aboutsummaryrefslogtreecommitdiffstats
diff options
context:
space:
mode:
authorDave Love2001-12-07 14:47:37 +0000
committerDave Love2001-12-07 14:47:37 +0000
commit269a5dd0adc17737ef2589084dc6ad1bb98de36a (patch)
tree17f95f1ec901030d1eafd708cef056c65d589605
parentaa2e3f49f337b3a14e208d1f2ddb34912730c535 (diff)
downloademacs-269a5dd0adc17737ef2589084dc6ad1bb98de36a.tar.gz
emacs-269a5dd0adc17737ef2589084dc6ad1bb98de36a.zip
Don't set word syntax (the default)
explicitly. Add a diacritic category. Add info for Unicode equivalents of characters in various Mule charsets and for extra Unicode characters. Don't define specific categories for Indian/Devanagari, since they aren't used.
Diffstat
-rw-r--r--lisp/international/characters.el554
1 files changed, 457 insertions, 97 deletions
diff --git a/lisp/international/characters.el b/lisp/international/characters.el
index f578d7ff264..cf3c9f92041 100644
--- a/lisp/international/characters.el
+++ b/lisp/international/characters.el
@@ -2,6 +2,7 @@
2 2
3;; Copyright (C) 1995, 1997 Electrotechnical Laboratory, JAPAN. 3;; Copyright (C) 1995, 1997 Electrotechnical Laboratory, JAPAN.
4;; Licensed to the Free Software Foundation. 4;; Licensed to the Free Software Foundation.
5;; Copyright (C) 2001 Free Software Foundation, Inc.
5 6
6;; Keywords: multibyte character, character set, syntax, category 7;; Keywords: multibyte character, character set, syntax, category
7 8
@@ -90,6 +91,8 @@
90(define-category ?> "A character which can't be placed at beginning of line.") 91(define-category ?> "A character which can't be placed at beginning of line.")
91(define-category ?< "A character which can't be placed at end of line.") 92(define-category ?< "A character which can't be placed at end of line.")
92 93
94;; Combining
95(define-category ?^ "Combining diacritic or mark")
93 96
94;;; Setting syntax and category. 97;;; Setting syntax and category.
95 98
@@ -108,13 +111,25 @@
108 arabic-1-column 111 arabic-1-column
109 arabic-2-column))) 112 arabic-2-column)))
110 (while charsets 113 (while charsets
111 (modify-syntax-entry (make-char (car charsets)) "w") 114;; (modify-syntax-entry (make-char (car charsets)) "w")
112 (modify-category-entry (make-char (car charsets)) ?b) 115 (modify-category-entry (make-char (car charsets)) ?b)
113 (setq charsets (cdr charsets)))) 116 (setq charsets (cdr charsets))))
117(let ((ch #x600))
118 (while (<= ch #x6ff)
119 (modify-category-entry (decode-char 'ucs ch) ?b)
120 (setq ch (1+ ch)))
121 (setq ch #xfb50)
122 (while (<= ch #xfdff)
123 (modify-category-entry (decode-char 'ucs ch) ?b)
124 (setq ch (1+ ch)))
125 (setq ch #xfe70)
126 (while (<= ch #xfefe)
127 (modify-category-entry (decode-char 'ucs ch) ?b)
128 (setq ch (1+ ch))))
114 129
115;; Chinese character set (GB2312) 130;; Chinese character set (GB2312)
116 131
117(modify-syntax-entry (make-char 'chinese-gb2312) "w") 132;; (modify-syntax-entry (make-char 'chinese-gb2312) "w")
118(modify-syntax-entry (make-char 'chinese-gb2312 33) "_") 133(modify-syntax-entry (make-char 'chinese-gb2312 33) "_")
119(modify-syntax-entry (make-char 'chinese-gb2312 34) "_") 134(modify-syntax-entry (make-char 'chinese-gb2312 34) "_")
120(modify-syntax-entry (make-char 'chinese-gb2312 41) "_") 135(modify-syntax-entry (make-char 'chinese-gb2312 41) "_")
@@ -132,6 +147,21 @@
132(modify-syntax-entry ?\$A!;(B ")$A!:(B") 147(modify-syntax-entry ?\$A!;(B ")$A!:(B")
133(modify-syntax-entry ?\$A!=(B ")$A!<(B") 148(modify-syntax-entry ?\$A!=(B ")$A!<(B")
134(modify-syntax-entry ?\$A!?(B ")$A!>(B") 149(modify-syntax-entry ?\$A!?(B ")$A!>(B")
150;; Unicode equivalents of above
151(modify-syntax-entry ?\$,2=T(B "($,2=U(B")
152(modify-syntax-entry ?\$,2=H(B "($,2=I(B")
153(modify-syntax-entry ?\$,2=J(B "($,2=K(B")
154(modify-syntax-entry ?\$,2=L(B "($,2=M(B")
155(modify-syntax-entry ?\$,2=N(B "($,2=O(B")
156(modify-syntax-entry ?\$,2=V(B "($,2=W(B")
157(modify-syntax-entry ?\$,2=P(B "($,2=Q(B")
158(modify-syntax-entry ?\$,2=U(B ")$,2=T(B")
159(modify-syntax-entry ?\$,2=I(B ")$,2=H(B")
160(modify-syntax-entry ?\$,2=K(B ")$,2=J(B")
161(modify-syntax-entry ?\$,2=M(B ")$,2=L(B")
162(modify-syntax-entry ?\$,2=O(B ")$,2=N(B")
163(modify-syntax-entry ?\$,2=W(B ")$,2=V(B")
164(modify-syntax-entry ?\$,2=Q(B ")$,2=P(B")
135 165
136(modify-category-entry (make-char 'chinese-gb2312) ?c) 166(modify-category-entry (make-char 'chinese-gb2312) ?c)
