1 ;;; composite.el --- support character composition
3 ;; Copyright (C) 1999, 2000, 2001, 2002, 2003, 2004, 2005, 2006, 2007,
5 ;; National Institute of Advanced Industrial Science and Technology (AIST)
6 ;; Registration Number H14PRO021
8 ;; Keywords: mule, multilingual, character composition
10 ;; This file is part of GNU Emacs.
12 ;; GNU Emacs is free software; you can redistribute it and/or modify
13 ;; it under the terms of the GNU General Public License as published by
14 ;; the Free Software Foundation; either version 3, or (at your option)
17 ;; GNU Emacs is distributed in the hope that it will be useful,
18 ;; but WITHOUT ANY WARRANTY; without even the implied warranty of
19 ;; MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
20 ;; GNU General Public License for more details.
22 ;; You should have received a copy of the GNU General Public License
23 ;; along with GNU Emacs; see the file COPYING. If not, write to the
24 ;; Free Software Foundation, Inc., 51 Franklin Street, Fifth Floor,
25 ;; Boston, MA 02110-1301, USA.
31 (defconst reference-point-alist
32 '((tl .
0) (tc .
1) (tr .
2)
33 (Bl .
3) (Bc .
4) (Br .
5)
34 (bl .
6) (bc .
7) (br .
8)
35 (cl .
9) (cc .
10) (cr .
11)
36 (top-left .
0) (top-center .
1) (top-right .
2)
37 (base-left .
3) (base-center .
4) (base-right .
5)
38 (bottom-left .
6) (bottom-center .
7) (bottom-right .
8)
39 (center-left .
9) (center-center .
10) (center-right .
11)
40 ;; For backward compatibility...
41 (ml .
3) (mc .
10) (mr .
5)
42 (mid-left .
3) (mid-center .
10) (mid-right .
5))
43 "Alist of symbols vs integer codes of glyph reference points.
44 A glyph reference point symbol is to be used to specify a composition
45 rule in COMPONENTS argument to such functions as `compose-region'.
47 Meanings of glyph reference point codes are as follows:
49 0----1----2 <---- ascent 0:tl or top-left
50 | | 1:tc or top-center
52 | | 3:Bl or base-left 9:cl or center-left
53 9 10 11 <---- center 4:Bc or base-center 10:cc or center-center
54 | | 5:Br or base-right 11:cr or center-right
55 --3----4----5-- <-- baseline 6:bl or bottom-left
56 | | 7:bc or bottom-center
57 6----7----8 <---- descent 8:br or bottom-right
59 Glyph reference point symbols are to be used to specify composition
60 rule of the form \(GLOBAL-REF-POINT . NEW-REF-POINT), where
61 GLOBAL-REF-POINT is a reference point in the overall glyphs already
62 composed, and NEW-REF-POINT is a reference point in the new glyph to
65 For instance, if GLOBAL-REF-POINT is `br' (bottom-right) and
66 NEW-REF-POINT is `tc' (top-center), the overall glyph is updated as
67 follows (the point `*' corresponds to both reference points):
69 +-------+--+ <--- new ascent
73 -- | | |-- <--- baseline \(doesn't change)
77 +----+-----+ <--- new descent
79 A composition rule may have the form \(GLOBAL-REF-POINT
80 NEW-REF-POINT XOFF YOFF), where XOFF and YOFF specifies how much
81 to shift NEW-REF-POINT from GLOBAL-REF-POINT. In this case, XOFF
82 and YOFF are integers in the range -100..100 representing the
83 shifting percentage against the font size.")
87 (defun encode-composition-rule (rule)
88 "Encode composition rule RULE into an integer value.
89 RULE is a cons of global and new reference point symbols
90 \(see `reference-point-alist')."
92 ;; This must be compatible with C macro COMPOSITION_ENCODE_RULE
93 ;; defined in composite.h.
