1 ;;; isd-turtle.el --- Utility to dump ideographic-structure as Turtle files
3 ;; Copyright (C) 2017, 2018 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
66 =cns11643-1 =cns11643-2 =cns11643-3
67 =cns11643-4 =cns11643-5 =cns11643-6 =cns11643-7
77 =zinbun-oracle =>zinbun-oracle
82 =+>ucs@iso =+>ucs@unicode
88 =>jis-x0208 =>jis-x0213-1
96 =>ucs@jis =>ucs@cns =>ucs@ks
99 =>>ucs@iso =>>ucs@unicode
100 =>>ucs@jis =>>ucs@cns =>>ucs@ks
118 ==cns11643-1 ==cns11643-2 ==cns11643-3
119 ==cns11643-4 ==cns11643-5 ==cns11643-6 ==cns11643-7
129 =>>jis-x0208 =>>jis-x0213-1 =>>jis-x0213-2
130 =+>jis-x0208 =+>jis-x0213-1 =+>jis-x0213-2
136 =jis-x0208@1983 =jis-x0208@1978
170 ;; (defvar isd-turtle-ccs-list nil)
171 (defvar chise-turtle-ccs-prefix-alist nil)
173 (defun charset-code-point-format-spec (ccs)
174 (cond ((memq ccs '(=ucs))
177 (let ((ccs-name (symbol-name ccs)))
180 "\\(shinjigen\\|daikanwa/ho\\|=>iwds-1\\)"
184 "\\(gt\\|daikanwa\\|adobe-japan1\\|cbeta\\|zinbun-oracle\\|hng\\)"
187 ((string-match "\\(hanyo-denshi/ks\\|koseki\\|mj\\)" ccs-name)
189 ((string-match "hanyo-denshi/tk" ccs-name)
194 ;; (defun isd-turtle-uri-encode-feature-name (feature-name)
196 ;; ((eq '=ucs feature-name)
198 ;; ((eq '==>ucs@bucs feature-name)
201 ;; (mapconcat (lambda (c)
204 ;; (char-to-string c)))
205 ;; (www-uri-encode-feature-name feature-name)
207 (defun chise-turtle-uri-encode-ccs-name (feature-name)
209 ((eq '=ucs feature-name)
211 ((eq '=big5 feature-name)
213 ((eq '==>ucs@bucs feature-name)
216 (mapconcat (lambda (c)
225 (char-to-string c))))
226 (www-uri-encode-feature-name feature-name)
229 ;; (defun isd-turtle-format-ccs-code-point (ccs code-point)
230 ;; (unless (memq ccs isd-turtle-ccs-list)
231 ;; (setq isd-turtle-ccs-list (cons ccs isd-turtle-ccs-list)))
233 ;; (isd-turtle-uri-encode-feature-name ccs)
234 ;; (format (charset-code-point-format-spec ccs)
236 (defun chise-turtle-format-ccs-code-point (ccs code-point)
237 (let ((ccs-uri (chise-turtle-uri-encode-ccs-name ccs)))
238 (unless (assoc ccs-uri chise-turtle-ccs-prefix-alist)
239 (setq chise-turtle-ccs-prefix-alist
240 (cons (cons ccs-uri ccs)
241 chise-turtle-ccs-prefix-alist)))
244 (format (charset-code-point-format-spec ccs)
247 (defun isd-turtle-encode-char (object)
248 (let ((ccs-list est-coded-charset-priority-list)
250 (if (setq ret (encode-char object '=ucs))
251 (chise-turtle-format-ccs-code-point '=ucs ret)
253 (setq ccs (pop ccs-list))
254 (not (setq ret (encode-char object ccs 'defined-only)))))
256 (chise-turtle-format-ccs-code-point ccs ret)
258 ((and (setq ccs (car (split-char object)))
259 (setq ret (encode-char object ccs)))
260 (chise-turtle-format-ccs-code-point ccs ret)
263 (format (if est-hide-cgi-mode
264 "system-char-id=0x%X"
265 "system-char-id:0x%X")
266 (encode-char object 'system-char-id))
269 (defun isd-turtle-format-component (component separator level prefix)
270 (cond ((characterp component)
272 (isd-turtle-encode-char component)
277 (let ((ret (find-char component)))
280 (isd-turtle-encode-char ret) separator ret))
281 ((setq ret (assq 'ideographic-structure component))
282 (if (eq separator ?\;)
284 (isd-turtle-format-char nil nil (cdr ret) (1+ level)