137(modify-category-entry (make-char 'chinese-gb2312) ?\|) 167(modify-category-entry (make-char 'chinese-gb2312) ?\|)
@@ -149,8 +179,8 @@
149 179
150(let ((generic-big5-1-char (make-char 'chinese-big5-1)) 180(let ((generic-big5-1-char (make-char 'chinese-big5-1))
151 (generic-big5-2-char (make-char 'chinese-big5-2))) 181 (generic-big5-2-char (make-char 'chinese-big5-2)))
152 (modify-syntax-entry generic-big5-1-char "w") 182;; (modify-syntax-entry generic-big5-1-char "w")
153 (modify-syntax-entry generic-big5-2-char "w") 183;; (modify-syntax-entry generic-big5-2-char "w")
154 184
155 (modify-category-entry generic-big5-1-char ?c) 185 (modify-category-entry generic-big5-1-char ?c)
156 (modify-category-entry generic-big5-2-char ?c) 186 (modify-category-entry generic-big5-2-char ?c)
@@ -174,7 +204,7 @@
174 generic-char) 204 generic-char)
175 (while cns-list 205 (while cns-list
176 (setq generic-char (make-char (car cns-list))) 206 (setq generic-char (make-char (car cns-list)))
177 (modify-syntax-entry generic-char "w") 207;; (modify-syntax-entry generic-char "w")
178 (modify-category-entry generic-char ?c) 208 (modify-category-entry generic-char ?c)
179 (modify-category-entry generic-char ?C) 209 (modify-category-entry generic-char ?C)
180 (modify-category-entry generic-char ?|) 210 (modify-category-entry generic-char ?|)
@@ -234,44 +264,103 @@
234 (set-case-syntax-pair ?,LL(B ?,Ll(B tbl) 264 (set-case-syntax-pair ?,LL(B ?,Ll(B tbl)
235 (set-case-syntax-pair ?,LM(B ?,Lm(B tbl) 265 (set-case-syntax-pair ?,LM(B ?,Lm(B tbl)
236 (set-case-syntax-pair ?,LN(B ?,Ln(B tbl) 266 (set-case-syntax-pair ?,LN(B ?,Ln(B tbl)
237 (set-case-syntax-pair ?,LO(B ?,Lo(B tbl)) 267 (set-case-syntax-pair ?,LO(B ?,Lo(B tbl)
268 (set-case-syntax-pair ?$,1(!(B ?$,1(q(B tbl)
269 (set-case-syntax-pair ?$,1("(B ?$,1(r(B tbl)
270 (set-case-syntax-pair ?$,1(#(B ?$,1(s(B tbl)
271 (set-case-syntax-pair ?$,1($(B ?$,1(t(B tbl)
272 (set-case-syntax-pair ?$,1(%(B ?$,1(u(B tbl)
273 (set-case-syntax-pair ?$,1(&(B ?$,1(v(B tbl)
274 (set-case-syntax-pair ?$,1('(B ?$,1(w(B tbl)
275 (set-case-syntax-pair ?$,1(((B ?$,1(x(B tbl)
276 (set-case-syntax-pair ?$,1()(B ?$,1(y(B tbl)
277 (set-case-syntax-pair ?$,1(*(B ?$,1(z(B tbl)
278 (set-case-syntax-pair ?$,1(+(B ?$,1({(B tbl)
279 (set-case-syntax-pair ?$,1(,(B ?$,1(|(B tbl)
280 (set-case-syntax-pair ?$,1(.(B ?$,1(~(B tbl)
281 (set-case-syntax-pair ?$,1(/(B ?$,1((B tbl)
282 (set-case-syntax-pair ?$,1(0(B ?$,1(P(B tbl)
283 (set-case-syntax-pair ?$,1(1(B ?$,1(Q(B tbl)
284 (set-case-syntax-pair ?$,1(2(B ?$,1(R(B tbl)
285 (set-case-syntax-pair ?$,1(3(B ?$,1(S(B tbl)
286 (set-case-syntax-pair ?$,1(4(B ?$,1(T(B tbl)
287 (set-case-syntax-pair ?$,1(5(B ?$,1(U(B tbl)
288 (set-case-syntax-pair ?$,1(6(B ?$,1(V(B tbl)
289 (set-case-syntax-pair ?$,1(7(B ?$,1(W(B tbl)
290 (set-case-syntax-pair ?$,1(8(B ?$,1(X(B tbl)
291 (set-case-syntax-pair ?$,1(9(B ?$,1(Y(B tbl)
292 (set-case-syntax-pair ?$,1(:(B ?$,1(Z(B tbl)
293 (set-case-syntax-pair ?$,1(;(B ?$,1([(B tbl)
294 (set-case-syntax-pair ?$,1(<(B ?$,1(\(B tbl)
295 (set-case-syntax-pair ?$,1(=(B ?$,1(](B tbl)
296 (set-case-syntax-pair ?$,1(>(B ?$,1(^(B tbl)
297 (set-case-syntax-pair ?$,1(?(B ?$,1(_(B tbl)
298 (set-case-syntax-pair ?$,1(@(B ?$,1(`(B tbl)
299 (set-case-syntax-pair ?$,1(A(B ?$,1(a(B tbl)
300 (set-case-syntax-pair ?$,1(B(B ?$,1(b(B tbl)
301 (set-case-syntax-pair ?$,1(C(B ?$,1(c(B tbl)
302 (set-case-syntax-pair ?$,1(D(B ?$,1(d(B tbl)
303 (set-case-syntax-pair ?$,1(E(B ?$,1(e(B tbl)
304 (set-case-syntax-pair ?$,1(F(B ?$,1(f(B tbl)
305 (set-case-syntax-pair ?$,1(G(B ?$,1(g(B tbl)
306 (set-case-syntax-pair ?$,1(H(B ?$,1(h(B tbl)
307 (set-case-syntax-pair ?$,1(I(B ?$,1(i(B tbl)
308 (set-case-syntax-pair ?$,1(J(B ?$,1(j(B tbl)
309 (set-case-syntax-pair ?$,1(K(B ?$,1(k(B tbl)
310 (set-case-syntax-pair ?$,1(L(B ?$,1(l(B tbl)
311 (set-case-syntax-pair ?$,1(M(B ?$,1(m(B tbl)
312 (set-case-syntax-pair ?$,1(N(B ?$,1(n(B tbl)
313 (set-case-syntax-pair ?$,1(O(B ?$,1(o(B tbl))
238 314
239;; Devanagari character set 315;; Devanagari character set
240 316
241(let ((deflist '(;; chars syntax category 317;;; Commented out since the categories appear not to be used anywhere
242 ("$(5!!!"!#(B" "w" ?7) ; vowel-modifying diacritical mark 318;;; and word syntax is the default.