95 (if (and (integerp rule
) (< rule
144))
99 (let ((gref (car rule
))
102 (if (consp nref
) ; (GREF NREF XOFF YOFF)
104 (setq xoff
(nth 1 nref
)
107 (or (and (>= xoff -
100) (<= xoff
100)
108 (>= yoff -
100) (<= yoff
100))
109 (error "Invalid compostion rule: %s" rule
))
110 (setq xoff
(+ xoff
128) yoff
(+ yoff
128)))
112 (setq xoff
0 yoff
0))
114 (setq gref
(cdr (assq gref reference-point-alist
))))
116 (setq nref
(cdr (assq nref reference-point-alist
))))
117 (or (and (>= gref
0) (< gref
12) (>= nref
0) (< nref
12))
118 (error "Invalid composition rule: %S" rule
))
119 (logior (lsh xoff
16) (lsh yoff
8) (+ (* gref
12) nref
)))
120 (error "Invalid composition rule: %S" rule
))))
122 ;; Decode encoded composition rule RULE-CODE. The value is a cons of
123 ;; global and new reference point symbols.
124 ;; This must be compatible with C macro COMPOSITION_DECODE_RULE
125 ;; defined in composite.h.
127 (defun decode-composition-rule (rule-code)
128 (or (and (natnump rule-code
) (< rule-code
#x1000000
))
129 (error "Invalid encoded composition rule: %S" rule-code
))
130 (let ((xoff (lsh rule-code -
16))
131 (yoff (logand (lsh rule-code -
8) #xFF
))
133 (setq rule-code
(logand rule-code
#xFF
)
134 gref
(car (rassq (/ rule-code
12) reference-point-alist
))
135 nref
(car (rassq (% rule-code
12) reference-point-alist
)))
136 (or (and gref
(symbolp gref
) nref
(symbolp nref
))
137 (error "Invalid composition rule code: %S" rule-code
))
138 (if (and (= xoff
0) (= yoff
0))
140 (setq xoff
(- xoff
128) yoff
(- yoff
128))
141 (list gref xoff yoff nref
))))
143 ;; Encode composition rules in composition components COMPONENTS. The
144 ;; value is a copy of COMPONENTS, where composition rules (cons of
145 ;; global and new glyph reference point symbols) are replaced with
146 ;; encoded composition rules. Optional 2nd argument NOCOPY non-nil
147 ;; means don't make a copy but modify COMPONENTS directly.
149 (defun encode-composition-components (components &optional nocopy
)
151 (setq components
(copy-sequence components
)))
152 (if (vectorp components
)
153 (let ((len (length components
))
157 (encode-composition-rule (aref components i
)))
159 (let ((tail (cdr components
)))
162 (encode-composition-rule (car tail
)))
163 (setq tail
(nthcdr 2 tail
)))))
166 ;; Decode composition rule codes in composition components COMPONENTS.
167 ;; The value is a copy of COMPONENTS, where composition rule codes are
168 ;; replaced with composition rules (cons of global and new glyph
169 ;; reference point symbols). Optional 2nd argument NOCOPY non-nil
170 ;; means don't make a copy but modify COMPONENTS directly.
171 ;; It is assumed that COMPONENTS is a vector and is for rule-base
172 ;; composition, thus (2N+1)th elements are rule codes.
174 (defun decode-composition-components (components &optional nocopy
)
176 (setq components
(copy-sequence components
)))
177 (let ((len (length components
))
181 (decode-composition-rule (aref components i
)))
185 (defun compose-region (start end
&optional components modification-func
)
186 "Compose characters in the current region.
188 Characters are composed relatively, i.e. composed by overstricking or
189 stacking depending on ascent, descent and other properties.
191 When called from a program, expects these four arguments.
193 First two arguments START and END are positions (integers or markers)
194 specifying the region.
196 Optional 3rd argument COMPONENTS, if non-nil, is a character, a string
197 or a vector or list of integers and rules.
199 If it is a character, it is an alternate character to display instead
200 of the text in the region.
202 If it is a string, the elements are alternate characters.
204 If it is a vector or list, it is a sequence of alternate characters and
205 composition rules, where (2N)th elements are characters and (2N+1)th
206 elements are composition rules to specify how to compose (2N+2)th
207 elements with previously composed N glyphs.
