X-Git-Url: http://git.chise.org/gitweb/?a=blobdiff_plain;f=ids-dump.el;h=c2e07988991fdef85331f6a286e7fa154ca6d676;hb=e13313e9957cb236f3823732cfa4cf36457f03fa;hp=03ada62d6033657959b12710f760fdc5c18afb69;hpb=233186c37f770a4f1b4c42d12cdbdf7f12519fd1;p=chise%2Fids.git

diff --git a/ids-dump.el b/ids-dump.el
index 03ada62..c2e0798 100644
--- a/ids-dump.el
+++ b/ids-dump.el
@@ -1,6 +1,6 @@
 ;;; ids-dump.el --- Dump utility of IDS-* files
 
-;; Copyright (C) 2002,2003 MORIOKA Tomohiko
+;; Copyright (C) 2002,2003,2004,2005,2009,2011,2019 MORIOKA Tomohiko
 
 ;; Author: MORIOKA Tomohiko <tomo@kanji.zinbun.kyoto-u.ac.jp>
 ;; Keywords: IDS, IDC, Ideographs, UCS, Unicode
@@ -26,6 +26,20 @@
 
 (require 'ids)
 
+(defun ids-dump-format-list (ids-list)
+  (if ids-list
+      (let (ucs)
+	(mapconcat
+	 (lambda (c)
+	   (char-to-string
+	    (if (setq ucs
+		      (unless (encode-char c '=ucs 'defined-only)
+			(or (get-char-attribute c '=ucs@unicode)
+			    (get-char-attribute c '=ucs@iso))))
+		(decode-char '=ucs ucs)
+	      c)))
+	 (ids-format-list ids-list) ""))))
+
 (defun ids-dump-insert-line (ccs line-spec code)
   (let ((chr (decode-char ccs code))
 	id-list)
@@ -34,7 +48,7 @@
       (insert (format line-spec
 		      code (decode-builtin-char ccs code)
 		      (if id-list
-			  (ids-format-list id-list)
+			  (ids-dump-format-list id-list)
 			(char-to-string chr)))))))
 
 (defun ids-dump-insert-ccs-ranges (ccs line-spec &rest ranges)
@@ -88,7 +102,7 @@
 	(insert
 	 (format "M-%05d \t%c\t%s\n"
 		 i (decode-builtin-char 'ideograph-daikanwa i)
-		 (or (ids-format-list
+		 (or (ids-dump-format-list
 		      (get-char-attribute chr 'ideographic-structure))
 		     ""))))
       (when (setq sal (assq i mdh-alist))
@@ -98,7 +112,7 @@
 	  (insert
 	   (format "M-%05d'\t%c\t%s\n"
 		   i chr
-		   (or (ids-format-list
+		   (or (ids-dump-format-list
 			(get-char-attribute chr 'ideographic-structure))
 		       ""))))
 	(when (setq chr (assq 2 sal))
@@ -106,7 +120,7 @@
 	  (insert
 	   (format "M-%05d\"\t%c\t%s\n"
 		   i chr
-		   (ids-format-list
+		   (ids-dump-format-list
 		    (get-char-attribute chr 'ideographic-structure)))))
 	)
       (setq i (1+ i)))))
@@ -127,7 +141,7 @@
        (format "MH-%04d \t%c\t%s\n"
 	       (car cell)
 	       chr
-	       (or (ids-format-list
+	       (or (ids-dump-format-list
 		    (get-char-attribute chr 'ideographic-structure))
 		   ""))))))
 
@@ -145,7 +159,7 @@
 		 h l
 		 (decode-builtin-char 'japanese-jisx0208-1990
 				      (logior (lsh h 8) l))
-		 (or (ids-format-list
+		 (or (ideographic-structure-to-ids
 		      (get-char-attribute chr 'ideographic-structure))
 		     "")))
 	(setq cell (1+ cell)))
@@ -160,7 +174,7 @@
 	       h l
 	       (decode-builtin-char 'japanese-jisx0208-1990
 				    (logior (lsh h 8) l))
-	       (or (ids-format-list
+	       (or (ideographic-structure-to-ids
 		    (get-char-attribute chr 'ideographic-structure))
 		   "")))
       (setq cell (1+ cell)))))
@@ -201,6 +215,48 @@
 	(setq l (1+ l)))
       (setq h (1+ h)))))
 
+(defun ids-dump-insert-big5-pua (ccs prefix)
+  (let ((line-spec (concat prefix "%04X\t%c\t%s\n"))
+	(h #x81)
+	l)
+    (while (<= h #xA0)
+      (setq l #x40)
+      (while (<= l #x7E)
+	(ids-dump-insert-line ccs line-spec (logior (lsh h 8) l))
+	(setq l (1+ l)))
+      (setq l #xA1)
+      (while (<= l #xFE)
+	(ids-dump-insert-line ccs line-spec (logior (lsh h 8) l))
+	(setq l (1+ l)))
+      (setq h (1+ h)))
+    (setq h #xC6)
+    (setq l #xDE)
+    (while (<= l #xFE)
+      (ids-dump-insert-line ccs line-spec (logior (lsh h 8) l))
+      (setq l (1+ l)))
+    (setq h #xC7)
+    (while (<= h #xC8)
+      (setq l #x40)
+      (while (<= l #x7E)
+	(ids-dump-insert-line ccs line-spec (logior (lsh h 8) l))
+	(setq l (1+ l)))
+      (setq l #xA1)
+      (while (<= l #xFE)
+	(ids-dump-insert-line ccs line-spec (logior (lsh h 8) l))
+	(setq l (1+ l)))
+      (setq h (1+ h)))
+    (setq h #xFA)
+    (while (<= h #xFE)
+      (setq l #x40)
+      (while (<= l #x7E)
+	(ids-dump-insert-line ccs line-spec (logior (lsh h 8) l))
+	(setq l (1+ l)))
+      (setq l #xA1)
+      (while (<= l #xFE)
+	(ids-dump-insert-line ccs line-spec (logior (lsh h 8) l))
+	(setq l (1+ l)))
+      (setq h (1+ h)))))
+
 (defun ids-dump-range (file path func &rest args)
   (with-temp-buffer
     (let* ((coding-system-for-write 'utf-8-mcs-er))
@@ -215,6 +271,14 @@
   (interactive "Fdump IDS-UCS-Basic : ")
   (ids-dump-range "IDS-UCS-Basic.txt" filename
 		  #'ids-dump-insert-ccs-ranges 'ucs "U+%04X\t%c\t%s\n"
+		  '(#x4E00 . #x9FEA)))
+
+;;;###autoload
+(defun ids-dump-ucs-basic@unicode (filename)
+  (interactive "Fdump IDS-UCS-Basic : ")
+  (ids-dump-range "IDS-UCS-Basic_u.txt" filename
+		  #'ids-dump-insert-ccs-ranges '=ucs@unicode
+		  "UU+%04X\t%c\t%s\n"
 		  '(#x4E00 . #x9FA5)))
 
 ;;;###autoload
@@ -406,7 +470,7 @@
 (defun ids-dump-big5-cdp (filename)
   (interactive "Fdump IDS-CDP : ")
   (ids-dump-range "IDS-CDP.txt" filename
-		  #'ids-dump-insert-big5
+		  #'ids-dump-insert-big5-pua
 		  '=big5-cdp "CDP-"))