1 ;; -*- coding: utf-8-mcs-er -*-
2 (require 'char-db-util)
4 (defvar chise-wiki-view-url "view.cgi")
5 (defvar chise-wiki-edit-url "edit/edit.cgi")
7 (defvar chise-wiki-bitmap-glyphs-url
8 "http://chise.zinbun.kyoto-u.ac.jp/glyphs")
10 (defvar chise-wiki-glyph-cgi-url
11 "http://chise.zinbun.kyoto-u.ac.jp/chisewiki/glyph.cgi")
13 (defun decode-uri-string (string &optional coding-system)
14 (if (> (length string) 0)
18 (mapconcat (lambda (char)
21 (char-to-string char)))
23 (while (string-match "%\\([0-9A-F][0-9A-F]\\)" string i)
24 (setq dest (concat dest
25 (substring string i (match-beginning 0))
28 (string-to-int (match-string 1 string) 16))))
31 (concat dest (substring string i))
34 (defun www-feature-type (feature-name)
35 (or (char-feature-property feature-name 'type)
36 (let ((str (symbol-name feature-name)))
38 ((string-match "\\*note\\(@[^*]+\\)?$" str)
40 ((string-match "\\*sources\\(@[^*]+\\)?$" str)
42 ((string-match "\\*" str)
44 ((string-match "^\\(->\\|<-\\)" str)
46 ((string-match "^ideographic-structure\\(@\\|$\\)" str)
50 (defun www-feature-value-format (feature-name)
51 (or (char-feature-property feature-name 'value-format)
52 (let ((type (www-feature-type feature-name)))
53 (cond ((eq type 'relation)
54 'space-separated-char-list)
60 (if (find-charset feature-name)
61 (if (and (= (charset-dimension feature-name) 2)
62 (= (charset-chars feature-name) 94))
64 " (" (decimal) ") <" (ku-ten) ">")
65 '("0x" (HEX) " (" (decimal) ")")))))
67 (defun char-feature-name-at-domain (feature-name domain)
68 (let ((name (symbol-name feature-name)))
70 ((string-match "@[^*]+$" name)
71 (intern (format "%s/%s" name domain))
74 (intern (format "%s@%s" name domain))
77 (defun char-feature-name-sans-versions (feature)
78 (let ((feature-name (symbol-name feature)))
79 (if (string-match "[@/]\\$rev=latest$" feature-name)
80 (intern (substring feature-name 0 (match-beginning 0)))
83 (defun www-char-feature (character feature)
84 (let ((latest-feature (char-feature-name-at-domain feature '$rev=latest)))
85 (mount-char-attribute-table latest-feature)
86 (or (char-feature character latest-feature)
87 (char-feature character feature))))
90 ;;; @ URI representation
93 (defun www-uri-decode-feature-name (uri-feature)
96 ((string-match "^from\\." uri-feature)
97 (intern (format "<-%s" (substring uri-feature (match-end 0))))
99 ((string-match "^to\\." uri-feature)
100 (intern (format "->%s" (substring uri-feature (match-end 0))))
102 ((string-match "^rep\\." uri-feature)
103 (intern (format "=%s" (substring uri-feature (match-end 0))))
105 ((string-match "^g\\." uri-feature)
106 (intern (format "=>>%s" (substring uri-feature (match-end 0))))
108 ((string-match "^gi\\." uri-feature)
109 (intern (format "=>>>%s" (substring uri-feature (match-end 0))))
111 ((string-match "^gi\\([0-9]+\\)\\." uri-feature)
112 (intern (format "=>>%s%s"
113 (make-string (string-to-int
114 (match-string 1 uri-feature))
116 (substring uri-feature (match-end 0))))
118 ((string-match "^a\\." uri-feature)
119 (intern (format "=>%s" (substring uri-feature (match-end 0))))
121 ((string-match "^a\\([0-9]+\\)\\." uri-feature)
122 (intern (format "%s>%s"
123 (make-string (string-to-int
124 (match-string 1 uri-feature))