209 A composition rule is a cons of global and new glyph reference point
210 symbols. See the documentation of `reference-point-alist' for more
213 Optional 4th argument MODIFICATION-FUNC is a function to call to
214 adjust the composition when it gets invalid because of a change of
215 text in the composition."
217 (let ((modified-p (buffer-modified-p))
218 (inhibit-read-only t
))
219 (if (or (vectorp components
) (listp components
))
220 (setq components
(encode-composition-components components
)))
221 (compose-region-internal start end components modification-func
)
222 (restore-buffer-modified-p modified-p
)))
224 (defun decompose-region (start end
)
225 "Decompose text in the current region.
227 When called from a program, expects two arguments,
228 positions (integers or markers) specifying the region."
230 (let ((modified-p (buffer-modified-p))
231 (inhibit-read-only t
))
232 (remove-text-properties start end
'(composition nil
))
233 (restore-buffer-modified-p modified-p
)))
235 (defun compose-string (string &optional start end components modification-func
)
236 "Compose characters in string STRING.
238 The return value is STRING with the `composition' property put on all
239 the characters in it.
241 Optional 2nd and 3rd arguments START and END specify the range of
242 STRING to be composed. They default to the beginning and the end of
245 Optional 4th argument COMPONENTS, if non-nil, is a character or a
246 sequence (vector, list, or string) of integers. See the function
247 `compose-region' for more detail.
249 Optional 5th argument MODIFICATION-FUNC is a function to call to
250 adjust the composition when it gets invalid because of a change of
251 text in the composition."
252 (if (or (vectorp components
) (listp components
))
253 (setq components
(encode-composition-components components
)))
254 (or start
(setq start
0))
255 (or end
(setq end
(length string
)))
256 (compose-string-internal string start end components modification-func
)
259 (defun decompose-string (string)
260 "Return STRING where `composition' property is removed."
261 (remove-text-properties 0 (length string
) '(composition nil
) string
)
264 (defun compose-chars (&rest args
)
265 "Return a string from arguments in which all characters are composed.
266 For relative composition, arguments are characters.
267 For rule-based composition, Mth \(where M is odd) arguments are
268 characters, and Nth \(where N is even) arguments are composition rules.
269 A composition rule is a cons of glyph reference points of the form
270 \(GLOBAL-REF-POINT . NEW-REF-POINT). See the documentation of
271 `reference-point-alist' for more detail."
272 (let (str components
)
273 (if (consp (car (cdr args
)))
274 ;; Rule-base composition.
275 (let ((len (length args
))
276 (tail (encode-composition-components args
'nocopy
)))
279 (setq str
(cons (car tail
) str
))
280 (setq tail
(nthcdr 2 tail
)))
281 (setq str
(concat (nreverse str
))
283 ;; Relative composition.
284 (setq str
(concat args
)))
285 (compose-string-internal str
0 (length str
) components
)))
287 (defun find-composition (pos &optional limit string detail-p
)
288 "Return information about a composition at or nearest to buffer position POS.
290 If the character at POS has `composition' property, the value is a list
291 of FROM, TO, and VALID-P.
293 FROM and TO specify the range of text that has the same `composition'
294 property, VALID-P is non-nil if and only if this composition is valid.
296 If there's no composition at POS, and the optional 2nd argument LIMIT
297 is non-nil, search for a composition toward LIMIT.
299 If no composition is found, return nil.
301 Optional 3rd argument STRING, if non-nil, is a string to look for a
302 composition in; nil means the current buffer.
304 If a valid composition is found and the optional 4th argument DETAIL-P
305 is non-nil, the return value is a list of FROM, TO, COMPONENTS,
306 RELATIVE-P, MOD-FUNC, and WIDTH.
308 COMPONENTS is a vector of integers, the meaning depends on RELATIVE-P.
310 RELATIVE-P is t if the composition method is relative, else nil.
312 If RELATIVE-P is t, COMPONENTS is a vector of characters to be
313 composed. If RELATIVE-P is nil, COMPONENTS is a vector of characters
314 and composition rules as described in `compose-region'.
