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
65 =cns11643-1 =cns11643-2 =cns11643-3
66 =cns11643-4 =cns11643-5 =cns11643-6 =cns11643-7
76 =zinbun-oracle =>zinbun-oracle
81 =+>ucs@iso =+>ucs@unicode
87 =>jis-x0208 =>jis-x0213-1
95 =>ucs@jis =>ucs@cns =>ucs@ks
98 =>>ucs@iso =>>ucs@unicode
99 =>>ucs@jis =>>ucs@cns =>>ucs@ks
117 ==cns11643-1 ==cns11643-2 ==cns11643-3
118 ==cns11643-4 ==cns11643-5 ==cns11643-6 ==cns11643-7
128 =>>jis-x0208 =>>jis-x0213-1 =>>jis-x0213-2
129 =+>jis-x0208 =+>jis-x0213-1 =+>jis-x0213-2
135 =jis-x0208@1983 =jis-x0208@1978
168 ;; (defvar isd-turtle-ccs-list nil)
169 (defvar chise-turtle-ccs-prefix-alist nil)
171 (defun charset-code-point-format-spec (ccs)
172 (cond ((memq ccs '(=ucs))
175 (let ((ccs-name (symbol-name ccs)))
178 "\\(shinjigen\\|daikanwa/ho\\|=>iwds-1\\)"
182 "\\(gt\\|daikanwa\\|adobe-japan1\\|cbeta\\|zinbun-oracle\\|hng\\)"
185 ((string-match "\\(hanyo-denshi/ks\\|koseki\\|mj\\)" ccs-name)
187 ((string-match "hanyo-denshi/tk" ccs-name)
192 ;; (defun isd-turtle-uri-encode-feature-name (feature-name)
194 ;; ((eq '=ucs feature-name)
196 ;; ((eq '==>ucs@bucs feature-name)
199 ;; (mapconcat (lambda (c)
202 ;; (char-to-string c)))
203 ;; (www-uri-encode-feature-name feature-name)
205 (defun chise-turtle-uri-encode-ccs-name (feature-name)
207 ((eq '=ucs feature-name)
209 ((eq '=big5 feature-name)
211 ((eq '==>ucs@bucs feature-name)
214 (mapconcat (lambda (c)
223 (char-to-string c))))
224 (www-uri-encode-feature-name feature-name)
227 ;; (defun isd-turtle-format-ccs-code-point (ccs code-point)
228 ;; (unless (memq ccs isd-turtle-ccs-list)
229 ;; (setq isd-turtle-ccs-list (cons ccs isd-turtle-ccs-list)))
231 ;; (isd-turtle-uri-encode-feature-name ccs)
232 ;; (format (charset-code-point-format-spec ccs)
234 (defun chise-turtle-format-ccs-code-point (ccs code-point)
235 (let ((ccs-uri (chise-turtle-uri-encode-ccs-name ccs)))
236 (unless (assoc ccs-uri chise-turtle-ccs-prefix-alist)
237 (setq chise-turtle-ccs-prefix-alist
238 (cons (cons ccs-uri ccs)
239 chise-turtle-ccs-prefix-alist)))
242 (format (charset-code-point-format-spec ccs)
245 (defun isd-turtle-encode-char (object)
246 (let ((ccs-list est-coded-charset-priority-list)
248 (if (setq ret (encode-char object '=ucs))
249 (chise-turtle-format-ccs-code-point '=ucs ret)
251 (setq ccs (pop ccs-list))
252 (not (setq ret (encode-char object ccs 'defined-only)))))
254 (chise-turtle-format-ccs-code-point ccs ret)
256 ((and (setq ccs (car (split-char object)))
257 (setq ret (encode-char object ccs)))
258 (chise-turtle-format-ccs-code-point ccs ret)
261 (format (if est-hide-cgi-mode
262 "system-char-id=0x%X"
263 "system-char-id:0x%X")
264 (encode-char object 'system-char-id))
267 (defun isd-turtle-format-component (component separator level prefix)
268 (cond ((characterp component)
270 (isd-turtle-encode-char component)
275 (let ((ret (find-char component)))
278 (isd-turtle-encode-char ret) separator ret))