126 (substring uri-feature (match-end 0))))
128 ((and (setq feature (intern (format "=>%s" uri-feature)))
129 (find-charset feature))
131 ((and (setq feature (intern (format "=>>%s" uri-feature)))
132 (find-charset feature))
134 ((and (setq feature (intern (format "=>>>%s" uri-feature)))
135 (find-charset feature))
137 ((and (setq feature (intern (format "=%s" uri-feature)))
138 (find-charset feature))
140 (t (intern uri-feature)))))
142 (defun www-uri-encode-feature-name (feature-name)
143 (setq feature-name (symbol-name feature-name))
145 ((string-match "^=\\([^=>]+\\)" feature-name)
146 (concat "rep." (substring feature-name (match-beginning 1)))
148 ((string-match "^=>>\\([^=>]+\\)" feature-name)
149 (concat "g." (substring feature-name (match-beginning 1)))
151 ((string-match "^=>>>\\([^=>]+\\)" feature-name)
152 (concat "gi." (substring feature-name (match-beginning 1)))
154 ((string-match "^=>>\\(>+\\)" feature-name)
156 (length (match-string 1 feature-name))
157 (substring feature-name (match-end 1)))
159 ((string-match "^=>\\([^=>]+\\)" feature-name)
160 (concat "a." (substring feature-name (match-beginning 1)))
162 ((string-match "^\\(=+\\)>" feature-name)
164 (length (match-string 1 feature-name))
165 (substring feature-name (match-end 0)))
167 ((string-match "^->" feature-name)
168 (concat "to." (substring feature-name (match-end 0)))
170 ((string-match "^<-" feature-name)
171 (concat "from." (substring feature-name (match-end 0)))
175 (defun www-uri-decode-char (char-rep)
178 ((string-match "\\(%3A\\|:\\)" char-rep)
179 (setq ccs (substring char-rep 0 (match-beginning 0))
180 cpos (substring char-rep (match-end 0)))
181 (setq ccs (www-uri-decode-feature-name ccs))
183 ((string-match "^0x" cpos)
185 (string-to-number (substring cpos (match-end 0)) 16))
188 (setq cpos (string-to-number cpos))
191 (decode-char ccs cpos))
194 (setq char-rep (decode-uri-string char-rep 'utf-8-mcs-er))
195 (when (= (length char-rep) 1)
199 (defun www-uri-encode-char (char)
200 (if (encode-char char '=ucs)
203 (format "%%%02X" byte))
204 (encode-coding-string (char-to-string char) 'utf-8-mcs-er)
206 (let ((ccs-list '(; =ucs
207 =cns11643-1 =cns11643-2 =cns11643-3
208 =cns11643-4 =cns11643-5 =cns11643-6 =cns11643-7
210 =jis-x0208 =jis-x0208@1990
213 =jis-x0213-1@2000 =jis-x0213-1@2004
214 =jis-x0208@1983 =jis-x0208@1978
218 =>>jis-x0208 =>>jis-x0213-1
219 =>jis-x0208 =>jis-x0213-1
225 (setq ccs (pop ccs-list))
226 (not (setq ret (encode-char char ccs 'defined-only)))))
229 (www-uri-encode-feature-name ccs)
231 ((and (setq ccs (car (split-char char)))
232 (setq ret (encode-char char ccs)))
234 (www-uri-encode-feature-name ccs)
237 (format "system-char-id:0x%X"
238 (encode-char char 'system-char-id))
242 ;;; @ Feature name presentation
245 (defun www-format-feature-name-default (feature-name)
249 (symbol-name feature-name)
253 (defun www-format-feature-name-as-metadata (feature-name &optional lang)
254 (let ((str (symbol-name feature-name))
257 ((string-match "\\*[^*]+$" str)
258 (setq base (substring str 0 (match-beginning 0))
259 meta (substring str (match-beginning 0)))
260 (concat (www-format-feature-name* (intern base) lang)
263 (www-format-feature-name-default feature-name)
