1 ;;; composite.el --- support character composition
3 ;; Copyright (C) 1999, 2000, 2001, 2002, 2003, 2004, 2005, 2006, 2007
4 ;; National Institute of Advanced Industrial Science and Technology (AIST)
5 ;; Registration Number H14PRO021
7 ;; Keywords: mule, multilingual, character composition
9 ;; This file is part of GNU Emacs.
11 ;; GNU Emacs is free software; you can redistribute it and/or modify
12 ;; it under the terms of the GNU General Public License as published by
13 ;; the Free Software Foundation; either version 3, or (at your option)
16 ;; GNU Emacs is distributed in the hope that it will be useful,
17 ;; but WITHOUT ANY WARRANTY; without even the implied warranty of
18 ;; MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
19 ;; GNU General Public License for more details.
21 ;; You should have received a copy of the GNU General Public License
22 ;; along with GNU Emacs; see the file COPYING. If not, write to the
23 ;; Free Software Foundation, Inc., 51 Franklin Street, Fifth Floor,
24 ;; Boston, MA 02110-1301, USA.
30 (defconst reference-point-alist
31 '((tl .
0) (tc .
1) (tr .
2)
32 (Bl .
3) (Bc .
4) (Br .
5)
33 (bl .
6) (bc .
7) (br .
8)
34 (cl .
9) (cc .
10) (cr .
11)
35 (top-left .
0) (top-center .
1) (top-right .
2)
36 (base-left .
3) (base-center .
4) (base-right .
5)
37 (bottom-left .
6) (bottom-center .
7) (bottom-right .
8)
38 (center-left .
9) (center-center .
10) (center-right .
11)
39 ;; For backward compatibility...
40 (ml .
3) (mc .
10) (mr .
5)
41 (mid-left .
3) (mid-center .
10) (mid-right .
5))
42 "Alist of symbols vs integer codes of glyph reference points.
43 A glyph reference point symbol is to be used to specify a composition
44 rule in COMPONENTS argument to such functions as `compose-region'.
46 Meanings of glyph reference point codes are as follows:
48 0----1----2 <---- ascent 0:tl or top-left
49 | | 1:tc or top-center
51 | | 3:Bl or base-left 9:cl or center-left
52 9 10 11 <---- center 4:Bc or base-center 10:cc or center-center
53 | | 5:Br or base-right 11:cr or center-right
54 --3----4----5-- <-- baseline 6:bl or bottom-left
55 | | 7:bc or bottom-center
56 6----7----8 <---- descent 8:br or bottom-right
58 Glyph reference point symbols are to be used to specify composition
59 rule of the form \(GLOBAL-REF-POINT . NEW-REF-POINT), where
60 GLOBAL-REF-POINT is a reference point in the overall glyphs already
61 composed, and NEW-REF-POINT is a reference point in the new glyph to
64 For instance, if GLOBAL-REF-POINT is `br' (bottom-right) and
65 NEW-REF-POINT is `tc' (top-center), the overall glyph is updated as
66 follows (the point `*' corresponds to both reference points):
68 +-------+--+ <--- new ascent
72 -- | | |-- <--- baseline \(doesn't change)
76 +----+-----+ <--- new descent
78 A composition rule may have the form \(GLOBAL-REF-POINT
79 NEW-REF-POINT XOFF YOFF), where XOFF and YOFF specifies how much
80 to shift NEW-REF-POINT from GLOBAL-REF-POINT. In this case, XOFF
81 and YOFF are integers in the range -100..100 representing the
82 shifting percentage against the font size.")
86 (defun encode-composition-rule (rule)
87 "Encode composition rule RULE into an integer value.
88 RULE is a cons of global and new reference point symbols
89 \(see `reference-point-alist')."
91 ;; This must be compatible with C macro COMPOSITION_ENCODE_RULE
92 ;; defined in composite.h.