243 ; chandrabindu, anuswar, visarga 319;; (let ((deflist '(;; chars syntax category
244 ("$(5!$(B-$(5!2(B" "w" ?1) ; independent vowel 320;; ("$(5!!!"!#(B" "w" ?7) ; vowel-modifying diacritical mark
245 ("$(5!3(B-$(5!X(B" "w" ?0) ; consonant 321;; ; chandrabindu, anuswar, visarga
246 ("$(5!Z(B-$(5!g(B" "w" ?8) ; matra 322;; ("$(5!$(B-$(5!2(B" "w" ?1) ; independent vowel
247 ("$(5!q(B-$(5!z(B" "w" ?6) ; digit 323;; ("$(5!3(B-$(5!X(B" "w" ?0) ; consonant
248 )) 324;; ("$(5!Z(B-$(5!g(B" "w" ?8) ; matra
249 elm chars len syntax category to ch i) 325;; ("$(5!q(B-$(5!z(B" "w" ?6) ; digit
250 (while deflist 326;; ;; Unicode equivalents
251 (setq elm (car deflist)) 327;; ("$,15A5B5C(B" "w" ?7) ; vowel-modifying diacritical mark
252 (setq chars (car elm) 328;; ; chandrabindu, anuswar, visarga
253 len (length chars) 329;; ("$,15E(B-$,15M(B" "w" ?1) ; independent vowel
254 syntax (nth 1 elm) 330;; ("$,15U(B-$,15y(B" "w" ?0) ; consonant
255 category (nth 2 elm) 331;; ("$,15~(B-$,16)(B" "w" ?8) ; matra
256 i 0) 332;; ("$,16F(B-$,16O(B" "w" ?6) ; digit
257 (while (< i len) 333;; ))
258 (if (= (aref chars i) ?-) 334;; elm chars len syntax category to ch i)
259 (setq i (1+ i) 335;; (while deflist
260 to (aref chars i)) 336;; (setq elm (car deflist))
261 (setq ch (aref chars i) 337;; (setq chars (car elm)
262 to ch)) 338;; len (length chars)
263 (while (<= ch to) 339;; syntax (nth 1 elm)
264 (modify-syntax-entry ch syntax) 340;; category (nth 2 elm)
265 (modify-category-entry ch category) 341;; i 0)
266 (setq ch (1+ ch))) 342;; (while (< i len)
267 (setq i (1+ i))) 343;; (if (= (aref chars i) ?-)
268 (setq deflist (cdr deflist)))) 344;; (setq i (1+ i)
345;; to (aref chars i))
346;; (setq ch (aref chars i)
347;; to ch))
348;; (while (<= ch to)
349;; (modify-syntax-entry ch syntax)
350;; (modify-category-entry ch category)
351;; (setq ch (1+ ch)))
352;; (setq i (1+ i)))
353;; (setq deflist (cdr deflist))))
269 354
270;; Ethiopic character set 355;; Ethiopic character set
271 356
272(modify-category-entry (make-char 'ethiopic) ?e) 357(modify-category-entry (make-char 'ethiopic) ?e)
273(modify-syntax-entry (make-char 'ethiopic) "w") 358;; (modify-syntax-entry (make-char 'ethiopic) "w")
274(let ((chars '(?$(3$h(B ?$(3$i(B ?$(3$j(B ?$(3$k(B ?$(3$l(B ?$(3$m(B ?$(3$n(B ?$(3$o(B ?$(3%i(B ?$(3%t(B ?$(3%u(B ?$(3%v(B ?$(3%w(B ?$(3%x(B))) 359(dotimes (i (1+ (- #x137c #x1200)))
360 (modify-category-entry (decode-char 'ucs (+ #x1200 i)) ?e))
361(let ((chars '(?$(3$h(B ?$(3$i(B ?$(3$j(B ?$(3$k(B ?$(3$l(B ?$(3$m(B ?$(3$n(B ?$(3$o(B ?$(3%i(B ?$(3%t(B ?$(3%u(B ?$(3%v(B ?$(3%w(B ?$(3%x(B
362 ;; Unicode equivalents of the above:
363 ?$,1Q!(B ?$,1Q"(B ?$,1Q#(B ?$,1Q$(B ?$,1Q%(B ?$,1Q&(B ?$,1Q'(B ?$,1Q((B ?$,3op(B ?$,3o{(B ?$,3o|(B ?$,3o}(B ?$,3o~(B ?$,3o(B)))
275 (while chars 364 (while chars
276 (modify-syntax-entry (car chars) ".") 365 (modify-syntax-entry (car chars) ".")
277 (setq chars (cdr chars)))) 366 (setq chars (cdr chars))))
@@ -279,16 +368,36 @@
279;; Greek character set (ISO-8859-7) 368;; Greek character set (ISO-8859-7)
280 369
281(modify-category-entry (make-char 'greek-iso8859-7) ?g) 370(modify-category-entry (make-char 'greek-iso8859-7) ?g)
282 371(let ((c #x370))
283(let ((c 182)) 372 (while (<= c #x3ff)
284 (while (< c 255) 373 (modify-category-entry (decode-char 'ucs c) ?g)
285 (modify-syntax-entry (make-char 'greek-iso8859-7 c) "w")
286 (setq c (1+ c)))) 374 (setq c (1+ c))))
287(modify-syntax-entry (make-char 'greek-iso8859-7 160) "w") ; NBSP 375
376;; (let ((c 182))
377;; (while (< c 255)
378;; (modify-syntax-entry (make-char 'greek-iso8859-7 c) "w")
379;; (setq c (1+ c))))
380;; (modify-syntax-entry (make-char 'greek-iso8859-7 160) "w") ; NBSP
288(modify-syntax-entry ?,F7(B ".") 381(modify-syntax-entry ?,F7(B ".")
289(modify-syntax-entry ?,F;(B ".") 382(modify-syntax-entry ?,F;(B ".")
290(modify-syntax-entry ?,F=(B ".") 383(modify-syntax-entry ?,F=(B ".")
291(let ((tbl (standard-case-table))) 384(let ((tbl (standard-case-table)))
385 ;; Fixme: non-letter syntax copied from latin-1, but that's dubious
386 ;; in several cases.
387 (set-case-syntax ?,F!(B "." tbl)
388 (set-case-syntax ?,F"(B "." tbl)
389 (set-case-syntax ?,F&(B "." tbl)
390 (set-case-syntax ?,F&(B "_" tbl)
391 (set-case-syntax ?,F'(B "." tbl)
392 (set-case-syntax ?,F)(B "_" tbl)
393 (set-case-syntax ?,F+(B "." tbl)
394 (set-case-syntax ?,F,(B "_" tbl)
395 (set-case-syntax ?,F-(B "_" tbl)
396 (set-case-syntax ?,F/(B "." tbl)
397 (set-case-syntax ?,F0(B "_" tbl)
398 (set-case-syntax ?,F1(B "_" tbl)
399;; (set-case-syntax ?,F7(B "_" tbl)
400;; (set-case-syntax ?,F=(B "_" tbl)
292 (set-case-syntax-pair ?,FA(B ?,Fa(B tbl) 401 (set-case-syntax-pair ?,FA(B ?,Fa(B tbl)
293 (set-case-syntax-pair ?,FB(B ?,Fb(B tbl) 402 (set-case-syntax-pair ?,FB(B ?,Fb(B tbl)
294 (set-case-syntax-pair ?,FC(B ?,Fc(B tbl) 403 (set-case-syntax-pair ?,FC(B ?,Fc(B tbl)
@@ -321,53 +430,94 @@
321 (set-case-syntax-pair ?,F6(B ?,F\(B tbl) 430 (set-case-syntax-pair ?,F6(B ?,F\(B tbl)
322 (set-case-syntax-pair ?,F8(B ?,F](B tbl) 431 (set-case-syntax-pair ?,F8(B ?,F](B tbl)
323 (set-case-syntax-pair ?,F9(B ?,F^(B tbl) 432 (set-case-syntax-pair ?,F9(B ?,F^(B tbl)
