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
63 =cns11643-1 =cns11643-2 =cns11643-3
64 =cns11643-4 =cns11643-5 =cns11643-6 =cns11643-7
74 =zinbun-oracle =>zinbun-oracle
79 =+>ucs@iso =+>ucs@unicode
85 =>jis-x0208 =>jis-x0213-1
92 =>ucs@jis =>ucs@cns =>ucs@ks
95 =>>ucs@iso =>>ucs@unicode
96 =>>ucs@jis =>>ucs@cns =>>ucs@ks
114 ==cns11643-1 ==cns11643-2 ==cns11643-3
115 ==cns11643-4 ==cns11643-5 ==cns11643-6 ==cns11643-7
125 =>>jis-x0208 =>>jis-x0213-1 =>>jis-x0213-2
126 =+>jis-x0208 =+>jis-x0213-1 =+>jis-x0213-2
132 =jis-x0208@1983 =jis-x0208@1978
162 ;; (defvar isd-turtle-ccs-list nil)
163 (defvar chise-turtle-ccs-prefix-alist nil)
165 (defun charset-code-point-format-spec (ccs)
166 (cond ((memq ccs '(=ucs))
169 (let ((ccs-name (symbol-name ccs)))
172 "\\(shinjigen\\|daikanwa/ho\\|=>iwds-1\\)"
176 "\\(gt\\|daikanwa\\|adobe-japan1\\|cbeta\\|zinbun-oracle\\|hng\\)"
179 ((string-match "\\(hanyo-denshi/ks\\|koseki\\|mj\\)" ccs-name)
181 ((string-match "hanyo-denshi/tk" ccs-name)
186 ;; (defun isd-turtle-uri-encode-feature-name (feature-name)
188 ;; ((eq '=ucs feature-name)
190 ;; ((eq '==>ucs@bucs feature-name)
193 ;; (mapconcat (lambda (c)
196 ;; (char-to-string c)))
197 ;; (www-uri-encode-feature-name feature-name)
199 (defun chise-turtle-uri-encode-ccs-name (feature-name)
201 ((eq '=ucs feature-name)
203 ((eq '=big5 feature-name)
205 ((eq '==>ucs@bucs feature-name)
208 (mapconcat (lambda (c)
217 (char-to-string c))))
218 (www-uri-encode-feature-name feature-name)
221 ;; (defun isd-turtle-format-ccs-code-point (ccs code-point)
222 ;; (unless (memq ccs isd-turtle-ccs-list)
223 ;; (setq isd-turtle-ccs-list (cons ccs isd-turtle-ccs-list)))
225 ;; (isd-turtle-uri-encode-feature-name ccs)
226 ;; (format (charset-code-point-format-spec ccs)
228 (defun chise-turtle-format-ccs-code-point (ccs code-point)
229 (let ((ccs-uri (chise-turtle-uri-encode-ccs-name ccs)))
230 (unless (assoc ccs-uri chise-turtle-ccs-prefix-alist)
231 (setq chise-turtle-ccs-prefix-alist
232 (cons (cons ccs-uri ccs)
233 chise-turtle-ccs-prefix-alist)))
236 (format (charset-code-point-format-spec ccs)
239 (defun isd-turtle-encode-char (object)
240 (let ((ccs-list est-coded-charset-priority-list)
242 (if (setq ret (encode-char object '=ucs))
243 (chise-turtle-format-ccs-code-point '=ucs ret)
245 (setq ccs (pop ccs-list))
246 (not (setq ret (encode-char object ccs 'defined-only)))))
248 (chise-turtle-format-ccs-code-point ccs ret)
250 ((and (setq ccs (car (split-char object)))
251 (setq ret (encode-char object ccs)))
252 (chise-turtle-format-ccs-code-point ccs ret)
255 (format (if est-hide-cgi-mode
256 "system-char-id=0x%X"
257 "system-char-id:0x%X")
258 (encode-char object 'system-char-id))
261 (defun isd-turtle-format-component (component separator level prefix)
262 (cond ((characterp component)
264 (isd-turtle-encode-char component)
269 (let ((ret (find-char component)))
272 (isd-turtle-encode-char ret) separator ret))
