdr (assq char
(cdr thai-word-table
))))
10782 ;; Look up the following character sequence in `thai-word-table'
10783 ;; character by character.
10786 char
(char-after pos
)
10787 category-set
(char-category-set char
))
10788 ;; If the current sequence is recorded in `thai-word-table'
10789 ;; (i.e. (car THIS) is 1) and the following Thai character is
10790 ;; not an upper-vowel, lower-vowel, or tone-mark, we have
10791 ;; found a possible word ending position.
10792 (if (and (eq (car this
) 1)
10793 (not (or (aref category-set ?
2)
10794 (aref category-set ?
3)
10795 (aref category-set ?
4))))
10797 ;; Skip possible Thai suffices.
10798 (while (or (eq char thai-MaiYaMok
) (eq char thai-PaiYanNoi
))
10800 char
(char-after pos
)))
10801 ;; Skip character positions in IGNORE list.
10802 (or (memq pos ignore
)
10803 (setq positions
(cons pos positions
)))))
10804 ;; Set up next loop.
10805 (setq this
(and (< pos limit
) (cdr (assq char this
)))))
10809 ;; Move point forward to the end of Thai word which follows point and
10810 ;; update VEC. VEC is a vector of three elements used to cache word
10811 ;; end positions. The Nth element, if non-nil, is a list of end
10812 ;; points of the Nth word, or t indicating that there is no Thai
10813 ;; character. LIMIT limits the point movement.
10815 (defun thai-forward-word-update-info (vec limit
)
10816 (let ((pos (point))
10822 ;; If four succeeding Thai words are found, throw t, otherwise
10825 ;; Start with first vector element.
10826 (setq v0
(aref vec
0))
10828 ;; Update VEC if V0 is empty.
10829 (setq v0
(thai-find-word-ends pos limit
))
10832 ;; In case we haven't found any wordbreaks resp. point has
10833 ;; reached LIMIT, exit the catch body.
10834 (if (symbolp v0
) ; i.e. nil or t?
10836 ;; OK, V0 holds possible word ends for the current position.
10837 ;; We save V0 for later reference.
10839 (setq v1
(aref vec
1))
10840 ;; Now we try all end word positions to find the next word.
10842 (setq pos
(car v0
))
10844 ;; Update VEC if V1 is empty, ignoring positions already
10845 ;; found -- for Thai, we need the longest match, so if
10848 ;; start(long-word) = start(short-word1)
10849 ;; end(short-word1) = start(short-word2)
10850 ;; end(short-word2) = end(long-word)
10852 ;; only long-word is used.
10853 (setq v1
(thai-find-word-ends pos limit tried
))
10856 ;; If point has reached LIMIT, exit the catch body.
10859 ;; Save SECOND-BEST, if this hasn't been done already.
10860 ;; The `second best' solution is the end position of the
10861 ;; longest first word followed by the longest second word.
10864 (setq second-best
(cons v0 v1
)))
10865 ;; Update the already tried end word positions.
10866 (setq tried
(append tried v1
))
10867 ;; Now repeat the whole process to find a third word.
10869 (setq v2
(aref vec
2))
10871 (setq pos
(car v1
))
10873 (setq v2
(thai-find-word-ends pos limit tried
))
10877 (setq tried
(append tried v2
))
10878 ;; And the same for a fourth word.
10880 (setq pos
(car v2
))
10881 (setq v3
(thai-find-word-ends pos limit tried
))
10884 (setq v2
(cdr v2
)))
10885 (setq v1
(cdr v1
))))
10886 (setq v0
(cdr v0
)))
10890 ;; We found four succeeding Thai words (or LIMIT has been
10891 ;; reached). Move to the end of the first word.
10892 (goto-char (car v0
))
10893 ;; Update VEC for the next function call. If no larger word
10894 ;; positions have been found, set the corresponding vector
10896 (if (and (consp v1
) (< (car v1
) (car (aref vec
1))))
10899 (if (and (consp v2
) (< (car v2
) (car (aref vec
2))))
10902 (aset vec
2 v3
)))) ; exit function successfully
10904 ;; We didn't find four consecutive words. If we have found a
10905 ;; `second best' solution and the length of those two words is
10906 ;; longer than the longest word we can see at the current point,
10907 ;; adopt the second best solution. This decision is based on
10908 ;; heuristic tests.
10909 (if (and second-best
10910 (< (car (aref vec
0)) (car (cdr second-best
))))
10912 (goto-char (car (car second-best
)))
10913 (aset vec
0 (cdr second-best
)))
10914 ;; We finally failed to find a word break. For Thai, the best
10915 ;; solution is to extend the first longest word so that the
10916 ;; end point starts a second word.