316 MOD-FUNC is a modification function of the composition.
318 WIDTH is a number of columns the composition occupies on the screen."
319 (let ((result (find-composition-internal pos limit string detail-p
)))
320 (if (and detail-p result
(nth 2 result
) (not (nth 3 result
)))
321 ;; This is a valid rule-base composition.
322 (decode-composition-components (nth 2 result
) 'nocopy
))
326 (defun compose-chars-after (pos &optional limit object
)
327 "Compose characters in current buffer after position POS.
329 It looks up the char-table `composition-function-table' (which see) by
330 a character after POS. If non-nil value is found, the format of the
331 value should be an alist of PATTERNs vs FUNCs, where PATTERNs are
332 regular expressions and FUNCs are functions. If the text after POS
333 matches one of PATTERNs, call the corresponding FUNC with three
334 arguments POS, TO, and PATTERN, where TO is the end position of text
335 matching PATTERN, and return what FUNC returns. Otherwise, return
338 FUNC is responsible for composing the text properly. The return value
340 nil -- if no characters were composed.
341 CHARS (integer) -- if CHARS characters were composed.
343 Optional 2nd arg LIMIT, if non-nil, limits the matching of text.
345 Optional 3rd arg OBJECT, if non-nil, is a string that contains the
346 text to compose. In that case, POS and LIMIT index into the string.
348 This function is the default value of `compose-chars-after-function'."
349 (let ((tail (aref composition-function-table
(char-after pos
)))
354 (while (and tail
(not func
))
355 (setq pattern
(car (car tail
))
356 func
(cdr (car tail
)))
359 (and (re-search-forward pattern limit t
)
360 (= (match-beginning 0) pos
))
361 (looking-at pattern
))
362 (setq result
(funcall func pos
(match-end 0) pattern nil
))
363 (setq func nil tail
(cdr tail
)))))))
366 (defun compose-last-chars (args)
367 "Compose last characters.
368 The argument is a parameterized event of the form
369 \(compose-last-chars N COMPONENTS),
370 where N is the number of characters before point to compose,
371 COMPONENTS, if non-nil, is the same as the argument to `compose-region'
372 \(which see). If it is nil, `compose-chars-after' is called,
373 and that function finds a proper rule to compose the target characters.
374 This function is intended to be used from input methods.
375 The global keymap binds special event `compose-last-chars' to this
376 function. Input method may generate an event (compose-last-chars N COMPONENTS)
377 after a sequence of character events."
379 (let ((chars (nth 1 args
)))
380 (if (and (numberp chars
)
381 (>= (- (point) (point-min)) chars
))
383 (compose-region (- (point) chars
) (point) (nth 2 args
))
384 (compose-chars-after (- (point) chars
) (point))))))
386 (global-set-key [compose-last-chars
] 'compose-last-chars
)
389 ;;; Automatic character composition.
391 (defvar composition-function-table
392 (make-char-table nil
)
393 "Char table of functions for automatic character composition.
394 For each character that has to be composed automatically with
395 preceding and/or following characters, this char table contains
396 a function to call to compose that character.
398 An element, if non-nil, is FUNC or an alist of PATTERNs vs FUNCs,
399 where PATTERNs are regular expressions and FUNCs are functions.
400 If the element is FUNC, FUNC itself determines the region to
403 Each function is called with 4 arguments, FROM, TO, FONT-OBJECT,
406 If STRING is nil, FROM and TO are positions specifying the region
407 maching with PATTERN in the current buffer, and the function has
408 to compose character in that region (possibly with characters
409 preceding FROM). The return value of the function is the end
410 position where characters are composed.
412 Otherwise, STRING is a string, and FROM and TO are indices into
413 the string. In this case, the function has to compose a
414 character in the string.
416 FONT-OBJECT may be nil if not available (e.g. for the case of
419 See also the command `toggle-auto-composition'.")
421 ;; Copied from font-lock.el.
423 ;; Borrowed from lazy-lock.el.
424 ;; We use this to preserve or protect things when modifying text properties.
425 (defmacro save-buffer-state
(varlist &rest body
)
426 "Bind variables according to VARLIST and eval BODY restoring buffer state."