324 (set-case-syntax-pair ?,F:(B ?,F_(B tbl)) 433 (set-case-syntax-pair ?,F:(B ?,F_(B tbl)
434 ;; Unicode equivalents
435 (set-case-syntax-pair ?$,1&q(B ?$,1'1(B tbl)
436 (set-case-syntax-pair ?$,1&r(B ?$,1'2(B tbl)
437 (set-case-syntax-pair ?$,1&s(B ?$,1'3(B tbl)
438 (set-case-syntax-pair ?$,1&t(B ?$,1'4(B tbl)
439 (set-case-syntax-pair ?$,1&u(B ?$,1'5(B tbl)
440 (set-case-syntax-pair ?$,1&v(B ?$,1'6(B tbl)
441 (set-case-syntax-pair ?$,1&w(B ?$,1'7(B tbl)
442 (set-case-syntax-pair ?$,1&x(B ?$,1'8(B tbl)
443 (set-case-syntax-pair ?$,1&y(B ?$,1'9(B tbl)
444 (set-case-syntax-pair ?$,1&z(B ?$,1':(B tbl)
445 (set-case-syntax-pair ?$,1&{(B ?$,1';(B tbl)
446 (set-case-syntax-pair ?$,1&|(B ?$,1'<(B tbl)
447 (set-case-syntax-pair ?$,1&}(B ?$,1'=(B tbl)
448 (set-case-syntax-pair ?$,1&~(B ?$,1'>(B tbl)
449 (set-case-syntax-pair ?$,1&(B ?$,1'?(B tbl)
450 (set-case-syntax-pair ?$,1' (B ?$,1'@(B tbl)
451 (set-case-syntax-pair ?$,1'!(B ?$,1'A(B tbl)
452 (set-case-syntax-pair ?$,1'#(B ?$,1'C(B tbl)
453 (set-case-syntax-pair ?$,1'$(B ?$,1'D(B tbl)
454 (set-case-syntax-pair ?$,1'%(B ?$,1'E(B tbl)
455 (set-case-syntax-pair ?$,1'&(B ?$,1'F(B tbl)
456 (set-case-syntax-pair ?$,1''(B ?$,1'G(B tbl)
457 (set-case-syntax-pair ?$,1'((B ?$,1'H(B tbl)
458 (set-case-syntax-pair ?$,1')(B ?$,1'I(B tbl)
459 (set-case-syntax-pair ?$,1'*(B ?$,1'J(B tbl)
460 (set-case-syntax-pair ?$,1'+(B ?$,1'K(B tbl)
461 (set-case-syntax-pair ?$,1&o(B ?$,1'N(B tbl)
462 (set-case-syntax-pair ?$,1&n(B ?$,1'M(B tbl)
463 (set-case-syntax-pair ?$,1&l(B ?$,1'L(B tbl)
464 (set-case-syntax-pair ?$,1&f(B ?$,1',(B tbl)
465 (set-case-syntax-pair ?$,1&h(B ?$,1'-(B tbl)
466 (set-case-syntax-pair ?$,1&i(B ?$,1'.(B tbl)
467 (set-case-syntax-pair ?$,1&j(B ?$,1'/(B tbl))
325 468
326;; Hebrew character set (ISO-8859-8) 469;; Hebrew character set (ISO-8859-8)
327 470
328(modify-category-entry (make-char 'hebrew-iso8859-8) ?w) 471;; (modify-category-entry (make-char 'hebrew-iso8859-8) ?w)
329 472
330(let ((c 224)) 473;; (let ((c 224))
331 (while (< c 251) 474;; (while (< c 251)
332 (modify-syntax-entry (make-char 'hebrew-iso8859-8 c) "w") 475;; (modify-syntax-entry (make-char 'hebrew-iso8859-8 c) "w")
333 (setq c (1+ c)))) 476;; (setq c (1+ c))))
334(modify-syntax-entry (make-char 'hebrew-iso8859-8 160) "w") ; NBSP 477;; (modify-syntax-entry (make-char 'hebrew-iso8859-8 160) "w") ; NBSP
335 478
336;; Indian character set (IS 13194 and other Emacs original Indian charsets) 479;; Indian character set (IS 13194 and other Emacs original Indian charsets)
337 480
338(modify-category-entry (make-char 'indian-is13194) ?i) 481(modify-category-entry (make-char 'indian-is13194) ?i)
339(modify-category-entry (make-char 'indian-2-column) ?I) 482(modify-category-entry (make-char 'indian-2-column) ?I)
340(modify-category-entry (make-char 'indian-glyph) ?I) 483(modify-category-entry (make-char 'indian-glyph) ?I)
484;; Unicode Devanagari block
485(let ((c #x901))
486 (while (<= c #x970)
487 (modify-category-entry (decode-char 'ucs c) ?i)
488 (setq c (1+ c))))
341 489
342(let ((deflist 490;;; Commented out since the categories appear not to be used anywhere
343 '(;; chars syntax category 491;;; and word syntax is the default.
344 ("(5!"#(B" "w" ?7) ; vowel-modifying diacritical mark 492;; (let ((deflist ;
345 ; chandrabindu, anuswar, visarga 493;; '(;; chars syntax category
346 ("(5$(B-(52(B" "w" ?1) ; base (independent) vowel 494;; ("(5!"#(B" "w" ?7) ; vowel-modifying diacritical mark
347 ("(53(B-(5X(B" "w" ?0) ; consonant 495;; ; chandrabindu, anuswar, visarga
348 ("(5Z(B-(5g(B" "w" ?8) ; matra 496;; ("(5$(B-(52(B" "w" ?1) ; base (independent) vowel
349 ("(5q(B-(5z(B" "w" ?6) ; digit 497;; ("(53(B-(5X(B" "w" ?0) ; consonant
350 )) 498;; ("(5Z(B-(5g(B" "w" ?8) ; matra
351 elm chars len syntax category to ch i) 499;; ("(5q(B-(5z(B" "w" ?6) ; digit
352 (while deflist 500;; ))
353 (setq elm (car deflist)) 501;; elm chars len syntax category to ch i)
354 (setq chars (car elm) 502;; (while deflist
355 len (length chars) 503;; (setq elm (car deflist))
356 syntax (nth 1 elm) 504;; (setq chars (car elm)
357 category (nth 2 elm) 505;; len (length chars)
358 i 0) 506;; syntax (nth 1 elm)
359 (while (< i len) 507;; category (nth 2 elm)
360 (if (= (aref chars i) ?-) 508;; i 0)
361 (setq i (1+ i) 509;; (while (< i len)
362 to (aref chars i)) 510;; (if (= (aref chars i) ?-)
363 (setq ch (aref chars i) 511;; (setq i (1+ i)
364 to ch)) 512;; to (aref chars i))
365 (while (<= ch to) 513;; (setq ch (aref chars i)
366 (modify-syntax-entry ch syntax) 514;; to ch))
367 (modify-category-entry ch category) 515;; (while (<= ch to)
368 (setq ch (1+ ch))) 516;; (modify-syntax-entry ch syntax)
369 (setq i (1+ i))) 517;; (modify-category-entry ch category)
370 (setq deflist (cdr deflist)))) 518;; (setq ch (1+ ch)))
519;; (setq i (1+ i)))
520;; (setq deflist (cdr deflist))))
371 521
372 522
373;; Japanese character set (JISX0201-kana, JISX0201-roman, JISX0208, JISX0212) 523;; Japanese character set (JISX0201-kana, JISX0201-roman, JISX0208, JISX0212)
@@ -381,15 +531,41 @@
381(modify-category-entry (make-char 'japanese-jisx0208) ?\|) 531(modify-category-entry (make-char 'japanese-jisx0208) ?\|)
382(modify-category-entry (make-char 'japanese-jisx0212) ?\|) 532(modify-category-entry (make-char 'japanese-jisx0212) ?\|)