94 (if (and (integerp rule
) (< rule
144))
98 (let ((gref (car rule
))
101 (if (consp nref
) ; (GREF NREF XOFF YOFF)
103 (setq xoff
(nth 1 nref
)
106 (or (and (>= xoff -
100) (<= xoff
100)
107 (>= yoff -
100) (<= yoff
100))
108 (error "Invalid compostion rule: %s" rule
))
109 (setq xoff
(+ xoff
128) yoff
(+ yoff
128)))
111 (setq xoff
0 yoff
0))
113 (setq gref
(cdr (assq gref reference-point-alist
))))
115 (setq nref
(cdr (assq nref reference-point-alist
))))
116 (or (and (>= gref
0) (< gref
12) (>= nref
0) (< nref
12))
117 (error "Invalid composition rule: %S" rule
))
118 (logior (lsh xoff
16) (lsh yoff
8) (+ (* gref
12) nref
)))
119 (error "Invalid composition rule: %S" rule
))))
121 ;; Decode encoded composition rule RULE-CODE. The value is a cons of
122 ;; global and new reference point symbols.
123 ;; This must be compatible with C macro COMPOSITION_DECODE_RULE
124 ;; defined in composite.h.
126 (defun decode-composition-rule (rule-code)
127 (or (and (natnump rule-code
) (< rule-code
#x1000000
))
128 (error "Invalid encoded composition rule: %S" rule-code
))
129 (let ((xoff (lsh rule-code -
16))
130 (yoff (logand (lsh rule-code -
8) #xFF
))
132 (setq rule-code
(logand rule-code
#xFF
)
133 gref
(car (rassq (/ rule-code
12) reference-point-alist
))
134 nref
(car (rassq (% rule-code
12) reference-point-alist
)))
135 (or (and gref
(symbolp gref
) nref
(symbolp nref
))
136 (error "Invalid composition rule code: %S" rule-code
))
137 (if (and (= xoff
0) (= yoff
0))
139 (setq xoff
(- xoff
128) yoff
(- yoff
128))
140 (list gref xoff yoff nref
))))
142 ;; Encode composition rules in composition components COMPONENTS. The
143 ;; value is a copy of COMPONENTS, where composition rules (cons of
144 ;; global and new glyph reference point symbols) are replaced with
145 ;; encoded composition rules. Optional 2nd argument NOCOPY non-nil
146 ;; means don't make a copy but modify COMPONENTS directly.
148 (defun encode-composition-components (components &optional nocopy
)
150 (setq components
(copy-sequence components
)))
151 (if (vectorp components
)
152 (let ((len (length components
))
156 (encode-composition-rule (aref components i
)))
158 (let ((tail (cdr components
)))
161 (encode-composition-rule (car tail
)))
162 (setq tail
(nthcdr 2 tail
)))))
165 ;; Decode composition rule codes in composition components COMPONENTS.
166 ;; The value is a copy of COMPONENTS, where composition rule codes are
167 ;; replaced with composition rules (cons of global and new glyph
168 ;; reference point symbols). Optional 2nd argument NOCOPY non-nil
169 ;; means don't make a copy but modify COMPONENTS directly.
170 ;; It is assumed that COMPONENTS is a vector and is for rule-base
171 ;; composition, thus (2N+1)th elements are rule codes.
173 (defun decode-composition-components (components &optional nocopy
)
175 (setq components
(copy-sequence components
)))
176 (let ((len (length components
))
180 (decode-composition-rule (aref components i
)))
184 (defun compose-region (start end
&optional components modification-func
)
185 "Compose characters in the current region.
187 Characters are composed relatively, i.e. composed by overstricking or
188 stacking depending on ascent, descent and other properties.
190 When called from a program, expects these four arguments.
192 First two arguments START and END are positions (integers or markers)
193 specifying the region.
195 Optional 3rd argument COMPONENTS, if non-nil, is a character, a string
196 or a vector or list of integers and rules.
198 If it is a character, it is an alternate character to display instead
199 of the text in the region.
201 If it is a string, the elements are alternate characters.
203 If it is a vector or list, it is a sequence of alternate characters and
204 composition rules, where (2N)th elements are characters and (2N+1)th
205 elements are composition rules to specify how to compose (2N+2)th
206 elements with previously composed N glyphs.
