1 ;;; isd-turtle.el --- Utility to dump ideographic-structure as Turtle files
3 ;; Copyright (C) 2017 MORIOKA Tomohiko
5 ;; Author: MORIOKA Tomohiko <tomo@kanji.zinbun.kyoto-u.ac.jp>
6 ;; Keywords: Ideographic Structures (漢字構造、解字), IDS, CHISE, RDF, Turtle
8 ;; This file is a part of CHISE-ISD (Ideographic Structure Database).
10 ;; This program is free software; you can redistribute it and/or
11 ;; modify it under the terms of the GNU General Public License as
12 ;; published by the Free Software Foundation; either version 2, or (at
13 ;; your option) any later version.
15 ;; This program is distributed in the hope that it will be useful, but
16 ;; WITHOUT ANY WARRANTY; without even the implied warranty of
17 ;; MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
18 ;; General Public License for more details.
20 ;; You should have received a copy of the GNU General Public License
21 ;; along with this program; see the file COPYING. If not, write to
22 ;; the Free Software Foundation, Inc., 59 Temple Place - Suite 330,
23 ;; Boston, MA 02111-1307, USA.
27 (require 'cwiki-common)
29 (defvar isd-url-prefix "http://rdf.chise.org/data/")
31 (setq est-coded-charset-priority-list
42 =jis-x0208 =jis-x0208@1990
44 =jis-x0213-1@2000 =jis-x0213-1@2004
62 =cns11643-1 =cns11643-2 =cns11643-3
63 =cns11643-4 =cns11643-5 =cns11643-6 =cns11643-7
73 =zinbun-oracle =>zinbun-oracle
78 =+>ucs@iso =+>ucs@unicode
84 =>jis-x0208 =>jis-x0213-1
91 =>ucs@jis =>ucs@cns =>ucs@ks
94 =>>ucs@iso =>>ucs@unicode
95 =>>ucs@jis =>>ucs@cns =>>ucs@ks
113 ==cns11643-1 ==cns11643-2 ==cns11643-3
114 ==cns11643-4 ==cns11643-5 ==cns11643-6 ==cns11643-7
124 =>>jis-x0208 =>>jis-x0213-1 =>>jis-x0213-2
125 =+>jis-x0208 =+>jis-x0213-1 =+>jis-x0213-2
131 =jis-x0208@1983 =jis-x0208@1978
160 ;; (defvar isd-turtle-ccs-list nil)
161 (defvar chise-turtle-ccs-prefix-alist nil)
163 (defun charset-code-point-format-spec (ccs)
164 (cond ((memq ccs '(=ucs))
167 =gt-k =daikanwa =adobe-japan1
168 =cbeta =zinbun-oracle))
170 ((memq ccs '(=hanyo-denshi/ks
173 ((memq ccs '(=hanyo-denshi/tk ==hanyo-denshi/tk))
178 ;; (defun isd-turtle-uri-encode-feature-name (feature-name)
180 ;; ((eq '=ucs feature-name)
182 ;; ((eq '==>ucs@bucs feature-name)
185 ;; (mapconcat (lambda (c)
188 ;; (char-to-string c)))
189 ;; (www-uri-encode-feature-name feature-name)
191 (defun chise-turtle-uri-encode-ccs-name (feature-name)
193 ((eq '=ucs feature-name)
195 ((eq '=big5 feature-name)
197 ((eq '==>ucs@bucs feature-name)
200 (mapconcat (lambda (c)
209 (char-to-string c))))
210 (www-uri-encode-feature-name feature-name)
213 ;; (defun isd-turtle-format-ccs-code-point (ccs code-point)
214 ;; (unless (memq ccs isd-turtle-ccs-list)
215 ;; (setq isd-turtle-ccs-list (cons ccs isd-turtle-ccs-list)))
217 ;; (isd-turtle-uri-encode-feature-name ccs)
218 ;; (format (charset-code-point-format-spec ccs)
220 (defun chise-turtle-format-ccs-code-point (ccs code-point)
221 (let ((ccs-uri (chise-turtle-uri-encode-ccs-name ccs)))
222 (unless (assoc ccs-uri chise-turtle-ccs-prefix-alist)
223 (setq chise-turtle-ccs-prefix-alist
224 (cons (cons ccs-uri ccs)
225 chise-turtle-ccs-prefix-alist)))
228 (format (charset-code-point-format-spec ccs)
231 (defun isd-turtle-encode-char (object)
232 (let ((ccs-list est-coded-charset-priority-list)
234 (if (setq ret (encode-char object '=ucs))
235 (chise-turtle-format-ccs-code-point '=ucs ret)
237 (setq ccs (pop ccs-list))
238 (not (setq ret (encode-char object ccs 'defined-only)))))