273 ((setq ret (assq 'ideographic-structure component))
274 (if (eq separator ?\;)
276 (isd-turtle-format-char nil nil (cdr ret) (1+ level)
278 (isd-turtle-format-char nil nil (cdr ret) (1+ level)
281 (defun isd-turtle-format-char (ccs code-point &optional ids-list level
282 prefix without-head-char)
287 (let ((indent (make-string (* level 4) ?\ ))
294 (if (and ccs code-point
295 (setq char (decode-char ccs code-point)))
296 (setq ids-list (get-char-attribute char 'ideographic-structure))))
297 (setq idc (car ids-list))
298 (setq c1 (nth 1 ids-list)
302 (setq idc (plist-get idc :char)))
304 (setq ret (find-char idc)))
307 (setq ret (find-char c1)))
310 (setq ret (find-char c2)))
313 (setq ret (find-char c3)))
316 ((eq idc ?\u2FF0) ; ⿰
334 ((memq idc '(?⿴ ?⿵ ?⿶ ?⿷ ?⿸ ?⿹ ?⿺))
345 %s %s:structure [ a idc:%c ;
350 (if without-head-char
352 (if (and ccs code-point)
354 (chise-turtle-format-ccs-code-point ccs code-point)
358 indent prefix p1 (isd-turtle-format-component c1 ?\; (1+ level) prefix)
359 indent prefix p2 (isd-turtle-format-component c2 ?\; (1+ level) prefix)
360 indent prefix p3 (isd-turtle-format-component c3 ?\ (1+ level) prefix)
362 (if without-head-char
371 %s %s:structure [ a idc:%c ;
375 (if without-head-char
377 (if (and ccs code-point)
379 (chise-turtle-format-ccs-code-point ccs code-point)
383 indent prefix p1 (isd-turtle-format-component c1 ?\; (1+ level) prefix)
384 indent prefix p2 (isd-turtle-format-component c2 ?\ (1+ level) prefix)
386 (if without-head-char
394 (defun isd-turtle-insert-char (ccs code-point)
395 (let ((ret (isd-turtle-format-char ccs code-point)))
400 (defun isd-turtle-insert-ccs-ranges (ccs &rest ranges)
401 (let (range code max-code)
403 (setq range (car ranges))
405 (setq code (car range)
406 max-code (cdr range))
407 (while (<= code max-code)
408 (isd-turtle-insert-char ccs code)
409 (setq code (1+ code)))
412 (isd-turtle-insert-char ccs range)
414 (t (error 'wrong-type-argument range)))
415 (setq ranges (cdr ranges)))))
417 (defun isd-turtle-dump-range (file path func &rest args)
419 (let ((coding-system-for-write 'utf-8-mcs-er)
420 ;; isd-turtle-ccs-list
421 chise-turtle-ccs-prefix-alist)
422 (if (file-directory-p path)
423 (setq path (expand-file-name file path)))
425 (goto-char (point-min))
426 ;; (dolist (ccs (sort isd-turtle-ccs-list
427 ;; #'char-attribute-name<))
428 ;; (insert (format "@prefix %s: <%s%s=> .\n"
429 ;; (isd-turtle-uri-encode-feature-name ccs)
430 ;; "http://www.chise.org/est/view/character/"
431 ;; (www-uri-encode-feature-name ccs))))
432 (dolist (cell (sort chise-turtle-ccs-prefix-alist
434 (char-attribute-name< (cdr a)(cdr b)))))
435 (insert (format "@prefix %s: <%s/%s=> .\n"
437 "http://www.chise.org/est/view/character"
438 (www-uri-encode-feature-name (cdr cell)))))
440 (goto-char (point-min))
441 (insert "# -*- coding: utf-8-mcs-er -*-\n")
442 (insert "@prefix : <http://rdf.chise.org/rdf/property/character/isd/> .