279 ((setq ret (assq 'ideographic-structure component))
280 (if (eq separator ?\;)
282 (isd-turtle-format-char nil nil (cdr ret) (1+ level)
284 (isd-turtle-format-char nil nil (cdr ret) (1+ level)
287 (defun isd-turtle-format-char (ccs code-point &optional ids-list level
288 prefix without-head-char)
293 (let ((indent (make-string (* level 4) ?\ ))
300 (if (and ccs code-point
301 (setq char (decode-char ccs code-point)))
302 (setq ids-list (get-char-attribute char 'ideographic-structure))))
303 (setq idc (car ids-list))
304 (setq c1 (nth 1 ids-list)
308 (setq idc (plist-get idc :char)))
310 (setq ret (find-char idc)))
313 (setq ret (find-char c1)))
316 (setq ret (find-char c2)))
319 (setq ret (find-char c3)))
322 ((eq idc ?\u2FF0) ; ⿰
340 ((memq idc '(?⿴ ?⿵ ?⿶ ?⿷ ?⿸ ?⿹ ?⿺))
351 %s %s:structure [ a idc:%c ;
356 (if without-head-char
358 (if (and ccs code-point)
360 (chise-turtle-format-ccs-code-point ccs code-point)
364 indent prefix p1 (isd-turtle-format-component c1 ?\; (1+ level) prefix)
365 indent prefix p2 (isd-turtle-format-component c2 ?\; (1+ level) prefix)
366 indent prefix p3 (isd-turtle-format-component c3 ?\ (1+ level) prefix)
368 (if without-head-char
377 %s %s:structure [ a idc:%c ;
381 (if without-head-char
383 (if (and ccs code-point)
385 (chise-turtle-format-ccs-code-point ccs code-point)
389 indent prefix p1 (isd-turtle-format-component c1 ?\; (1+ level) prefix)
390 indent prefix p2 (isd-turtle-format-component c2 ?\ (1+ level) prefix)
392 (if without-head-char
400 (defun isd-turtle-insert-char (ccs code-point)
401 (let ((ret (isd-turtle-format-char ccs code-point)))
406 (defun isd-turtle-insert-ccs-ranges (ccs &rest ranges)
407 (let (range code max-code)
409 (setq range (car ranges))
411 (setq code (car range)
412 max-code (cdr range))
413 (while (<= code max-code)
414 (isd-turtle-insert-char ccs code)
415 (setq code (1+ code)))
418 (isd-turtle-insert-char ccs range)
420 (t (error 'wrong-type-argument range)))
421 (setq ranges (cdr ranges)))))
423 (defun isd-turtle-dump-range (file path func &rest args)
425 (let ((coding-system-for-write 'utf-8-mcs-er)
426 ;; isd-turtle-ccs-list
427 chise-turtle-ccs-prefix-alist)
428 (if (file-directory-p path)
429 (setq path (expand-file-name file path)))
431 (goto-char (point-min))
432 ;; (dolist (ccs (sort isd-turtle-ccs-list
433 ;; #'char-attribute-name<))
434 ;; (insert (format "@prefix %s: <%s%s=> .\n"
435 ;; (isd-turtle-uri-encode-feature-name ccs)
436 ;; "http://www.chise.org/est/view/character/"
437 ;; (www-uri-encode-feature-name ccs))))
438 (dolist (cell (sort chise-turtle-ccs-prefix-alist
440 (char-attribute-name< (cdr a)(cdr b)))))
441 (insert (format "@prefix %s: <%s/%s=> .\n"
443 "http://www.chise.org/est/view/character"
444 (www-uri-encode-feature-name (cdr cell)))))
446 (goto-char (point-min))
447 (insert "# -*- coding: utf-8-mcs-er -*-\n")
448 (insert "@prefix : <http://rdf.chise.org/rdf/property/character/isd/> .