266 (defun www-format-feature-name-as-rel-to (feature-name)
267 (concat "\u2192" (substring (symbol-name feature-name) 2)))
269 (defun www-format-feature-name-as-rel-from (feature-name)
270 (concat "\u2190" (substring (symbol-name feature-name) 2)))
272 (defun www-format-feature-name-as-CCS (feature-name)
275 (symbol-name feature-name)
277 (dest (upcase (pop rest))))
278 (when (string-match "^=+>*" dest)
279 (setq dest (concat (substring dest 0 (match-end 0))
281 (substring dest (match-end 0)))))
285 (setq dest (concat dest " " (upcase (pop rest)))))
286 (if (string-match "^[0-9]+$" (car rest))
287 (concat dest "-" (car rest))
288 (concat dest " " (upcase (car rest))))
292 (defun www-format-feature-name* (feature-name &optional lang)
296 (char-feature-property
298 (intern (format "name@%s" lang))))
299 (char-feature-property
300 feature-name 'name)))
301 ((find-charset feature-name)
302 (www-format-feature-name-as-CCS feature-name))
303 ((and (setq name (symbol-name feature-name))
304 (string-match "\\*" name))
305 (www-format-feature-name-as-metadata feature-name lang))
306 ((string-match "^\\(->\\)" name)
307 (www-format-feature-name-as-rel-to feature-name))
308 ((string-match "^\\(<-\\)" name)
309 (www-format-feature-name-as-rel-from feature-name))
311 (www-format-feature-name-default feature-name)))))
313 (defun www-format-feature-name (feature-name &optional lang)
314 (www-format-encode-string
315 (www-format-feature-name* feature-name lang)))
318 ;;; @ Feature value presentation
321 (defun www-format-value-as-kuten (value)
323 (- (lsh value -8) 32)
324 (- (logand value 255) 32)))
326 (defun www-format-value-default (value &optional without-tags)
330 (www-format-encode-string
334 (www-format-encode-string (format "%S" value) without-tags)))
336 (defun www-format-value-as-char-list (value &optional without-tags)
341 (www-format-encode-string
342 (format (if (characterp unit)
348 (if (characterp unit)
349 (format "<a href=\"%s?char=%s\">%s</a>"
351 (www-uri-encode-char unit)
352 (www-format-encode-string (char-to-string unit)))
353 (www-format-encode-string (format "%s" unit)))))
355 (www-format-encode-string (format "%s" value) without-tags)))
357 (defun www-format-value-as-ids (value &optional without-tags)
362 (www-format-encode-string
363 (format (if (characterp unit)
369 (if (characterp unit)
370 (format "<a href=\"%s?char=%s\">%s</a>"
372 (www-uri-encode-char unit)
373 (www-format-encode-string (char-to-string unit)))
374 (www-format-encode-string (format "%s" unit)))))
375 (ideographic-structure-to-ids value) " ")
376 (www-format-encode-string (format "%s" value) without-tags)))
378 (defun www-format-value-as-S-exp (value &optional without-tags)
379 (www-format-encode-string (format "%S" value) without-tags))
381 (defun www-format-value-as-HEX (value)
384 (www-format-value-as-S-exp value)))
386 (defun www-format-value-as-CCS-default (value)
389 (www-format-value-as-HEX value)
391 (www-format-value-as-S-exp value)))
393 (defun www-format-value-as-CCS-94x94 (value)
395 (format "0x%s [%s] (%d)"
396 (www-format-value-as-HEX value)
397 (www-format-value-as-kuten value)
399 (www-format-value-as-S-exp value)))
401 (defun www-format-value (object feature-name
402 &optional value format without-tags)