383 533
534;; Unicode equivalents of JISX0201-kana
535(let ((c #xff61))
536 (while (<= c #xff9f)
537 (modify-category-entry (decode-char 'ucs c) ?k)
538 (modify-category-entry (decode-char 'ucs c) ?j)
539 (modify-category-entry (decode-char 'ucs c) ?\|)
540 (setq c (1+ c))))
541
542;; Katakana block
543(let ((c #x30a0))
544 (while (<= c #x30ff)
545 ;; ?K is double width, ?k isn't specified
546 (modify-category-entry (decode-char 'ucs c) ?k)
547 (modify-category-entry (decode-char 'ucs c) ?j)
548 (modify-category-entry (decode-char 'ucs c) ?\|)
549 (setq c (1+ c))))
550
551;; Hiragana block
552(let ((c #x3040))
553 (while (<= c #x309f)
554 ;; ?H is actually defined to be double width
555 (modify-category-entry (decode-char 'ucs c) ?H)
556 ;;(modify-category-entry (decode-char 'ucs c) ?j)
557 (modify-category-entry (decode-char 'ucs c) ?\|)
558 (setq c (1+ c))))
559
384;; JISX0208 560;; JISX0208
385(modify-syntax-entry (make-char 'japanese-jisx0208) "w") 561;; (modify-syntax-entry (make-char 'japanese-jisx0208) "w")
386(modify-syntax-entry (make-char 'japanese-jisx0208 33) "_") 562(modify-syntax-entry (make-char 'japanese-jisx0208 33) "_")
387(modify-syntax-entry (make-char 'japanese-jisx0208 34) "_") 563(modify-syntax-entry (make-char 'japanese-jisx0208 34) "_")
388(modify-syntax-entry (make-char 'japanese-jisx0208 40) "_") 564(modify-syntax-entry (make-char 'japanese-jisx0208 40) "_")
389(let ((chars '(?$B!<(B ?$B!+(B ?$B!,(B ?$B!3(B ?$B!4(B ?$B!5(B ?$B!6(B ?$B!7(B ?$B!8(B ?$B!9(B ?$B!:(B ?$B!;(B))) 565;; (let ((chars '(?$B!<(B ?$B!+(B ?$B!,(B ?$B!3(B ?$B!4(B ?$B!5(B ?$B!6(B ?$B!7(B ?$B!8(B ?$B!9(B ?$B!:(B ?$B!;(B)))
390 (while chars 566;; (while chars
391 (modify-syntax-entry (car chars) "w") 567;; (modify-syntax-entry (car chars) "w")
392 (setq chars (cdr chars)))) 568;; (setq chars (cdr chars))))
393(modify-syntax-entry ?\$B!J(B "($B!K(B") 569(modify-syntax-entry ?\$B!J(B "($B!K(B")
394(modify-syntax-entry ?\$B!N(B "($B!O(B") 570(modify-syntax-entry ?\$B!N(B "($B!O(B")
395(modify-syntax-entry ?\$B!P(B "($B!Q(B") 571(modify-syntax-entry ?\$B!P(B "($B!Q(B")
@@ -422,7 +598,7 @@
422 (setq chars (cdr chars)))) 598 (setq chars (cdr chars))))
423 599
424;; JISX0212 600;; JISX0212
425(modify-syntax-entry (make-char 'japanese-jisx0212) "w") 601;; (modify-syntax-entry (make-char 'japanese-jisx0212) "w")
426(modify-syntax-entry (make-char 'japanese-jisx0212 33) "_") 602(modify-syntax-entry (make-char 'japanese-jisx0212 33) "_")
427(modify-syntax-entry (make-char 'japanese-jisx0212 34) "_") 603(modify-syntax-entry (make-char 'japanese-jisx0212 34) "_")
428(modify-syntax-entry (make-char 'japanese-jisx0212 35) "_") 604(modify-syntax-entry (make-char 'japanese-jisx0212 35) "_")
@@ -430,8 +606,10 @@
430(modify-category-entry (make-char 'japanese-jisx0212 ) ?C) 606(modify-category-entry (make-char 'japanese-jisx0212 ) ?C)
431 607
432;; JISX0201-Kana 608;; JISX0201-Kana
433(modify-syntax-entry (make-char 'katakana-jisx0201) "w") 609;; (modify-syntax-entry (make-char 'katakana-jisx0201) "w")
434(let ((chars '(?(I!(B ?(I$(B ?(I%(B))) 610(let ((chars '(?(I!(B ?(I$(B ?(I%(B
611 ;; Unicode:
612 ?$,3sa(B ?$,3sd(B ?$,3se(B)))
435 (while chars 613 (while chars
436 (modify-syntax-entry (car chars) ".") 614 (modify-syntax-entry (car chars) ".")
437 (setq chars (cdr chars)))) 615 (setq chars (cdr chars))))
@@ -441,7 +619,7 @@
441 619
442;; Korean character set (KSC5601) 620;; Korean character set (KSC5601)
443 621
444(modify-syntax-entry (make-char 'korean-ksc5601) "w") 622;; (modify-syntax-entry (make-char 'korean-ksc5601) "w")
445(modify-syntax-entry (make-char 'korean-ksc5601 33) "_") 623(modify-syntax-entry (make-char 'korean-ksc5601 33) "_")
446(modify-syntax-entry (make-char 'korean-ksc5601 34) "_") 624(modify-syntax-entry (make-char 'korean-ksc5601 34) "_")
447(modify-syntax-entry (make-char 'korean-ksc5601 38) "_") 625(modify-syntax-entry (make-char 'korean-ksc5601 38) "_")
@@ -477,6 +655,8 @@
477;; Lao character set 655;; Lao character set
478 656
479(modify-category-entry (make-char 'lao) ?o) 657(modify-category-entry (make-char 'lao) ?o)
658(dotimes (i (1+ (- #xeff #xe80)))
659 (modify-category-entry (decode-char 'ucs (+ i #xe80)) ?o))
480 660
481(let ((deflist '(;; chars syntax category 661(let ((deflist '(;; chars syntax category
482 ("(1!(B-(1N(B" "w" ?0) ; consonant 662 ("(1!(B-(1N(B" "w" ?0) ; consonant
@@ -487,6 +667,15 @@
487 ("(1\(B" "w" ?9) ; semivowel lower 667 ("(1\(B" "w" ?9) ; semivowel lower
488 ("(1p(B-(1y(B" "w" ?6) ; digit 668 ("(1p(B-(1y(B" "w" ?6) ; digit
489 ("(1Of(B" "_" ?5) ; symbol 669 ("(1Of(B" "_" ?5) ; symbol
670 ;; Unicode equivalents
671 ("$,1D!(B-$,1DN(B" "w" ?0) ; consonant
672 ("$,1DPDRDSD]D`(B-$,1Dd(B" "w" ?1) ; vowel base
673 ("$,1DQDT(B-$,1DWD[Dm(B" "w" ?2) ; vowel upper
674 ("$,1DXDY(B" "w" ?3) ; vowel lower
675 ("$,1Dh(B-$,1Dk(B" "w" ?4) ; tone mark
676 ("$,1D\D](B" "w" ?9) ; semivowel lower
677 ("$,1Dp(B-$,1Dy(B" "w" ?6) ; digit
678 ("$,1DODf(B" "_" ?5) ; symbol
490 )) 679 ))
491 elm chars len syntax category to ch i) 680 elm chars len syntax category to ch i)
492 (while deflist 681 (while deflist
@@ -503,7 +692,8 @@