286 (isd-turtle-format-char nil nil (cdr ret) (1+ level)
289 (defun isd-turtle-format-char (ccs code-point &optional ids-list level
290 prefix without-head-char)
295 (let ((indent (make-string (* level 4) ?\ ))
302 (if (and ccs code-point
303 (setq char (decode-char ccs code-point)))
304 (setq ids-list (get-char-attribute char 'ideographic-structure))))
305 (setq idc (car ids-list))
306 (setq c1 (nth 1 ids-list)
310 (setq idc (plist-get idc :char)))
312 (setq ret (find-char idc)))
315 (setq ret (find-char c1)))
318 (setq ret (find-char c2)))
321 (setq ret (find-char c3)))
324 ((eq idc ?\u2FF0) ; ⿰
342 ((memq idc '(?⿴ ?⿵ ?⿶ ?⿷ ?⿸ ?⿹ ?⿺))
350 ((and idc (eq (encode-char idc '=ucs-itaiji-001) #x2FF6))
355 ((and idc (eq (encode-char idc '=ucs-var-001) #x2FF0))
356 (setq idc-str "⿰・SLR")
360 ((and idc (eq (encode-char idc '=>iwds-1) 307))
365 ((and idc (eq (encode-char idc '=>iwds-1) 305))
370 ((and idc (eq (encode-char idc '=>ucs@component) #x2FF5))
379 %s %s:structure [ a idc:%s ;
384 (if without-head-char
386 (if (and ccs code-point)
388 (chise-turtle-format-ccs-code-point ccs code-point)
391 indent prefix (or idc-str (char-to-string idc))
392 indent prefix p1 (isd-turtle-format-component c1 ?\; (1+ level) prefix)
393 indent prefix p2 (isd-turtle-format-component c2 ?\; (1+ level) prefix)
394 indent prefix p3 (isd-turtle-format-component c3 ?\ (1+ level) prefix)
396 (if without-head-char
405 %s %s:structure [ a idc:%s ;
409 (if without-head-char
411 (if (and ccs code-point)
413 (chise-turtle-format-ccs-code-point ccs code-point)
416 indent prefix (or idc-str (char-to-string idc))
417 indent prefix p1 (isd-turtle-format-component c1 ?\; (1+ level) prefix)
418 indent prefix p2 (isd-turtle-format-component c2 ?\ (1+ level) prefix)
420 (if without-head-char
428 (defun isd-turtle-insert-char (ccs code-point)
429 (let ((ret (isd-turtle-format-char ccs code-point)))
434 (defun isd-turtle-insert-ccs-ranges (ccs &rest ranges)
435 (let (range code max-code)
437 (setq range (car ranges))
439 (setq code (car range)
440 max-code (cdr range))
441 (while (<= code max-code)
442 (isd-turtle-insert-char ccs code)
443 (setq code (1+ code)))
446 (isd-turtle-insert-char ccs range)
448 (t (error 'wrong-type-argument range)))
449 (setq ranges (cdr ranges)))))
451 (defun isd-turtle-dump-range (file path func &rest args)
453 (let ((coding-system-for-write 'utf-8-mcs-er)
454 ;; isd-turtle-ccs-list
455 chise-turtle-ccs-prefix-alist)
456 (if (file-directory-p path)
457 (setq path (expand-file-name file path)))
459 (goto-char (point-min))
460 ;; (dolist (ccs (sort isd-turtle-ccs-list
461 ;; #'char-attribute-name<))
462 ;; (insert (format "@prefix %s: <%s%s=> .\n"
463 ;; (isd-turtle-uri-encode-feature-name ccs)
464 ;; "http://www.chise.org/est/view/character/"
465 ;; (www-uri-encode-feature-name ccs))))
466 (dolist (cell (sort chise-turtle-ccs-prefix-alist
468 (char-attribute-name< (cdr a)(cdr b)))))
469 (insert (format "@prefix %s: <%s/%s=> .\n"
471 "http://www.chise.org/est/view/character"
472 (www-uri-encode-feature-name (cdr cell)))))
474 (goto-char (point-min))
475 (insert "# -*- coding: utf-8-mcs-er -*-\n")
476 (insert "@prefix : <http://rdf.chise.org/rdf/property/character/isd/> .