240 (chise-turtle-format-ccs-code-point ccs ret)
242 ((and (setq ccs (car (split-char object)))
243 (setq ret (encode-char object ccs)))
244 (chise-turtle-format-ccs-code-point ccs ret)
247 (format (if est-hide-cgi-mode
248 "system-char-id=0x%X"
249 "system-char-id:0x%X")
250 (encode-char object 'system-char-id))
253 (defun isd-turtle-format-component (component separator level)
254 (cond ((characterp component)
256 (isd-turtle-encode-char component)
261 (let ((ret (find-char component)))
264 (isd-turtle-encode-char ret) separator ret))
265 ((setq ret (assq 'ideographic-structure component))
266 (if (eq separator ?\;)
268 (isd-turtle-format-char nil nil (cdr ret) (1+ level)))
269 (isd-turtle-format-char nil nil (cdr ret) (1+ level)))))))))
271 (defun isd-turtle-format-char (ccs code-point &optional ids-list level
272 prefix without-head-char)
277 (let ((indent (make-string (* level 4) ?\ ))
284 (if (and ccs code-point
285 (setq char (decode-char ccs code-point)))
286 (setq ids-list (get-char-attribute char 'ideographic-structure))))
287 (setq idc (car ids-list))
288 (setq c1 (nth 1 ids-list)
292 (setq idc (plist-get idc :char)))
294 (setq ret (find-char idc)))
297 (setq ret (find-char c1)))
300 (setq ret (find-char c2)))
303 (setq ret (find-char c3)))
306 ((eq idc ?\u2FF0) ; ⿰
324 ((memq idc '(?⿴ ?⿵ ?⿶ ?⿷ ?⿸ ?⿹ ?⿺))
335 %s %s:structure [ a idc:%c ;
340 (if without-head-char
342 (if (and ccs code-point)
344 (chise-turtle-format-ccs-code-point ccs code-point)
348 indent prefix p1 (isd-turtle-format-component c1 ?\; (1+ level))
349 indent prefix p2 (isd-turtle-format-component c2 ?\; (1+ level))
350 indent prefix p3 (isd-turtle-format-component c3 ?\ (1+ level))
352 (if without-head-char
361 %s %s:structure [ a idc:%c ;
365 (if without-head-char
367 (if (and ccs code-point)
369 (chise-turtle-format-ccs-code-point ccs code-point)
373 indent prefix p1 (isd-turtle-format-component c1 ?\; (1+ level))
374 indent prefix p2 (isd-turtle-format-component c2 ?\ (1+ level))
376 (if without-head-char
384 (defun isd-turtle-insert-char (ccs code-point)
385 (let ((ret (isd-turtle-format-char ccs code-point)))
390 (defun isd-turtle-insert-ccs-ranges (ccs &rest ranges)
391 (let (range code max-code)
393 (setq range (car ranges))
395 (setq code (car range)
396 max-code (cdr range))
397 (while (<= code max-code)
398 (isd-turtle-insert-char ccs code)
399 (setq code (1+ code)))
402 (isd-turtle-insert-char ccs range)
404 (t (error 'wrong-type-argument range)))
405 (setq ranges (cdr ranges)))))
407 (defun isd-turtle-dump-range (file path func &rest args)
409 (let ((coding-system-for-write 'utf-8-mcs-er)
410 ;; isd-turtle-ccs-list
411 chise-turtle-ccs-prefix-alist)
412 (if (file-directory-p path)
413 (setq path (expand-file-name file path)))
415 (goto-char (point-min))
416 ;; (dolist (ccs (sort isd-turtle-ccs-list
417 ;; #'char-attribute-name<))
418 ;; (insert (format "@prefix %s: <%s%s=> .\n"
419 ;; (isd-turtle-uri-encode-feature-name ccs)
420 ;; "http://www.chise.org/est/view/character/"
421 ;; (www-uri-encode-feature-name ccs))))
422 (dolist (cell (sort chise-turtle-ccs-prefix-alist
424 (char-attribute-name< (cdr a)(cdr b)))))
425 (insert (format "@prefix %s: <%s/%s=> .\n"
427 "http://www.chise.org/est/view/character"
428 (www-uri-encode-feature-name (cdr cell)))))
430 (goto-char (point-min))
431 (insert "# -*- coding: utf-8-mcs-er -*-\n")
432 (insert "@prefix : <http://rdf.chise.org/rdf/property/character/isd/> .