503 (setq ch (aref chars i) 692 (setq ch (aref chars i)
504 to ch)) 693 to ch))
505 (while (<= ch to) 694 (while (<= ch to)
506 (modify-syntax-entry ch syntax) 695 (unless (string-equal syntax "w")
696 (modify-syntax-entry ch syntax))
507 (modify-category-entry ch category) 697 (modify-category-entry ch category)
508 (setq ch (1+ ch))) 698 (setq ch (1+ ch)))
509 (setq i (1+ i))) 699 (setq i (1+ i)))
@@ -512,6 +702,8 @@
512;; Thai character set (TIS620) 702;; Thai character set (TIS620)
513 703
514(modify-category-entry (make-char 'thai-tis620) ?t) 704(modify-category-entry (make-char 'thai-tis620) ?t)
705(dotimes (i (1+ (- #xe7f #xe00)))
706 (modify-category-entry (decode-char 'ucs (+ i #xe00)) ?t))
515 707
516(let ((deflist '(;; chars syntax category 708(let ((deflist '(;; chars syntax category
517 (",T!(B-,TCEG(B-,TN(B" "w" ?0) ; consonant 709 (",T!(B-,TCEG(B-,TN(B" "w" ?0) ; consonant
@@ -521,6 +713,14 @@
521 (",Th(B-,Tm(B" "w" ?4) ; tone mark 713 (",Th(B-,Tm(B" "w" ?4) ; tone mark
522 (",Tp(B-,Ty(B" "w" ?6) ; digit 714 (",Tp(B-,Ty(B" "w" ?6) ; digit
523 (",TOf_oz{(B" "_" ?5) ; symbol 715 (",TOf_oz{(B" "_" ?5) ; symbol
716 ;; Unicode equivalents
717 ("$,1Ba(B-$,1C#C%C'(B-$,1C.(B" "w" ?0) ; consonant
718 ("$,1C$C&C0C2C3C@(B-$,1CE(B" "w" ?1) ; vowel base
719 ("$,1C1C4(B-$,1C7CGCN(B" "w" ?2) ; vowel upper
720 ("$,1C8(B-$,1C:(B" "w" ?3) ; vowel lower
721 ("$,1CH(B-$,1CM(B" "w" ?4) ; tone mark
722 ("$,1CP(B-$,1CY(B" "w" ?6) ; digit
723 ("$,1C/CFC?COCZC[(B" "_" ?5) ; symbol
524 )) 724 ))
525 elm chars len syntax category to ch i) 725 elm chars len syntax category to ch i)
526 (while deflist 726 (while deflist
@@ -537,7 +737,8 @@
537 (setq ch (aref chars i) 737 (setq ch (aref chars i)
538 to ch)) 738 to ch))
539 (while (<= ch to) 739 (while (<= ch to)
540 (modify-syntax-entry ch syntax) 740 (unless (string-equal syntax "w")
741 (modify-syntax-entry ch syntax))
541 (modify-category-entry ch category) 742 (modify-category-entry ch category)
542 (setq ch (1+ ch))) 743 (setq ch (1+ ch)))
543 (setq i (1+ i))) 744 (setq i (1+ i)))
@@ -547,9 +748,11 @@
547 748
548(modify-category-entry (make-char 'tibetan) ?q) 749(modify-category-entry (make-char 'tibetan) ?q)
549(modify-category-entry (make-char 'tibetan-1-column) ?q) 750(modify-category-entry (make-char 'tibetan-1-column) ?q)
751(dotimes (i (1+ (- #xfff #xf00)))
752 (modify-category-entry (decode-char 'ucs (+ i #xf00)) ?q))
550 753
551(let ((deflist '(;; chars syntax category 754(let ((deflist '(;; chars syntax category
552 ("$(7"!(B-$(7"J"K(B" "w" ?0) ; consonant 755 ("4$(7"!0"!1(B-4$(7"J0"J14"K0"K1(B" "w" ?0) ; consonant
553 ("$(7#!(B-$(7#J#K#L#M!"!#(B" "w" ?0) ; 756 ("$(7#!(B-$(7#J#K#L#M!"!#(B" "w" ?0) ;
554 ("$(7$!(B-$(7$e(B" "w" ?0) ; 757 ("$(7$!(B-$(7$e(B" "w" ?0) ;
555 ("$(7%!(B-$(7%u(B" "w" ?0) ; 758 ("$(7%!(B-$(7%u(B" "w" ?0) ;
@@ -563,6 +766,18 @@
563 ("$(8!;!=!?!@!A!D"`(B" "." ?>) ; 766 ("$(8!;!=!?!@!A!D"`(B" "." ?>) ;
564 ("$(7!0(B-$(7!:!l#R#S"f(B" "." ?<) ; prohibition 767 ("$(7!0(B-$(7!:!l#R#S"f(B" "." ?<) ; prohibition
565 ("$(7!C!E(B-$(7!H!J(B-$(7!O!f!h(B-$(7!k!n!o#O#P(B-$(7#`(B" "." ?q) ; others 768 ("$(7!C!E(B-$(7!H!J(B-$(7!O!f!h(B-$(7!k!n!o#O#P(B-$(7#`(B" "." ?q) ; others
769
770 ;; Unicode version (not complete)
771 ("$,1F (B-$,1FIFJ(B" "w" ?0) ; consonant
772 ("$,1Fp(B-$,1G9G:G;G<(B" "w" ?0) ;
773 ("$,1FRFZF[F\F]F`(B" "w" ?2) ; upper vowel
774 ("$,1F^FbFcFfFgFhFiFjFk(B" "w" ?2) ; upper modifier
775 ("$,1EYFPFQFTFdEuEw(B" "w" ?3) ; lowel vowel/modifier
776 ("$,1E`(B-$,1EiEj(B-$,1Es(B" "w" ?6) ; digit
777 ("$,1EKEM(B-$,1ERETF_(B" "." ?|) ; line-break char
778 ("$,1EHEKEM(B-$,1ERETF_E}Et(B" "." ?>) ; prohibition
779 ("$,1E@(B-$,1EJE|GAGBFe(B" "." ?<) ; prohibition
780 ("$,1ESEU(B-$,1EXEZ(B-$,1E_EvEx(B-$,1E{E~EG>G?(B-$,1GO(B" "." ?q) ; others
566 )) 781 ))
567 elm chars len syntax category to ch i) 782 elm chars len syntax category to ch i)
568 (while deflist 783 (while deflist
@@ -579,7 +794,8 @@
579 (setq ch (aref chars i) 794 (setq ch (aref chars i)
580 to ch)) 795 to ch))
581 (while (<= ch to) 796 (while (<= ch to)
582 (modify-syntax-entry ch syntax) 797 (unless (string-equal syntax "w")
798 (modify-syntax-entry ch syntax))
583 (modify-category-entry ch category) 799 (modify-category-entry ch category)
584 (setq ch (1+ ch))) 800 (setq ch (1+ ch)))
585 (setq i (1+ i))) 801 (setq i (1+ i)))
@@ -589,8 +805,8 @@
589 805
590(let ((lower (make-char 'vietnamese-viscii-lower)) 806(let ((lower (make-char 'vietnamese-viscii-lower))
591 (upper (make-char 'vietnamese-viscii-upper))) 807 (upper (make-char 'vietnamese-viscii-upper)))
592 (modify-syntax-entry lower "w") 808;; (modify-syntax-entry lower "w")
593 (modify-syntax-entry upper "w") 809;; (modify-syntax-entry upper "w")
594 (modify-category-entry lower ?v) 810 (modify-category-entry lower ?v)
595 (modify-category-entry upper ?v) 811 (modify-category-entry upper ?v)
596 (modify-category-entry lower ?l) ; To make a word with 812 (modify-category-entry lower ?l) ; To make a word with
@@ -615,8 +831,6 @@
615;; Thus we have to check language-environment to handle casing 831;; Thus we have to check language-environment to handle casing
616;; correctly. Currently only I<->i is available. 832;; correctly. Currently only I<->i is available.