404 (setq value (www-char-feature object feature-name)))
405 (www-format-apply-value object feature-name
406 format nil value nil nil without-tags)
410 ;;; @ format evaluator
413 (defun www-format-encode-string (string &optional without-tags)
416 (let (plane code start end char variants ret)
417 (goto-char (point-min))
418 (while (search-forward "<" nil t)
419 (replace-match "<" nil t))
420 (goto-char (point-min))
421 (while (search-forward ">" nil t)
422 (replace-match ">" nil t))
424 (encode-coding-region (point-min)(point-max) 'utf-8-mcs-er)
425 (let ((coded-charset-entity-reference-alist
428 '(=cns11643-1 "C1-" 4 X)
429 '(=cns11643-2 "C2-" 4 X)
430 '(=cns11643-3 "C3-" 4 X)
431 '(=cns11643-4 "C4-" 4 X)
432 '(=cns11643-5 "C5-" 4 X)
433 '(=cns11643-6 "C6-" 4 X)
434 '(=cns11643-7 "C7-" 4 X)
436 '(=gb12345 "G1-" 4 X)
437 '(=jis-x0208@1990 "J90-" 4 X)
438 '(=jis-x0212 "JSP-" 4 X)
440 '(=jis-x0208@1997 "J97-" 4 X)
441 '(=jis-x0208@1978 "J78-" 4 X)
442 '(=jis-x0208@1983 "J83-" 4 X)
443 '(=zinbun-oracle "ZOB-" 4 d)
444 '(=jef-china3 "JC3-" 4 X)
445 '(=daikanwa "M-" 5 d)
446 coded-charset-entity-reference-alist)))
447 (encode-coding-region (point-min)(point-max) 'utf-8-mcs-er)
449 (goto-char (point-min))
450 (while (re-search-forward "&CB\\([0-9]+\\);" nil t)
451 (setq code (string-to-int (match-string 1)))
453 (format "<img alt=\"CB%05d\" src=\"%s/cb-gaiji/%02d/CB%05d.gif\">"
455 chise-wiki-bitmap-glyphs-url
459 (goto-char (point-min))
460 (while (re-search-forward "&J\\(78\\|83\\|90\\|97\\|SP\\)-\\([0-9A-F][0-9A-F][0-9A-F][0-9A-F]\\);" nil t)
461 (setq plane (match-string 1)
462 code (string-to-int (match-string 2) 16))
464 (format "<img alt=\"J%s-%04X\" src=\"%s/JIS-%s/%02d-%02d.gif\">"
466 chise-wiki-bitmap-glyphs-url
469 (- (logand code 255) 32))
472 (goto-char (point-min))
473 (while (re-search-forward "&G\\([01]\\)-\\([0-9A-F][0-9A-F][0-9A-F][0-9A-F]\\);" nil t)
474 (setq plane (string-to-int (match-string 1))
475 code (string-to-int (match-string 2) 16))
477 (format "<img alt=\"GB%d-%04X\" src=\"%s/GB%d/%02d-%02d.gif\">"
479 chise-wiki-bitmap-glyphs-url
482 (- (logand code 255) 32))
485 (goto-char (point-min))
486 (while (re-search-forward "&C\\([1-7]\\)-\\([0-9A-F][0-9A-F][0-9A-F][0-9A-F]\\);" nil t)
487 (setq plane (string-to-int (match-string 1))
488 code (string-to-int (match-string 2) 16))
490 (format "<img alt=\"CNS%d-%04X\" src=\"%s/CNS%d/%04X.gif\">"
492 chise-wiki-bitmap-glyphs-url
496 (goto-char (point-min))
497 (while (re-search-forward "&JC3-\\([0-9A-F][0-9A-F][0-9A-F][0-9A-F]\\);" nil t)
498 (setq code (string-to-int (match-string 1) 16))
500 (format "<img alt=\"JC3-%04X\" src=\"http://kanji.zinbun.kyoto-u.ac.jp/db/CHINA3/Gaiji/%04x.gif\">"
504 (goto-char (point-min))
505 (while (re-search-forward "&ZOB-\\([0-9]+\\);" nil t)
506 (setq code (string-to-int (match-string 1)))
508 (format "<img alt=\"ZOB-%04d\" src=\"%s/ZOB-1968/%04d.png\">"
510 chise-wiki-bitmap-glyphs-url
514 (goto-char (point-min))
515 (while (re-search-forward "&\\(G-\\)?GT-\\([0-9]+\\);" nil t)
516 (setq code (string-to-int (match-string 2)))
518 (format "<img alt=\"GT-%05d\" src=\"%s?char=GT-%05d\">"
520 chise-wiki-glyph-cgi-url
524 (goto-char (point-min))