443 @prefix idc: <http://rdf.chise.org/rdf/type/character/idc/> .\n")
444 (write-region (point-min)(point-max) path))))
447 (defun isd-turtle-dump-ucs-basic (filename)
448 (interactive "Fdump ISD-UCS-Basic : ")
449 (isd-turtle-dump-range "ISD-UCS-Basic.ttl" filename
450 #'isd-turtle-insert-ccs-ranges
451 '=ucs '(#x4E00 . #x9FA5)))
454 (defun isd-turtle-dump-ucs-ext-a (filename)
455 (interactive "Fdump ISD-UCS-Ext-A : ")
456 (isd-turtle-dump-range "ISD-UCS-Ext-A.ttl" filename
457 #'isd-turtle-insert-ccs-ranges
458 '=ucs '(#x3400 . #x4DB5) #xFA1F #xFA23))
461 (defun isd-turtle-dump-ucs-ext-b-1 (filename)
462 (interactive "Fdump IDS-UCS-Ext-B-1 : ")
463 (isd-turtle-dump-range "ISD-UCS-Ext-B-1.ttl" filename
464 #'isd-turtle-insert-ccs-ranges
465 'ucs '(#x20000 . #x21FFF)))
468 (defun isd-turtle-dump-ucs-ext-b-2 (filename)
469 (interactive "Fdump IDS-UCS-Ext-B-2 : ")
470 (isd-turtle-dump-range "ISD-UCS-Ext-B-2.ttl" filename
471 #'isd-turtle-insert-ccs-ranges
472 'ucs '(#x22000 . #x23FFF)))
475 (defun isd-turtle-dump-ucs-ext-b-3 (filename)
476 (interactive "Fdump IDS-UCS-Ext-B-3 : ")
477 (isd-turtle-dump-range "ISD-UCS-Ext-B-3.ttl" filename
478 #'isd-turtle-insert-ccs-ranges
479 'ucs '(#x24000 . #x25FFF)))
482 (defun isd-turtle-dump-ucs-ext-b-4 (filename)
483 (interactive "Fdump IDS-UCS-Ext-B-4 : ")
484 (isd-turtle-dump-range "ISD-UCS-Ext-B-4.ttl" filename
485 #'isd-turtle-insert-ccs-ranges
486 'ucs '(#x26000 . #x27FFF)))
489 (defun isd-turtle-dump-ucs-ext-b-5 (filename)
490 (interactive "Fdump IDS-UCS-Ext-B-5 : ")
491 (isd-turtle-dump-range "ISD-UCS-Ext-B-5.ttl" filename
492 #'isd-turtle-insert-ccs-ranges
493 'ucs '(#x28000 . #x29FFF)))
496 (defun isd-turtle-dump-ucs-ext-b-6 (filename)
497 (interactive "Fdump IDS-UCS-Ext-B-6 : ")
498 (isd-turtle-dump-range "ISD-UCS-Ext-B-6.ttl" filename
499 #'isd-turtle-insert-ccs-ranges
500 'ucs '(#x2A000 . #x2A6D6)))
503 (defun isd-turtle-dump-ucs-ext-c (filename)
504 (interactive "Fdump IDS-UCS-Ext-C : ")
505 (isd-turtle-dump-range "ISD-UCS-Ext-C.ttl" filename
506 #'isd-turtle-insert-ccs-ranges
507 'ucs '(#x2A700 . #x2B734)))
510 (defun isd-turtle-dump-ucs-ext-d (filename)
511 (interactive "Fdump IDS-UCS-Ext-D : ")
512 (isd-turtle-dump-range "ISD-UCS-Ext-D.ttl" filename
513 #'isd-turtle-insert-ccs-ranges
514 'ucs '(#x2B740 . #x2B81D)))
517 (defun isd-turtle-dump-ucs-ext-e (filename)