208 A composition rule is a cons of global and new glyph reference point
209 symbols. See the documentation of `reference-point-alist' for more
212 Optional 4th argument MODIFICATION-FUNC is a function to call to
213 adjust the composition when it gets invalid because of a change of
214 text in the composition."
216 (let ((modified-p (buffer-modified-p))
217 (inhibit-read-only t
))
218 (if (or (vectorp components
) (listp components
))
219 (setq components
(encode-composition-components components
)))
220 (compose-region-internal start end components modification-func
)
221 (restore-buffer-modified-p modified-p
)))
223 (defun decompose-region (start end
)
224 "Decompose text in the current region.
226 When called from a program, expects two arguments,
227 positions (integers or markers) specifying the region."
229 (let ((modified-p (buffer-modified-p))
230 (inhibit-read-only t
))
231 (remove-text-properties start end
'(composition nil
))
232 (restore-buffer-modified-p modified-p
)))
234 (defun compose-string (string &optional start end components modification-func
)
235 "Compose characters in string STRING.
237 The return value is STRING with the `composition' property put on all
238 the characters in it.
240 Optional 2nd and 3rd arguments START and END specify the range of
241 STRING to be composed. They default to the beginning and the end of
244 Optional 4th argument COMPONENTS, if non-nil, is a character or a
245 sequence (vector, list, or string) of integers. See the function
246 `compose-region' for more detail.
248 Optional 5th argument MODIFICATION-FUNC is a function to call to
249 adjust the composition when it gets invalid because of a change of
250 text in the composition."
251 (if (or (vectorp components
) (listp components
))
252 (setq components
(encode-composition-components components
)))
253 (or start
(setq start
0))
254 (or end
(setq end
(length string
)))
255 (compose-string-internal string start end components modification-func
)
258 (defun decompose-string (string)
259 "Return STRING where `composition' property is removed."
260 (remove-text-properties 0 (length string
) '(composition nil
) string
)
263 (defun compose-chars (&rest args
)
264 "Return a string from arguments in which all characters are composed.
265 For relative composition, arguments are characters.
266 For rule-based composition, Mth \(where M is odd) arguments are
267 characters, and Nth \(where N is even) arguments are composition rules.
268 A composition rule is a cons of glyph reference points of the form
269 \(GLOBAL-REF-POINT . NEW-REF-POINT). See the documentation of
270 `reference-point-alist' for more detail."
271 (let (str components
)
272 (if (consp (car (cdr args
)))
273 ;; Rule-base composition.
274 (let ((len (length args
))
275 (tail (encode-composition-components args
'nocopy
)))
278 (setq str
(cons (car tail
) str
))
279 (setq tail
(nthcdr 2 tail
)))
280 (setq str
(concat (nreverse str
))
282 ;; Relative composition.
283 (setq str
(concat args
)))
284 (compose-string-internal str
0 (length str
) components
)))
286 (defun find-composition (pos &optional limit string detail-p
)
287 "Return information about a composition at or nearest to buffer position POS.
289 If the character at POS has `composition' property, the value is a list
290 of FROM, TO, and VALID-P.
292 FROM and TO specify the range of text that has the same `composition'
293 property, VALID-P is non-nil if and only if this composition is valid.
295 If there's no composition at POS, and the optional 2nd argument LIMIT
296 is non-nil, search for a composition toward LIMIT.
298 If no composition is found, return nil.
300 Optional 3rd argument STRING, if non-nil, is a string to look for a
301 composition in; nil means the current buffer.
303 If a valid composition is found and the optional 4th argument DETAIL-P
304 is non-nil, the return value is a list of FROM, TO, COMPONENTS,
305 RELATIVE-P, MOD-FUNC, and WIDTH.
307 COMPONENTS is a vector of integers, the meaning depends on RELATIVE-P.
309 RELATIVE-P is t if the composition method is relative, else nil.
311 If RELATIVE-P is t, COMPONENTS is a vector of characters to be
312 composed. If RELATIVE-P is nil, COMPONENTS is a vector of characters
313 and composition rules as described in `compose-region'.