525 (while (re-search-forward "&\\(G-\\)?GT-K\\([0-9]+\\);" nil t)
526 (setq code (string-to-int (match-string 2)))
528 (format "<img alt=\"GT-K%05d\" src=\"%s?char=GT-K%05d\">"
530 chise-wiki-glyph-cgi-url
534 (goto-char (point-min))
535 (while (re-search-forward "&B-\\([0-9A-F]+\\);" nil t)
536 (setq code (string-to-int (match-string 1) 16))
538 (format "<img alt=\"B-%04X\" src=\"%s?char=B-%04X\">"
540 chise-wiki-glyph-cgi-url
544 (goto-char (point-min))
545 (while (re-search-forward "&CDP-\\([0-9A-F]+\\);" nil t)
546 (setq code (string-to-int (match-string 1) 16))
548 (format "<img alt=\"CDP-%04X\" src=\"%s?char=CDP-%04X\">"
550 chise-wiki-glyph-cgi-url
554 (goto-char (point-min))
555 (while (re-search-forward "&\\(UU\\+\\|U-\\)\\([0-9A-F]+\\);" nil t)
556 (setq code (string-to-int (match-string 2) 16))
558 (format "<img alt=\"UU+%04X\" src=\"http://www.unicode.org/cgi-bin/refglyph?24-%04X\">"
563 (goto-char (point-min))
564 (while (re-search-forward "&MCS-\\([0-9A-F]+\\);" nil t)
565 (setq code (string-to-int (match-string 1) 16))
566 (setq start (match-beginning 0)
568 (setq char (decode-char 'system-char-id code))
569 (setq variants (or (www-char-feature char '->subsumptive)
570 (www-char-feature char '->denotational)))
572 (setq ret (www-format-encode-string
573 (char-to-string (car variants))))
574 (string-match "&MCS-\\([0-9A-F]+\\);" ret))
575 (setq variants (cdr variants)))
576 (unless (string-match "&MCS-\\([0-9A-F]+\\);" ret)
578 (delete-region start end)
581 ;; (goto-char (point-min))
582 ;; (while (search-forward ">-" nil t)
583 ;; (replace-match "&GT-" t 'literal))
586 (defun www-format-props-to-string (props &optional format)
588 (setq format (plist-get props :format)))
590 (plist-get props :flag)
591 (if (plist-get props :zero-padding)
593 (if (plist-get props :len)
594 (format "%d" (plist-get props :len)))
596 ((eq format 'decimal) "d")
597 ((eq format 'hex) "x")
598 ((eq format 'HEX) "X")
599 ((eq format 'S-exp) "S")
602 (defun www-format-apply-value (object feature-name
604 &optional uri-char uri-feature
609 ((memq format '(decimal hex HEX))
611 (format (www-format-props-to-string props format)
613 (www-format-encode-string
617 ((eq format 'wiki-text)
619 (www-xml-format-list value)
620 (www-format-eval-list value object feature-name nil uri-char))
623 (www-format-encode-string
624 (format (www-format-props-to-string props format)
628 (www-format-value-as-kuten value))
629 ((eq format 'space-separated-char-list)
630 (www-format-value-as-char-list value without-tags))
631 ((eq format 'space-separated-ids)
632 (www-format-value-as-ids value without-tags))
634 (www-format-encode-string (format "%s" value) without-tags)
637 (www-format-value-default value without-tags)
640 (if (or without-tags (eq (plist-get props :mode) 'peek))
642 (format "%s <a href=\"%s?char=%s&feature=%s&format=%s\"
643 ><input type=\"submit\" value=\"edit\" /></a>"
646 uri-char uri-feature format))))
648 (defun www-format-eval-feature-value (char
650 &optional format lang uri-char value)
652 (setq value (www-char-feature char feature-name)))
654 (setq format (www-feature-value-format feature-name)))
657 (www-format-apply-value
660 uri-char (www-uri-encode-feature-name feature-name))
663 (cond ((null (cdr format))