449 @prefix idc: <http://rdf.chise.org/rdf/type/character/idc/> .\n")
450 (write-region (point-min)(point-max) path))))
453 (defun isd-turtle-dump-ucs-basic (filename)
454 (interactive "Fdump ISD-UCS-Basic : ")
455 (isd-turtle-dump-range "ISD-UCS-Basic.ttl" filename
456 #'isd-turtle-insert-ccs-ranges
457 '=ucs '(#x4E00 . #x9FA5)))
460 (defun isd-turtle-dump-ucs-ext-a (filename)
461 (interactive "Fdump ISD-UCS-Ext-A : ")
462 (isd-turtle-dump-range "ISD-UCS-Ext-A.ttl" filename
463 #'isd-turtle-insert-ccs-ranges
464 '=ucs '(#x3400 . #x4DB5) #xFA1F #xFA23))
467 (defun isd-turtle-dump-ucs-ext-b-1 (filename)
468 (interactive "Fdump IDS-UCS-Ext-B-1 : ")
469 (isd-turtle-dump-range "ISD-UCS-Ext-B-1.ttl" filename
470 #'isd-turtle-insert-ccs-ranges
471 'ucs '(#x20000 . #x21FFF)))
474 (defun isd-turtle-dump-ucs-ext-b-2 (filename)
475 (interactive "Fdump IDS-UCS-Ext-B-2 : ")
476 (isd-turtle-dump-range "ISD-UCS-Ext-B-2.ttl" filename
477 #'isd-turtle-insert-ccs-ranges
478 'ucs '(#x22000 . #x23FFF)))
481 (defun isd-turtle-dump-ucs-ext-b-3 (filename)
482 (interactive "Fdump IDS-UCS-Ext-B-3 : ")
483 (isd-turtle-dump-range "ISD-UCS-Ext-B-3.ttl" filename
484 #'isd-turtle-insert-ccs-ranges
485 'ucs '(#x24000 . #x25FFF)))
488 (defun isd-turtle-dump-ucs-ext-b-4 (filename)
489 (interactive "Fdump IDS-UCS-Ext-B-4 : ")
490 (isd-turtle-dump-range "ISD-UCS-Ext-B-4.ttl" filename
491 #'isd-turtle-insert-ccs-ranges
492 'ucs '(#x26000 . #x27FFF)))
495 (defun isd-turtle-dump-ucs-ext-b-5 (filename)
496 (interactive "Fdump IDS-UCS-Ext-B-5 : ")
497 (isd-turtle-dump-range "ISD-UCS-Ext-B-5.ttl" filename
498 #'isd-turtle-insert-ccs-ranges
499 'ucs '(#x28000 . #x29FFF)))
502 (defun isd-turtle-dump-ucs-ext-b-6 (filename)
503 (interactive "Fdump IDS-UCS-Ext-B-6 : ")
504 (isd-turtle-dump-range "ISD-UCS-Ext-B-6.ttl" filename
505 #'isd-turtle-insert-ccs-ranges
506 'ucs '(#x2A000 . #x2A6D6)))
509 (defun isd-turtle-dump-ucs-ext-c (filename)
510 (interactive "Fdump IDS-UCS-Ext-C : ")
511 (isd-turtle-dump-range "ISD-UCS-Ext-C.ttl" filename
512 #'isd-turtle-insert-ccs-ranges
513 'ucs '(#x2A700 . #x2B734)))
516 (defun isd-turtle-dump-ucs-ext-d (filename)
517 (interactive "Fdump IDS-UCS-Ext-D : ")
518 (isd-turtle-dump-range "ISD-UCS-Ext-D.ttl" filename
519 #'isd-turtle-insert-ccs-ranges
520 'ucs '(#x2B740 . #x2B81D)))
523 (defun isd-turtle-dump-ucs-ext-e (filename)