427 `(let* ,(append varlist
428 '((modified (buffer-modified-p)) (buffer-undo-list t
)
429 (inhibit-read-only t
) (inhibit-point-motion-hooks t
)
430 (inhibit-modification-hooks t
)
431 deactivate-mark buffer-file-name buffer-file-truename
))
434 (restore-buffer-modified-p nil
))))
435 ;; Fixme: This makes bootstrapping fail with this error.
436 ;; Symbol's function definition is void: eval-defun
437 ;;(def-edebug-spec save-buffer-state let)
440 (put 'save-buffer-state
'lisp-indent-function
1)
442 (defun terminal-composition-function (from to font-object string
)
443 "General composition function used on terminal.
444 Non-spacing characters are composed with the preceding spacing
445 character. All non-spacing characters has this function in
446 `terminal-composition-function-table'."
447 (let ((pos (1+ from
)))
450 (while (and (< pos to
)
451 (= (aref char-width-table
(aref string pos
)) 0))
454 (compose-string string
(1- from
) pos
)
455 (compose-string string from pos
456 (concat " " (buffer-substring from pos
)))))
457 (while (and (< pos to
)
458 (= (aref char-width-table
(char-after pos
)) 0))
460 (if (> from
(point-min))
461 (compose-region (1- from
) pos
(buffer-substring from pos
))
462 (compose-region from pos
463 (concat " " (buffer-substring from pos
)))))
466 (defvar terminal-composition-function-table
467 (let ((table (make-char-table nil
)))
470 (if (= val
0) (set-char-table-range table key
471 'terminal-composition-function
)))
474 "Char table of functions for automatic character composition on terminal.
475 This is like `composition-function-table' but used when Emacs is running
478 (defun auto-compose-chars (from to window string
)
479 "Compose characters in the region between FROM and TO.
480 WINDOW is a window displaying the current buffer.
481 If STRING is non-nil, it is a string, and FROM and TO are indices
482 into the string. In that case, compose characters in the string.
484 This function is the default value of `auto-composition-function' (which see)."
485 (save-buffer-state nil
489 (let ((table (if (display-graphic-p)
490 composition-function-table
491 terminal-composition-function-table
))
493 (setq to
(or (text-property-any (1+ from
) to
'auto-composed t
498 (let* ((ch (aref string from
))
499 (elt (aref table ch
))
503 (setq font-obj
(font-at from window string
)))
505 (setq newpos
(funcall elt from to font-obj string
))
507 (or (not (eq (string-match (caar elt
) string
511 (funcall (cdar elt
) from
514 (setq elt
(cdr elt
)))))
515 (if (and newpos
(> newpos from
))
517 (setq from
(1+ from
)))))
518 (narrow-to-region from to
)
520 (let* ((ch (char-after from
))
521 (elt (aref table ch
))
522 func pattern font-obj newpos
)
525 (setq font-obj
(font-at from window
)))
527 (setq newpos
(funcall elt from to font-obj nil
))
530 (or (not (looking-at (caar elt
)))
532 (funcall (cdar elt
) from
535 (setq elt
(cdr elt
)))))
536 (if (and newpos
(> newpos from
))
538 (setq from
(1+ from
))))))
539 (put-text-property start to
'auto-composed t string
)))))))
541 (make-variable-buffer-local 'auto-composition-function
)
544 (define-minor-mode auto-composition-mode
545 "Toggle Auto Compostion mode.
546 With arg, turn Auto Compostion mode off if and only if arg is a non-positive
547 number; if arg is nil, toggle Auto Compostion mode; anything else turns Auto
550 When Auto Composition is enabled, text characters are automatically composed
551 by functions registered in `composition-function-table' (which see).