617 833
618;; case-syntax-pair's are not yet given for Latin Extendet-B
619
620 ;; Latin Extended-A, Latin Extended-B 834 ;; Latin Extended-A, Latin Extended-B
621 (setq c #x0100) 835 (setq c #x0100)
622 (while (<= c #x0233) 836 (while (<= c #x0233)
@@ -640,6 +854,102 @@
640 (set-case-syntax-pair ?$,1!;(B ?$,1!<(B tbl) 854 (set-case-syntax-pair ?$,1!;(B ?$,1!<(B tbl)
641 (set-case-syntax-pair ?$,1!=(B ?$,1!>(B tbl) 855 (set-case-syntax-pair ?$,1!=(B ?$,1!>(B tbl)
642 856
857 ;; Latin Extended-B
858 (set-case-syntax-pair ?$,1!A(B ?$,1#S(B tbl)
859 (set-case-syntax-pair ?$,1!B(B ?$,1!C(B tbl)
860 (set-case-syntax-pair ?$,1!D(B ?$,1!E(B tbl)
861 (set-case-syntax-pair ?$,1!F(B ?$,1#T(B tbl)
862 (set-case-syntax-pair ?$,1!G(B ?$,1!H(B tbl)
863 (set-case-syntax-pair ?$,1!I(B ?$,1#V(B tbl)
864 (set-case-syntax-pair ?$,1!J(B ?$,1#W(B tbl)
865 (set-case-syntax-pair ?$,1!K(B ?$,1!L(B tbl)
866 (set-case-syntax-pair ?$,1!N(B ?$,1"=(B tbl)
867 (set-case-syntax-pair ?$,1!O(B ?$,1#Y(B tbl)
868 (set-case-syntax-pair ?$,1!P(B ?$,1#[(B tbl)
869 (set-case-syntax-pair ?$,1!Q(B ?$,1!R(B tbl)
870 (set-case-syntax-pair ?$,1!S(B ?$,1#`(B tbl)
871 (set-case-syntax-pair ?$,1!T(B ?$,1#c(B tbl)
872 (set-case-syntax-pair ?$,1!V(B ?$,1#i(B tbl)
873 (set-case-syntax-pair ?$,1!W(B ?$,1#h(B tbl)
874 (set-case-syntax-pair ?$,1!X(B ?$,1!Y(B tbl)
875 (set-case-syntax-pair ?$,1!\(B ?$,1#o(B tbl)
876 (set-case-syntax-pair ?$,1!](B ?$,1#r(B tbl)
877 (set-case-syntax-pair ?$,1!_(B ?$,1#u(B tbl)
878 (set-case-syntax-pair ?$,1!`(B ?$,1!a(B tbl)
879 (set-case-syntax-pair ?$,1!b(B ?$,1!c(B tbl)
880 (set-case-syntax-pair ?$,1!d(B ?$,1!e(B tbl)
881 (set-case-syntax-pair ?$,1!f(B ?$,1$ (B tbl)
882 (set-case-syntax-pair ?$,1!g(B ?$,1!h(B tbl)
883 (set-case-syntax-pair ?$,1!i(B ?$,1$#(B tbl)
884 (set-case-syntax-pair ?$,1!l(B ?$,1!m(B tbl)
885 (set-case-syntax-pair ?$,1!n(B ?$,1$((B tbl)
886 (set-case-syntax-pair ?$,1!o(B ?$,1!p(B tbl)
887 (set-case-syntax-pair ?$,1!q(B ?$,1$*(B tbl)
888 (set-case-syntax-pair ?$,1!r(B ?$,1$+(B tbl)
889 (set-case-syntax-pair ?$,1!s(B ?$,1!t(B tbl)
890 (set-case-syntax-pair ?$,1!u(B ?$,1!v(B tbl)
891 (set-case-syntax-pair ?$,1!w(B ?$,1$2(B tbl)
892 (set-case-syntax-pair ?$,1!x(B ?$,1!y(B tbl)
893 (set-case-syntax-pair ?$,1!|(B ?$,1!}(B tbl)
894 (set-case-syntax-pair ?$,1"$(B ?$,1"&(B tbl)
895 (set-case-syntax-pair ?$,1"%(B ?$,1"&(B tbl)
896 (set-case-syntax-pair ?$,1"'(B ?$,1")(B tbl)
897 (set-case-syntax-pair ?$,1"((B ?$,1")(B tbl)
898 (set-case-syntax-pair ?$,1"*(B ?$,1",(B tbl)
899 (set-case-syntax-pair ?$,1"+(B ?$,1",(B tbl)
900 (set-case-syntax-pair ?$,1"-(B ?$,1".(B tbl)
901 (set-case-syntax-pair ?$,1"/(B ?$,1"0(B tbl)
902 (set-case-syntax-pair ?$,1"1(B ?$,1"2(B tbl)
903 (set-case-syntax-pair ?$,1"3(B ?$,1"4(B tbl)
904 (set-case-syntax-pair ?$,1"5(B ?$,1"6(B tbl)
905 (set-case-syntax-pair ?$,1"7(B ?$,1"8(B tbl)
906 (set-case-syntax-pair ?$,1"9(B ?$,1":(B tbl)
907 (set-case-syntax-pair ?$,1";(B ?$,1"<(B tbl)
908 (set-case-syntax-pair ?$,1">(B ?$,1"?(B tbl)
909 (set-case-syntax-pair ?$,1"@(B ?$,1"A(B tbl)
910 (set-case-syntax-pair ?$,1"B(B ?$,1"C(B tbl)
911 (set-case-syntax-pair ?$,1"D(B ?$,1"E(B tbl)
912 (set-case-syntax-pair ?$,1"F(B ?$,1"G(B tbl)
913 (set-case-syntax-pair ?$,1"H(B ?$,1"I(B tbl)
914 (set-case-syntax-pair ?$,1"J(B ?$,1"K(B tbl)
915 (set-case-syntax-pair ?$,1"L(B ?$,1"M(B tbl)
916 (set-case-syntax-pair ?$,1"N(B ?$,1"O(B tbl)
917 ;; 01F0; F; 006A 030C; # LATIN SMALL LETTER J WITH CARON
918 (set-case-syntax-pair ?$,1"Q(B ?$,1"S(B tbl)
919 (set-case-syntax-pair ?$,1"R(B ?$,1"S(B tbl)
920 (set-case-syntax-pair ?$,1"T(B ?$,1"U(B tbl)
921 (set-case-syntax-pair ?$,1"V(B ?$,1!U(B tbl)
922 (set-case-syntax-pair ?$,1"W(B ?$,1!(B tbl)
923 (set-case-syntax-pair ?$,1"X(B ?$,1"Y(B tbl)
924 (set-case-syntax-pair ?$,1"Z(B ?$,1"[(B tbl)
925 (set-case-syntax-pair ?$,1"\(B ?$,1"](B tbl)
926 (set-case-syntax-pair ?$,1"^(B ?$,1"_(B tbl)
927 (set-case-syntax-pair ?$,1"`(B ?$,1"a(B tbl)
928 (set-case-syntax-pair ?$,1"b(B ?$,1"c(B tbl)
929 (set-case-syntax-pair ?$,1"d(B ?$,1"e(B tbl)