433 @prefix idc: <http://rdf.chise.org/rdf/type/character/idc/> .\n")
434 (write-region (point-min)(point-max) path))))
437 (defun isd-turtle-dump-ucs-basic (filename)
438 (interactive "Fdump ISD-UCS-Basic : ")
439 (isd-turtle-dump-range "ISD-UCS-Basic.ttl" filename
440 #'isd-turtle-insert-ccs-ranges
441 '=ucs '(#x4E00 . #x9FA5)))
444 (defun isd-turtle-dump-ucs-ext-a (filename)
445 (interactive "Fdump ISD-UCS-Ext-A : ")
446 (isd-turtle-dump-range "ISD-UCS-Ext-A.ttl" filename
447 #'isd-turtle-insert-ccs-ranges
448 '=ucs '(#x3400 . #x4DB5) #xFA1F #xFA23))
451 (defun isd-turtle-dump-ucs-ext-b-1 (filename)
452 (interactive "Fdump IDS-UCS-Ext-B-1 : ")
453 (isd-turtle-dump-range "ISD-UCS-Ext-B-1.ttl" filename
454 #'isd-turtle-insert-ccs-ranges
455 'ucs '(#x20000 . #x21FFF)))
458 (defun isd-turtle-dump-ucs-ext-b-2 (filename)
459 (interactive "Fdump IDS-UCS-Ext-B-2 : ")
460 (isd-turtle-dump-range "ISD-UCS-Ext-B-2.ttl" filename
461 #'isd-turtle-insert-ccs-ranges
462 'ucs '(#x22000 . #x23FFF)))
465 (defun isd-turtle-dump-ucs-ext-b-3 (filename)
466 (interactive "Fdump IDS-UCS-Ext-B-3 : ")
467 (isd-turtle-dump-range "ISD-UCS-Ext-B-3.ttl" filename
468 #'isd-turtle-insert-ccs-ranges
469 'ucs '(#x24000 . #x25FFF)))
472 (defun isd-turtle-dump-ucs-ext-b-4 (filename)
473 (interactive "Fdump IDS-UCS-Ext-B-4 : ")
474 (isd-turtle-dump-range "ISD-UCS-Ext-B-4.ttl" filename
475 #'isd-turtle-insert-ccs-ranges
476 'ucs '(#x26000 . #x27FFF)))
479 (defun isd-turtle-dump-ucs-ext-b-5 (filename)
480 (interactive "Fdump IDS-UCS-Ext-B-5 : ")
481 (isd-turtle-dump-range "ISD-UCS-Ext-B-5.ttl" filename
482 #'isd-turtle-insert-ccs-ranges
483 'ucs '(#x28000 . #x29FFF)))
486 (defun isd-turtle-dump-ucs-ext-b-6 (filename)
487 (interactive "Fdump IDS-UCS-Ext-B-6 : ")
488 (isd-turtle-dump-range "ISD-UCS-Ext-B-6.ttl" filename
489 #'isd-turtle-insert-ccs-ranges
490 'ucs '(#x2A000 . #x2A6D6)))
493 (defun isd-turtle-dump-ucs-ext-c (filename)
494 (interactive "Fdump IDS-UCS-Ext-C : ")
495 (isd-turtle-dump-range "ISD-UCS-Ext-C.ttl" filename
496 #'isd-turtle-insert-ccs-ranges
497 'ucs '(#x2A700 . #x2B734)))
500 (defun isd-turtle-dump-ucs-ext-d (filename)
501 (interactive "Fdump IDS-UCS-Ext-D : ")
502 (isd-turtle-dump-range "ISD-UCS-Ext-D.ttl" filename
503 #'isd-turtle-insert-ccs-ranges
504 'ucs '(#x2B740 . #x2B81D)))
507 (defun isd-turtle-dump-ucs-ext-e (filename)