553 You can use Global Auto Composition mode to automagically turn on
554 Auto Composition mode in all buffers (this is the default)."
557 (setq auto-composition-mode nil
))
558 (cond (auto-composition-mode
559 (add-hook 'after-change-functions
'auto-composition-after-change nil t
)
560 (setq auto-composition-function
'auto-compose-chars
))
562 (remove-hook 'after-change-functions
'auto-composition-after-change t
)
563 (setq auto-composition-function nil
)))
564 (save-buffer-state nil
567 (remove-text-properties (point-min) (point-max) '(auto-composed nil
))
568 (decompose-region (point-min) (point-max)))))
570 (defun auto-composition-after-change (start end old-len
)
571 (save-buffer-state nil
572 (if (< start
(point-min))
573 (setq start
(point-min)))
574 (if (> end
(point-max))
575 (setq end
(point-max)))
576 (when (and auto-composition-mode
(not memory-full
))
578 (when (and (> start
(point-min))
579 (setq func2
(aref composition-function-table
580 (char-after (1- start
))))
581 (or (= start
(point-max))
582 (not (setq func1
(aref composition-function-table
583 (char-after start
))))
585 (setq start
(1- start
)
587 (while (eq func1 func2
)
588 (if (> start
(point-min))
589 (setq start
(1- start
)
590 func2
(aref composition-function-table
593 (when (and (< end
(point-max))
594 (setq func2
(aref composition-function-table
596 (or (= end
(point-min))
597 (not (setq func1
(aref composition-function-table
598 (char-after (1- end
)))))
602 (while (eq func1 func2
)
603 (if (< end
(point-max))
604 (setq func2
(aref composition-function-table
609 (remove-text-properties start end
'(auto-composed nil
)))))))
611 (defun turn-on-auto-composition-if-enabled ()
612 (if enable-multibyte-characters
613 (auto-composition-mode 1)))
616 (define-global-minor-mode global-auto-composition-mode
617 auto-composition-mode turn-on-auto-composition-if-enabled
619 :initialize
'custom-initialize-safe-default
620 :init-value
(not noninteractive
)
621 :group
'auto-composition
624 (defun toggle-auto-composition (&optional arg
)
625 "Change whether automatic character composition is enabled in this buffer.
626 With arg, enable it iff arg is positive."
628 (let ((enable (if (null arg
) (not auto-composition-function
)
629 (> (prefix-numeric-value arg
) 0))))
631 (kill-local-variable 'auto-composition-function
)
632 (make-local-variable 'auto-composition-function
)
633 (setq auto-composition-function nil
)
634 (save-buffer-state nil
637 (decompose-region (point-min) (point-max)))))
639 (save-buffer-state nil
642 (remove-text-properties (point-min) (point-max)
643 '(auto-composed nil
))))))
645 (defun auto-compose-region (from to
)
646 "Force automatic character composition on the region FROM and TO."
648 (if (get-text-property from
'auto-composed
)
649 (setq from
(next-single-property-change from
'auto-composed nil to
)))
651 (let ((modified-p (buffer-modified-p))
652 (inhibit-read-only '(composition auto-composed
))
653 (stop (next-single-property-change (point) 'auto-composed nil to
)))
654 (while (< (point) to
)
657 (goto-char (next-single-property-change (point)
658 'auto-composed nil to
))
659 (setq stop
(next-single-property-change (point)
660 'auto-composed nil to
)))
661 (let ((func (aref composition-function-table
(following-char)))
664 (goto-char (funcall func
(point) nil
)))
667 (put-text-property from to
'auto-composed t
)
668 (set-buffer-modified-p modified-p
))))
671 ;; The following codes are only for backward compatibility with Emacs
674 (defun decompose-composite-char (char &optional type with-composition-rule
)
675 "Convert CHAR to string.
677 If optional 2nd arg TYPE is non-nil, it is `string', `list', or
678 `vector'. In this case, CHAR is converted to string, list of CHAR, or
679 vector of CHAR respectively.
680 Optional 3rd arg WITH-COMPOSITION-RULE is ignored."
681 (cond ((or (null type
) (eq type
'string
)) (char-to-string char
))
682 ((eq type
'list
) (list char
))
685 (make-obsolete 'decompose-composite-char
'char-to-string
"21.1")
689 ;; arch-tag: ee703d77-1723-45d4-a31f-e9f0f867aa33
690 ;;; composite.el ends here