315 MOD-FUNC is a modification function of the composition.
317 WIDTH is a number of columns the composition occupies on the screen."
318 (let ((result (find-composition-internal pos limit string detail-p
)))
319 (if (and detail-p result
(nth 2 result
) (not (nth 3 result
)))
320 ;; This is a valid rule-base composition.
321 (decode-composition-components (nth 2 result
) 'nocopy
))
325 (defun compose-chars-after (pos &optional limit object
)
326 "Compose characters in current buffer after position POS.
328 It looks up the char-table `composition-function-table' (which see) by
329 a character after POS. If non-nil value is found, the format of the
330 value should be an alist of PATTERNs vs FUNCs, where PATTERNs are
331 regular expressions and FUNCs are functions. If the text after POS
332 matches one of PATTERNs, call the corresponding FUNC with three
333 arguments POS, TO, and PATTERN, where TO is the end position of text
334 matching PATTERN, and return what FUNC returns. Otherwise, return
337 FUNC is responsible for composing the text properly. The return value
339 nil -- if no characters were composed.
340 CHARS (integer) -- if CHARS characters were composed.
342 Optional 2nd arg LIMIT, if non-nil, limits the matching of text.
344 Optional 3rd arg OBJECT, if non-nil, is a string that contains the
345 text to compose. In that case, POS and LIMIT index into the string.
347 This function is the default value of `compose-chars-after-function'."
348 (let ((tail (aref composition-function-table
(char-after pos
)))
353 (while (and tail
(not func
))
354 (setq pattern
(car (car tail
))
355 func
(cdr (car tail
)))
358 (and (re-search-forward pattern limit t
)
359 (= (match-beginning 0) pos
))
360 (looking-at pattern
))
361 (setq result
(funcall func pos
(match-end 0) pattern nil
))
362 (setq func nil tail
(cdr tail
)))))))
365 (defun compose-last-chars (args)
366 "Compose last characters.
367 The argument is a parameterized event of the form
368 \(compose-last-chars N COMPONENTS),
369 where N is the number of characters before point to compose,
370 COMPONENTS, if non-nil, is the same as the argument to `compose-region'
371 \(which see). If it is nil, `compose-chars-after' is called,
372 and that function finds a proper rule to compose the target characters.
373 This function is intended to be used from input methods.
374 The global keymap binds special event `compose-last-chars' to this
375 function. Input method may generate an event (compose-last-chars N COMPONENTS)
376 after a sequence of character events."
378 (let ((chars (nth 1 args
)))
379 (if (and (numberp chars
)
380 (>= (- (point) (point-min)) chars
))
382 (compose-region (- (point) chars
) (point) (nth 2 args
))
383 (compose-chars-after (- (point) chars
) (point))))))
385 (global-set-key [compose-last-chars
] 'compose-last-chars
)
388 ;;; Automatic character composition.
390 (defvar composition-function-table
391 (make-char-table nil
)
392 "Char table of functions for automatic character composition.
393 For each character that has to be composed automatically with
394 preceding and/or following characters, this char table contains
395 a function to call to compose that character.
397 An element, if non-nil, is FUNC or an alist of PATTERNs vs FUNCs,
398 where PATTERNs are regular expressions and FUNCs are functions.
399 If the element is FUNC, FUNC itself determines the region to
402 Each function is called with 5 arguments, FROM, TO, FONT-OBJECT,
405 If STRING is nil, FROM and TO are positions specifying the region
406 maching with PATTERN in the current buffer, and the function has
407 to compose character in that region (possibly with characters
408 preceding FROM). The return value of the function is the end
409 position where characters are composed.
411 Otherwise, STRING is a string, and FROM and TO are indices into
412 the string. In this case, the function has to compose a
413 character in the string.
415 FONT-OBJECT may be nil if not available (e.g. for the case of
418 See also the command `toggle-auto-composition'.")
420 ;; Copied from font-lock.el.
422 ;; Borrowed from lazy-lock.el.
423 ;; We use this to preserve or protect things when modifying text properties.
424 (defmacro save-buffer-state
(varlist &rest body
)
425 "Bind variables according to VARLIST and eval BODY restoring buffer state."