477 @prefix idc: <http://rdf.chise.org/rdf/type/character/idc/> .\n")
478 (write-region (point-min)(point-max) path))))
481 (defun isd-turtle-dump-ucs-basic (filename)
482 (interactive "Fdump ISD-UCS-Basic : ")
483 (isd-turtle-dump-range "ISD-UCS-Basic.ttl" filename
484 #'isd-turtle-insert-ccs-ranges
485 '=ucs '(#x4E00 . #x9FA5)))
488 (defun isd-turtle-dump-ucs-ext-a (filename)
489 (interactive "Fdump ISD-UCS-Ext-A : ")
490 (isd-turtle-dump-range "ISD-UCS-Ext-A.ttl" filename
491 #'isd-turtle-insert-ccs-ranges
492 '=ucs '(#x3400 . #x4DB5) #xFA1F #xFA23))
495 (defun isd-turtle-dump-ucs-ext-b-1 (filename)
496 (interactive "Fdump IDS-UCS-Ext-B-1 : ")
497 (isd-turtle-dump-range "ISD-UCS-Ext-B-1.ttl" filename
498 #'isd-turtle-insert-ccs-ranges
499 'ucs '(#x20000 . #x21FFF)))
502 (defun isd-turtle-dump-ucs-ext-b-2 (filename)
503 (interactive "Fdump IDS-UCS-Ext-B-2 : ")
504 (isd-turtle-dump-range "ISD-UCS-Ext-B-2.ttl" filename
505 #'isd-turtle-insert-ccs-ranges
506 'ucs '(#x22000 . #x23FFF)))
509 (defun isd-turtle-dump-ucs-ext-b-3 (filename)
510 (interactive "Fdump IDS-UCS-Ext-B-3 : ")
511 (isd-turtle-dump-range "ISD-UCS-Ext-B-3.ttl" filename
512 #'isd-turtle-insert-ccs-ranges
513 'ucs '(#x24000 . #x25FFF)))
516 (defun isd-turtle-dump-ucs-ext-b-4 (filename)
517 (interactive "Fdump IDS-UCS-Ext-B-4 : ")
518 (isd-turtle-dump-range "ISD-UCS-Ext-B-4.ttl" filename
519 #'isd-turtle-insert-ccs-ranges
520 'ucs '(#x26000 . #x27FFF)))
523 (defun isd-turtle-dump-ucs-ext-b-5 (filename)
524 (interactive "Fdump IDS-UCS-Ext-B-5 : ")
525 (isd-turtle-dump-range "ISD-UCS-Ext-B-5.ttl" filename
526 #'isd-turtle-insert-ccs-ranges
527 'ucs '(#x28000 . #x29FFF)))
530 (defun isd-turtle-dump-ucs-ext-b-6 (filename)
531 (interactive "Fdump IDS-UCS-Ext-B-6 : ")
532 (isd-turtle-dump-range "ISD-UCS-Ext-B-6.ttl" filename
533 #'isd-turtle-insert-ccs-ranges
534 'ucs '(#x2A000 . #x2A6D6)))
537 (defun isd-turtle-dump-ucs-ext-c (filename)
538 (interactive "Fdump IDS-UCS-Ext-C : ")
539 (isd-turtle-dump-range "ISD-UCS-Ext-C.ttl" filename
540 #'isd-turtle-insert-ccs-ranges
541 'ucs '(#x2A700 . #x2B734)))
544 (defun isd-turtle-dump-ucs-ext-d (filename)
545 (interactive "Fdump IDS-UCS-Ext-D : ")
546 (isd-turtle-dump-range "ISD-UCS-Ext-D.ttl" filename
547 #'isd-turtle-insert-ccs-ranges
548 'ucs '(#x2B740 . #x2B81D)))
551 (defun isd-turtle-dump-ucs-ext-e (filename)