10917 (setq pos
(or (car (aref vec
0)) pos
))
10918 (while (and (< pos limit
)
10919 (not (setq positions
(thai-find-word-ends pos limit
))))
10920 (setq pos
(1+ pos
)))
10922 (aset vec
0 positions
))
10927 ;; Return a list of Thai word boundary positions after the current
10928 ;; point. LIMIT, if non-nil, limits the region to check.
10930 (defun thai-find-word-boundaries (&optional limit
)
10932 (setq limit
(point-max)))
10934 (let ((vec (make-vector 3 nil
))
10937 ;; Loop over all (consecutive) Thai regions by using the
10938 ;; character property `t' until LIMIT is reached.
10939 (while (and (< (point) limit
)
10940 (re-search-forward "\\ct+" nil t
))
10941 (setq this-limit
(point))
10942 (goto-char (match-beginning 0))
10943 (fillarray vec nil
)
10944 ;; Check the first word, initializing VEC.
10945 (thai-forward-word-update-info vec this-limit
)
10946 ;; Then loop over the remaining words in the current Thai
10947 ;; region, collecting the boundaries.
10948 (while (< (point) this-limit
)
10949 (setq boundaries
(cons (point) boundaries
))
10950 (thai-forward-word-update-info vec this-limit
)))
10954 (defun thai-break-words (separator &optional limit
)
10955 "Break Thai words by inserting a separator string at word boundaries."
10956 (interactive "sSeparator: ")
10958 (let ((boundaries (thai-find-word-boundaries limit
)))
10960 (goto-char (car boundaries
))
10962 (setq boundaries
(cdr boundaries
))))))
10965 (defun thai-forward-word (count)
10966 "Move point forward COUNT words considering Thai word boundaries.
10967 If COUNT is negative, move point backward (- COUNT) words."
10971 (skip-syntax-forward "^w")
10972 (if (looking-at "\\ct+")
10973 ;; We have reached a Thai region, so we must do something
10974 ;; special instead of using forward-word.
10975 (let ((start (point))
10976 (limit (match-end 0))
10979 ;; If thai-forward-word has been called within a Thai
10980 ;; region, we must go back until the Thai region starts
10981 ;; to do the contextual analysis for finding word
10983 (while (aref (char-category-set (preceding-char)) ?t
)
10985 ;; OK, we ask for the list of word boundaries in
10987 (setq boundaries
(nreverse (thai-find-word-boundaries limit
)))
10988 ;; Now we search for the next boundary after START.
10989 (while (and boundaries
(<= (car boundaries
) start
))
10990 (setq boundaries
(cdr boundaries
)))
10991 ;; Adjust loop for next while loop.
10992 (setq count
(1- count
))
10993 ;; Now we skip Thai words until the BOUNDARIES list is
10994 ;; empty or count-1 words have been passed.
10995 (if (not boundaries
)
10997 (while (and (> count
0)
10999 (setq boundaries
(cdr boundaries
)
11001 ;; If BOUNDARIES is empty, the word counter is not
11002 ;; zero yet (remember that we have decreased COUNT by
11003 ;; one), so we go to LIMIT. Otherwise go to next
11006 (goto-char (car boundaries
))
11007 (goto-char limit
))))
11010 (setq count
(1- count
))))
11011 ;; The symmetrical action for negative values.
11014 (skip-syntax-backward "^w")
11015 (if (aref (char-category-set (preceding-char)) ?t
)
11016 (let ((start (point))
11017 (limit (if (looking-at "\\ct+") (match-end 0)
11021 (while (aref (char-category-set (preceding-char)) ?t
)
11023 (setq boundaries
(thai-find-word-boundaries limit
))
11024 (while (and boundaries
(>= (car boundaries
) start
))
11025 (setq boundaries
(cdr boundaries
)))
11026 (setq count
(1+ count
))
11029 (while (and (< count
0) boundaries
)
11030 (setq boundaries
(cdr boundaries
)
11033 (goto-char (car boundaries
))))))
11035 (setq count
(1+ count
)))))))
11038 (defun thai-backward-word (count)
11039 "Move point backward COUNT words considering Thai word boundaries.
11040 If COUNT is negative, move point forward (- COUNT) words."
11042 (thai-forward-word (- count
)))
11045 (provide 'thai-word
)
11048 ;; Local Variables:
11052 ;; end of thai-word.el
11054 ;; arch-tag: 29927f02-e177-4224-a270-7e67210b038a