518 (interactive "Fdump IDS-UCS-Ext-E : ")
519 (isd-turtle-dump-range "ISD-UCS-Ext-E.ttl" filename
520 #'isd-turtle-insert-ccs-ranges
521 'ucs '(#x2B820 . #x2CEA1)))
524 (defun isd-turtle-dump-mj-0 (filename)
525 (interactive "Fdump ISD-MJ-0 : ")
526 (isd-turtle-dump-range "ISD-MJ-0.ttl" filename
527 #'isd-turtle-insert-ccs-ranges
531 (defun isd-turtle-dump-mj-1 (filename)
532 (interactive "Fdump ISD-MJ-1 : ")
533 (isd-turtle-dump-range "ISD-MJ-1.ttl" filename
534 #'isd-turtle-insert-ccs-ranges
535 '=mj '(10000 . 19999)))
538 (defun isd-turtle-dump-mj-2 (filename)
539 (interactive "Fdump ISD-MJ-2 : ")
540 (isd-turtle-dump-range "ISD-MJ-2.ttl" filename
541 #'isd-turtle-insert-ccs-ranges
542 '=mj '(20000 . 29999)))
545 (defun isd-turtle-dump-mj-3 (filename)
546 (interactive "Fdump ISD-MJ-3 : ")
547 (isd-turtle-dump-range "ISD-MJ-3.ttl" filename
548 #'isd-turtle-insert-ccs-ranges
549 '=mj '(30000 . 39999)))
552 (defun isd-turtle-dump-mj-4 (filename)
553 (interactive "Fdump ISD-MJ-4 : ")
554 (isd-turtle-dump-range "ISD-MJ-4.ttl" filename
555 #'isd-turtle-insert-ccs-ranges
556 '=mj '(40000 . 49999)))
559 (defun isd-turtle-dump-mj-5 (filename)
560 (interactive "Fdump ISD-MJ-5 : ")
561 (isd-turtle-dump-range "ISD-MJ-5.ttl" filename
562 #'isd-turtle-insert-ccs-ranges
563 '=mj '(50000 . 59999)))
566 (defun isd-turtle-dump-mj-6 (filename)
567 (interactive "Fdump ISD-MJ-6 : ")
568 (isd-turtle-dump-range "ISD-MJ-6.ttl" filename
569 #'isd-turtle-insert-ccs-ranges
570 '=mj '(60000 . 69999)))
573 (defun isd-turtle-dump-all (directory)
574 (interactive "DISD directory : ")
575 (isd-turtle-dump-ucs-basic directory)
576 (isd-turtle-dump-ucs-ext-a directory)
577 (isd-turtle-dump-ucs-ext-b-1 directory)
578 (isd-turtle-dump-ucs-ext-b-2 directory)
579 (isd-turtle-dump-ucs-ext-b-3 directory)
580 (isd-turtle-dump-ucs-ext-b-4 directory)
581 (isd-turtle-dump-ucs-ext-b-5 directory)
582 (isd-turtle-dump-ucs-ext-b-6 directory)
583 (isd-turtle-dump-ucs-ext-c directory)
584 (isd-turtle-dump-ucs-ext-d directory)
585 (isd-turtle-dump-ucs-ext-e directory)
586 (isd-turtle-dump-mj-0 directory)
587 (isd-turtle-dump-mj-1 directory)
588 (isd-turtle-dump-mj-2 directory)
589 (isd-turtle-dump-mj-3 directory)
590 (isd-turtle-dump-mj-4 directory)
591 (isd-turtle-dump-mj-5 directory)
592 (isd-turtle-dump-mj-6 directory)
599 (provide 'isd-turtle)
601 ;;; isd-turtle.el ends here