524 (interactive "Fdump IDS-UCS-Ext-E : ")
525 (isd-turtle-dump-range "ISD-UCS-Ext-E.ttl" filename
526 #'isd-turtle-insert-ccs-ranges
527 'ucs '(#x2B820 . #x2CEA1)))
530 (defun isd-turtle-dump-mj-0 (filename)
531 (interactive "Fdump ISD-MJ-0 : ")
532 (isd-turtle-dump-range "ISD-MJ-0.ttl" filename
533 #'isd-turtle-insert-ccs-ranges
537 (defun isd-turtle-dump-mj-1 (filename)
538 (interactive "Fdump ISD-MJ-1 : ")
539 (isd-turtle-dump-range "ISD-MJ-1.ttl" filename
540 #'isd-turtle-insert-ccs-ranges
541 '=mj '(10000 . 19999)))
544 (defun isd-turtle-dump-mj-2 (filename)
545 (interactive "Fdump ISD-MJ-2 : ")
546 (isd-turtle-dump-range "ISD-MJ-2.ttl" filename
547 #'isd-turtle-insert-ccs-ranges
548 '=mj '(20000 . 29999)))
551 (defun isd-turtle-dump-mj-3 (filename)
552 (interactive "Fdump ISD-MJ-3 : ")
553 (isd-turtle-dump-range "ISD-MJ-3.ttl" filename
554 #'isd-turtle-insert-ccs-ranges
555 '=mj '(30000 . 39999)))
558 (defun isd-turtle-dump-mj-4 (filename)
559 (interactive "Fdump ISD-MJ-4 : ")
560 (isd-turtle-dump-range "ISD-MJ-4.ttl" filename
561 #'isd-turtle-insert-ccs-ranges
562 '=mj '(40000 . 49999)))
565 (defun isd-turtle-dump-mj-5 (filename)
566 (interactive "Fdump ISD-MJ-5 : ")
567 (isd-turtle-dump-range "ISD-MJ-5.ttl" filename
568 #'isd-turtle-insert-ccs-ranges
569 '=mj '(50000 . 59999)))
572 (defun isd-turtle-dump-mj-6 (filename)
573 (interactive "Fdump ISD-MJ-6 : ")
574 (isd-turtle-dump-range "ISD-MJ-6.ttl" filename
575 #'isd-turtle-insert-ccs-ranges
576 '=mj '(60000 . 69999)))
579 (defun isd-turtle-dump-all (directory)
580 (interactive "DISD directory : ")
581 (isd-turtle-dump-ucs-basic directory)
582 (isd-turtle-dump-ucs-ext-a directory)
583 (isd-turtle-dump-ucs-ext-b-1 directory)
584 (isd-turtle-dump-ucs-ext-b-2 directory)
585 (isd-turtle-dump-ucs-ext-b-3 directory)
586 (isd-turtle-dump-ucs-ext-b-4 directory)
587 (isd-turtle-dump-ucs-ext-b-5 directory)
588 (isd-turtle-dump-ucs-ext-b-6 directory)
589 (isd-turtle-dump-ucs-ext-c directory)
590 (isd-turtle-dump-ucs-ext-d directory)
591 (isd-turtle-dump-ucs-ext-e directory)
592 (isd-turtle-dump-mj-0 directory)
593 (isd-turtle-dump-mj-1 directory)
594 (isd-turtle-dump-mj-2 directory)
595 (isd-turtle-dump-mj-3 directory)
596 (isd-turtle-dump-mj-4 directory)
597 (isd-turtle-dump-mj-5 directory)
598 (isd-turtle-dump-mj-6 directory)
605 (provide 'isd-turtle)
607 ;;; isd-turtle.el ends here