426 `(let* ,(append varlist
427 '((modified (buffer-modified-p)) (buffer-undo-list t
)
428 (inhibit-read-only t
) (inhibit-point-motion-hooks t
)
429 (inhibit-modification-hooks t
)
430 deactivate-mark buffer-file-name buffer-file-truename
))
433 (restore-buffer-modified-p nil
))))
434 ;; Fixme: This makes bootstrapping fail with this error.
435 ;; Symbol's function definition is void: eval-defun
436 ;;(def-edebug-spec save-buffer-state let)
439 (put 'save-buffer-state
'lisp-indent-function
1)
441 (defun terminal-composition-function (from to font-object string
)
442 "General composition function used on terminal.
443 Non-spacing characters are composed with the preceding spacing
444 character. All non-spacing characters has this function in
445 `terminal-composition-function-table'."
446 (let ((pos (1+ from
)))
449 (while (and (< pos to
)
450 (= (aref char-width-table
(aref string pos
)) 0))
453 (compose-string string
(1- from
) pos
)
454 (compose-string string from pos
455 (concat " " (buffer-substring from pos
)))))
456 (while (and (< pos to
)
457 (= (aref char-width-table
(char-after pos
)) 0))
459 (if (> from
(point-min))
460 (compose-region (1- from
) pos
(buffer-substring from pos
))
461 (compose-region from pos
462 (concat " " (buffer-substring from pos
)))))
465 (defvar terminal-composition-function-table
466 (let ((table (make-char-table nil
)))
469 (if (= val
0) (set-char-table-range table key
470 'terminal-composition-function
)))
473 "Char table of functions for automatic character composition on terminal.
474 This is like `composition-function-table' but used when Emacs is running
477 (defun auto-compose-chars (from to window string
)
478 "Compose characters in the region between FROM and TO.
479 WINDOW is a window displaying the current buffer.
480 If STRING is non-nil, it is a string, and FROM and TO are indices
481 into the string. In that case, compose characters in the string.
483 This function is the default value of `auto-composition-function' (which see)."
484 (save-buffer-state nil
488 (let ((table (if (display-graphic-p)
489 composition-function-table
490 terminal-composition-function-table
))
492 (setq to
(or (text-property-any (1+ from
) to
'auto-composed t
497 (let* ((ch (aref string from
))
498 (elt (aref table ch
))
502 (setq font-obj
(font-at from window string
)))
504 (setq newpos
(funcall elt from to font-obj string
))
506 (or (not (eq (string-match (caar elt
) string
510 (funcall (cdar elt
) from
513 (setq elt
(cdr elt
)))))
514 (if (and newpos
(> newpos from
))
516 (setq from
(1+ from
)))))
517 (narrow-to-region from to
)
519 (let* ((ch (char-after from
))
520 (elt (aref table ch
))
521 func pattern font-obj newpos
)
524 (setq font-obj
(font-at from window
)))
526 (setq newpos
(funcall elt from to font-obj nil
))
529 (or (not (looking-at (caar elt
)))
531 (funcall (cdar elt
) from
534 (setq elt
(cdr elt
)))))
535 (if (and newpos
(> newpos from
))
537 (setq from
(1+ from
))))))
538 (put-text-property start to
'auto-composed t string
)))))))
540 (make-variable-buffer-local 'auto-composition-function
)
543 (define-minor-mode auto-composition-mode
544 "Toggle Auto Compostion mode.
545 With arg, turn Auto Compostion mode off if and only if arg is a non-positive
546 number; if arg is nil, toggle Auto Compostion mode; anything else turns Auto
549 When Auto Composition is enabled, text characters are automatically composed
550 by functions registered in `composition-function-table' (which see).