508 (interactive "Fdump IDS-UCS-Ext-E : ")
509 (isd-turtle-dump-range "ISD-UCS-Ext-E.ttl" filename
510 #'isd-turtle-insert-ccs-ranges
511 'ucs '(#x2B820 . #x2CEA1)))
514 (defun isd-turtle-dump-mj-0 (filename)
515 (interactive "Fdump ISD-MJ-0 : ")
516 (isd-turtle-dump-range "ISD-MJ-0.ttl" filename
517 #'isd-turtle-insert-ccs-ranges
521 (defun isd-turtle-dump-mj-1 (filename)
522 (interactive "Fdump ISD-MJ-1 : ")
523 (isd-turtle-dump-range "ISD-MJ-1.ttl" filename
524 #'isd-turtle-insert-ccs-ranges
525 '=mj '(10000 . 19999)))
528 (defun isd-turtle-dump-mj-2 (filename)
529 (interactive "Fdump ISD-MJ-2 : ")
530 (isd-turtle-dump-range "ISD-MJ-2.ttl" filename
531 #'isd-turtle-insert-ccs-ranges
532 '=mj '(20000 . 29999)))
535 (defun isd-turtle-dump-mj-3 (filename)
536 (interactive "Fdump ISD-MJ-3 : ")
537 (isd-turtle-dump-range "ISD-MJ-3.ttl" filename
538 #'isd-turtle-insert-ccs-ranges
539 '=mj '(30000 . 39999)))
542 (defun isd-turtle-dump-mj-4 (filename)
543 (interactive "Fdump ISD-MJ-4 : ")
544 (isd-turtle-dump-range "ISD-MJ-4.ttl" filename
545 #'isd-turtle-insert-ccs-ranges
546 '=mj '(40000 . 49999)))
549 (defun isd-turtle-dump-mj-5 (filename)
550 (interactive "Fdump ISD-MJ-5 : ")
551 (isd-turtle-dump-range "ISD-MJ-5.ttl" filename
552 #'isd-turtle-insert-ccs-ranges
553 '=mj '(50000 . 59999)))
556 (defun isd-turtle-dump-mj-6 (filename)
557 (interactive "Fdump ISD-MJ-6 : ")
558 (isd-turtle-dump-range "ISD-MJ-6.ttl" filename
559 #'isd-turtle-insert-ccs-ranges
560 '=mj '(60000 . 69999)))
563 (defun isd-turtle-dump-all (directory)
564 (interactive "DISD directory : ")
565 (isd-turtle-dump-ucs-basic directory)
566 (isd-turtle-dump-ucs-ext-a directory)
567 (isd-turtle-dump-ucs-ext-b-1 directory)
568 (isd-turtle-dump-ucs-ext-b-2 directory)
569 (isd-turtle-dump-ucs-ext-b-3 directory)
570 (isd-turtle-dump-ucs-ext-b-4 directory)
571 (isd-turtle-dump-ucs-ext-b-5 directory)
572 (isd-turtle-dump-ucs-ext-b-6 directory)
573 (isd-turtle-dump-ucs-ext-c directory)
574 (isd-turtle-dump-ucs-ext-d directory)
575 (isd-turtle-dump-ucs-ext-e directory)
576 (isd-turtle-dump-mj-0 directory)
577 (isd-turtle-dump-mj-1 directory)
578 (isd-turtle-dump-mj-2 directory)
579 (isd-turtle-dump-mj-3 directory)
580 (isd-turtle-dump-mj-4 directory)
581 (isd-turtle-dump-mj-5 directory)
582 (isd-turtle-dump-mj-6 directory)
589 (provide 'isd-turtle)
591 ;;; isd-turtle.el ends here