552 (interactive "Fdump IDS-UCS-Ext-E : ")
553 (isd-turtle-dump-range "ISD-UCS-Ext-E.ttl" filename
554 #'isd-turtle-insert-ccs-ranges
555 'ucs '(#x2B820 . #x2CEA1)))
558 (defun isd-turtle-dump-mj-0 (filename)
559 (interactive "Fdump ISD-MJ-0 : ")
560 (isd-turtle-dump-range "ISD-MJ-0.ttl" filename
561 #'isd-turtle-insert-ccs-ranges
565 (defun isd-turtle-dump-mj-1 (filename)
566 (interactive "Fdump ISD-MJ-1 : ")
567 (isd-turtle-dump-range "ISD-MJ-1.ttl" filename
568 #'isd-turtle-insert-ccs-ranges
569 '=mj '(10000 . 19999)))
572 (defun isd-turtle-dump-mj-2 (filename)
573 (interactive "Fdump ISD-MJ-2 : ")
574 (isd-turtle-dump-range "ISD-MJ-2.ttl" filename
575 #'isd-turtle-insert-ccs-ranges
576 '=mj '(20000 . 29999)))
579 (defun isd-turtle-dump-mj-3 (filename)
580 (interactive "Fdump ISD-MJ-3 : ")
581 (isd-turtle-dump-range "ISD-MJ-3.ttl" filename
582 #'isd-turtle-insert-ccs-ranges
583 '=mj '(30000 . 39999)))
586 (defun isd-turtle-dump-mj-4 (filename)
587 (interactive "Fdump ISD-MJ-4 : ")
588 (isd-turtle-dump-range "ISD-MJ-4.ttl" filename
589 #'isd-turtle-insert-ccs-ranges
590 '=mj '(40000 . 49999)))
593 (defun isd-turtle-dump-mj-5 (filename)
594 (interactive "Fdump ISD-MJ-5 : ")
595 (isd-turtle-dump-range "ISD-MJ-5.ttl" filename
596 #'isd-turtle-insert-ccs-ranges
597 '=mj '(50000 . 59999)))
600 (defun isd-turtle-dump-mj-6 (filename)
601 (interactive "Fdump ISD-MJ-6 : ")
602 (isd-turtle-dump-range "ISD-MJ-6.ttl" filename
603 #'isd-turtle-insert-ccs-ranges
604 '=mj '(60000 . 69999)))
607 (defun isd-turtle-dump-all (directory)
608 (interactive "DISD directory : ")
609 (isd-turtle-dump-ucs-basic directory)
610 (isd-turtle-dump-ucs-ext-a directory)
611 (isd-turtle-dump-ucs-ext-b-1 directory)
612 (isd-turtle-dump-ucs-ext-b-2 directory)
613 (isd-turtle-dump-ucs-ext-b-3 directory)
614 (isd-turtle-dump-ucs-ext-b-4 directory)
615 (isd-turtle-dump-ucs-ext-b-5 directory)
616 (isd-turtle-dump-ucs-ext-b-6 directory)
617 (isd-turtle-dump-ucs-ext-c directory)
618 (isd-turtle-dump-ucs-ext-d directory)
619 (isd-turtle-dump-ucs-ext-e directory)
620 (isd-turtle-dump-mj-0 directory)
621 (isd-turtle-dump-mj-1 directory)
622 (isd-turtle-dump-mj-2 directory)
623 (isd-turtle-dump-mj-3 directory)
624 (isd-turtle-dump-mj-4 directory)
625 (isd-turtle-dump-mj-5 directory)
626 (isd-turtle-dump-mj-6 directory)
633 (provide 'isd-turtle)
635 ;;; isd-turtle.el ends here