930 (set-case-syntax-pair ?$,1"f(B ?$,1"g(B tbl)
931 (set-case-syntax-pair ?$,1"h(B ?$,1"i(B tbl)
932 (set-case-syntax-pair ?$,1"j(B ?$,1"k(B tbl)
933 (set-case-syntax-pair ?$,1"l(B ?$,1"m(B tbl)
934 (set-case-syntax-pair ?$,1"n(B ?$,1"o(B tbl)
935 (set-case-syntax-pair ?$,1"p(B ?$,1"q(B tbl)
936 (set-case-syntax-pair ?$,1"r(B ?$,1"s(B tbl)
937 (set-case-syntax-pair ?$,1"t(B ?$,1"u(B tbl)
938 (set-case-syntax-pair ?$,1"v(B ?$,1"w(B tbl)
939 (set-case-syntax-pair ?$,1"x(B ?$,1"y(B tbl)
940 (set-case-syntax-pair ?$,1"z(B ?$,1"{(B tbl)
941 (set-case-syntax-pair ?$,1"|(B ?$,1"}(B tbl)
942 (set-case-syntax-pair ?$,1"~(B ?$,1"(B tbl)
943 (set-case-syntax-pair ?$,1#"(B ?$,1##(B tbl)
944 (set-case-syntax-pair ?$,1#$(B ?$,1#%(B tbl)
945 (set-case-syntax-pair ?$,1#&(B ?$,1#'(B tbl)
946 (set-case-syntax-pair ?$,1#((B ?$,1#)(B tbl)
947 (set-case-syntax-pair ?$,1#*(B ?$,1#+(B tbl)
948 (set-case-syntax-pair ?$,1#,(B ?$,1#-(B tbl)
949 (set-case-syntax-pair ?$,1#.(B ?$,1#/(B tbl)
950 (set-case-syntax-pair ?$,1#0(B ?$,1#1(B tbl)
951 (set-case-syntax-pair ?$,1#2(B ?$,1#3(B tbl)
952
643 ;; Latin Extended Additional 953 ;; Latin Extended Additional
644 (setq c #x1e00) 954 (setq c #x1e00)
645 (while (<= c #x1ef9) 955 (while (<= c #x1ef9)
@@ -672,6 +982,13 @@
672 (set-case-syntax-pair ?$,1&n(B ?$,1'M(B tbl) 982 (set-case-syntax-pair ?$,1&n(B ?$,1'M(B tbl)
673 (set-case-syntax-pair ?$,1&o(B ?$,1'N(B tbl) 983 (set-case-syntax-pair ?$,1&o(B ?$,1'N(B tbl)
674 984
985 ;; Armenian
986 (setq c #x531)
987 (while (<= c #x556)
988 (set-case-syntax-pair (decode-char 'ucs c)
989 (decode-char 'ucs (+ c #x30)) tbl)
990 (setq c (1+ c)))
991
675 ;; Greek Extended 992 ;; Greek Extended
676 (setq c #x1f00) 993 (setq c #x1f00)
677 (while (<= c #x1fff) 994 (while (<= c #x1fff)
@@ -742,8 +1059,51 @@
742 (while (<= c #x2027) 1059 (while (<= c #x2027)
743 (set-case-syntax c "_" tbl) 1060 (set-case-syntax c "_" tbl)
744 (setq c (1+ c))) 1061 (setq c (1+ c)))
745 )
746 1062
1063 ;; Roman numerals
1064 (setq c #x2160)
1065 (while (<= c #x216f)
1066 (set-case-syntax-pair (decode-char 'ucs c)
1067 (decode-char 'ucs (+ c #x10)) tbl)
1068 (setq c (1+ c)))
1069
1070 ;; Circled Latin
1071 (setq c #x24b6)
1072 (while (<= c #x24cf)
1073 (set-case-syntax-pair (decode-char 'ucs c)
1074 (decode-char 'ucs (+ c 26)) tbl)
1075 (modify-category-entry (decode-char 'ucs c) ?l)
1076 (modify-category-entry (decode-char 'ucs (+ c 26)) ?l)
1077 (setq c (1+ c)))
1078
1079 ;; Fullwidth Latin
1080 (setq c #xff21)
1081 (while (<= c #xff3a)
1082 (set-case-syntax-pair (decode-char 'ucs c)
1083 (decode-char 'ucs (+ c #x20)) tbl)
1084 (modify-category-entry (decode-char 'ucs c) ?l)
1085 (modify-category-entry (decode-char 'ucs (+ c #x20)) ?l)
1086 (setq c (1+ c)))
1087
1088 ;; Ohm, Kelvin, Angstrom
1089 (set-case-syntax-pair ?$,1uf(B ?$,1'I(B tbl)
1090 (set-case-syntax-pair ?$,1uj(B ?k tbl)
1091 (set-case-syntax-pair ?$,1uk(B ?,Ae(B tbl)
1092
1093 ;; Combining diacritics
1094 (setq c #x300)
1095 (while (<= c #x362)
1096 (modify-category-entry (decode-char 'ucs c) ?^)
1097 (setq c (1+ c)))
1098
1099 ;; Combining marks
1100 (setq c #x20d0)
1101 (while (<= c #x20e3)
1102 (modify-category-entry (decode-char 'ucs c) ?^)
1103 (setq c (1+ c)))
1104
1105 ;; Fixme: syntax for symbols &c
1106 )
747 1107
748;;; Setting word boundary. 1108;;; Setting word boundary.
749 1109
@@ -815,8 +1175,8 @@
815 1175
816 1176
817;; Setup auto-fill-chars for charsets that should invoke auto-filling. 1177;; Setup auto-fill-chars for charsets that should invoke auto-filling.
818;; SPACE and NEWLIE are already set. Also put `nospace-between-words' 1178;; SPACE and NEWLINE are already set. Also put `nospace-between-words'
819;; property to the charsets. 1179;; property on the charsets.
820(let ((l '(katakana-jisx0201 1180(let ((l '(katakana-jisx0201
821 japanese-jisx0208 japanese-jisx0212 1181 japanese-jisx0208 japanese-jisx0212
822 chinese-gb2312 chinese-big5-1 chinese-big5-2))) 1182 chinese-gb2312 chinese-big5-1 chinese-big5-2)))
generated by cgit v1.2.1 (git 2.18.0) at 2026-01-31 13:30:12 +0000