552 You can use Global Auto Composition mode to automagically turn on
553 Auto Composition mode in all buffers (this is the default)."
556 (setq auto-composition-mode nil
))
557 (cond (auto-composition-mode
558 (add-hook 'after-change-functions
'auto-composition-after-change nil t
)
559 (setq auto-composition-function
'auto-compose-chars
))
561 (remove-hook 'after-change-functions
'auto-composition-after-change t
)
562 (setq auto-composition-function nil
)))
563 (save-buffer-state nil
566 (remove-text-properties (point-min) (point-max) '(auto-composed nil
))
567 (decompose-region (point-min) (point-max)))))
569 (defun auto-composition-after-change (start end old-len
)
570 (save-buffer-state nil
571 (if (< start
(point-min))
572 (setq start
(point-min)))
573 (if (> end
(point-max))
574 (setq end
(point-max)))
575 (when (and auto-composition-mode
(not memory-full
))
577 (when (and (> start
(point-min))
578 (setq func2
(aref composition-function-table
579 (char-after (1- start
))))
580 (or (= start
(point-max))
581 (not (setq func1
(aref composition-function-table
582 (char-after start
))))
584 (setq start
(1- start
)
586 (while (eq func1 func2
)
587 (if (> start
(point-min))
588 (setq start
(1- start
)
589 func2
(aref composition-function-table
592 (when (and (< end
(point-max))
593 (setq func2
(aref composition-function-table
595 (or (= end
(point-min))
596 (not (setq func1
(aref composition-function-table
597 (char-after (1- end
)))))
601 (while (eq func1 func2
)
602 (if (< end
(point-max))
603 (setq func2
(aref composition-function-table
608 (remove-text-properties start end
'(auto-composed nil
)))))))
610 (defun turn-on-auto-composition-if-enabled ()
611 (if enable-multibyte-characters
612 (auto-composition-mode 1)))
615 (define-global-minor-mode global-auto-composition-mode
616 auto-composition-mode turn-on-auto-composition-if-enabled
618 :initialize
'custom-initialize-safe-default
619 :init-value
(not noninteractive
)
620 :group
'auto-composition
623 (defun toggle-auto-composition (&optional arg
)
624 "Change whether automatic character composition is enabled in this buffer.
625 With arg, enable it iff arg is positive."
627 (let ((enable (if (null arg
) (not auto-composition-function
)
628 (> (prefix-numeric-value arg
) 0))))
630 (kill-local-variable 'auto-composition-function
)
631 (make-local-variable 'auto-composition-function
)
632 (setq auto-composition-function nil
)
633 (save-buffer-state nil
636 (decompose-region (point-min) (point-max)))))
638 (save-buffer-state nil
641 (remove-text-properties (point-min) (point-max)
642 '(auto-composed nil
))))))
644 (defun auto-compose-region (from to
)
645 "Force automatic character composition on the region FROM and TO."
647 (if (get-text-property from
'auto-composed
)
648 (setq from
(next-single-property-change from
'auto-composed nil to
)))
650 (let ((modified-p (buffer-modified-p))
651 (inhibit-read-only '(composition auto-composed
))
652 (stop (next-single-property-change (point) 'auto-composed nil to
)))
653 (while (< (point) to
)
656 (goto-char (next-single-property-change (point)
657 'auto-composed nil to
))
658 (setq stop
(next-single-property-change (point)
659 'auto-composed nil to
)))
660 (let ((func (aref composition-function-table
(following-char)))
663 (goto-char (funcall func
(point) nil
)))
666 (put-text-property from to
'auto-composed t
)
667 (set-buffer-modified-p modified-p
))))
670 ;; The following codes are only for backward compatibility with Emacs
673 (defun decompose-composite-char (char &optional type with-composition-rule
)
674 "Convert CHAR to string.
676 If optional 2nd arg TYPE is non-nil, it is `string', `list', or
677 `vector'. In this case, CHAR is converted to string, list of CHAR, or
678 vector of CHAR respectively.
679 Optional 3rd arg WITH-COMPOSITION-RULE is ignored."
680 (cond ((or (null type
) (eq type
'string
)) (char-to-string char
))
681 ((eq type
'list
) (list char
))
684 (make-obsolete 'decompose-composite-char
'char-to-string
"21.1")
688 ;; arch-tag: ee703d77-1723-45d4-a31f-e9f0f867aa33
689 ;;; composite.el ends here