664 (setq format (car format))
665 (www-format-apply-value
667 (car format) (nth 1 format) value
668 uri-char (www-uri-encode-feature-name feature-name))
671 (www-format-eval-list format char feature-name lang uri-char)
674 (defun www-format-eval-unit (exp char feature-name
675 &optional lang uri-char value)
677 (setq value (www-char-feature char feature-name)))
679 (setq uri-char (www-uri-encode-char char)))
681 ((stringp exp) (www-format-encode-string exp))
685 ((memq (car exp) '(value decimal hex HEX ku-ten S-exp default))
686 (if (eq (car exp) 'value)
687 (www-format-eval-feature-value char feature-name
688 (plist-get (nth 1 exp) :format)
690 (www-format-apply-value
692 (car exp) (nth 1 exp) value
693 uri-char (www-uri-encode-feature-name feature-name)))
695 ((eq (car exp) 'name)
696 (format "<a href=\"%s?feature=%s&char=%s\">%s</a>"
698 (www-uri-encode-feature-name feature-name)
700 (www-format-feature-name feature-name lang))
702 ((eq (car exp) 'link)
707 (www-format-eval-list (plist-get (nth 1 exp) :ref)
708 char feature-name lang uri-char)
709 (www-format-eval-list (nthcdr 2 exp)
710 char feature-name lang uri-char)))
716 (www-format-eval-list (nthcdr 2 exp) char feature-name
720 (defun www-format-eval-list (format-list char feature-name
721 &optional lang uri-char)
722 (if (consp format-list)
725 (www-format-eval-unit exp char feature-name lang uri-char))
727 (www-format-eval-unit format-list char feature-name lang uri-char)))
733 (defun www-xml-format-props (props)
737 (setq key (pop props)
740 (setq key (symbol-name key)))
741 (if (eq (aref key 0) ?:)
742 (setq key (substring key 1)))
744 (format "%s %s=\"%s\""
746 (www-format-encode-string
747 (format "%s" val) 'without-tags))))
750 (defun www-xml-format-unit (format-unit)
751 (let (name props children ret)
753 ((stringp format-unit)
754 (mapconcat (lambda (c)
757 ;; ((eq c ?<) "&lt;")
758 ;; ((eq c ?>) "&gt;")
760 (char-to-string c))))
761 (www-format-encode-string format-unit 'without-tags)
765 (setq name (car format-unit)
766 props (nth 1 format-unit)
767 children (nthcdr 2 format-unit))
768 (when (eq name 'link)
769 (setq ret (plist-get props :ref))
770 (unless (stringp ret)
771 (setq props (plist-remprop (copy-list props) :ref))
773 (cons (list* 'ref nil ret)
776 (format "<%s%s>%s</%s>"
779 (www-xml-format-props props)
781 (www-xml-format-list children)
784 name (www-xml-format-props props)))
787 (format "%s" format-unit)))))
789 (defun www-xml-format-list (format-list)
790 (if (atom format-list)
791 (www-xml-format-unit format-list)
792 (mapconcat #'www-xml-format-unit
799 (defun www-html-display-text (text)
803 (goto-char (point-min))
804 (while (search-forward "<" nil t)
805 (replace-match "<" nil t))
806 (goto-char (point-min))
807 (while (search-forward ">" nil t)
808 (replace-match ">" nil t))
809 (goto-char (point-min))
810 (while (re-search-forward "\\[\\[\\([^]|[]+\\)|\\([^][]+\\)\\]\\]" nil t)
812 (format "<a href=\"%s\">%s</a>"
816 (encode-coding-region (point-min)(point-max) 'utf-8-mcs-er)
817 (goto-char (point-min))
818 (while (search-forward ">-" nil t)
819 (replace-match "&GT-" nil t))
822 (defun www-html-display-paragraph (text)
824 (www-html-display-text text)
827 (provide 'cwiki-common)