X-Git-Url: http://git.chise.org/gitweb/?a=blobdiff_plain;f=lisp%2Futf-2000%2Fchar-db-util.el;h=290e69459fdd3242bb1d50e1eb476a125278ed5a;hb=c17f21f4cc9fd196a4433709e74f0ea00acede7e;hp=ff807744e3258e279425dfb132c6dcf9f4d7b50a;hpb=56d1227c2c03af55779b36687d23df7bd1396db3;p=chise%2Fxemacs-chise.git

diff --git a/lisp/utf-2000/char-db-util.el b/lisp/utf-2000/char-db-util.el
index ff80774..290e694 100644
--- a/lisp/utf-2000/char-db-util.el
+++ b/lisp/utf-2000/char-db-util.el
@@ -1,6 +1,6 @@
 ;;; char-db-util.el --- Character Database utility
 
-;; Copyright (C) 1998,1999,2000,2001,2002,2003,2004 MORIOKA Tomohiko.
+;; Copyright (C) 1998,1999,2000,2001,2002,2003,2004,2005 MORIOKA Tomohiko.
 
 ;; Author: MORIOKA Tomohiko <tomo@kanji.zinbun.kyoto-u.ac.jp>
 ;; Keywords: CHISE, Character Database, ISO/IEC 10646, Unicode, UCS-4, MULE.
@@ -75,7 +75,7 @@
   '(ucs daikanwa cns gt jis jis/alt jis/a jis/b
 	jis-x0212 jis-x0213 cdp shinjigen misc unknown))
 
-(defvar char-db-ignored-attributes nil)
+(defvar char-db-ignored-attributes '(ideographic-products))
 
 (defun char-attribute-name< (ka kb)
   (cond
@@ -127,9 +127,9 @@
     greek-iso8859-7
     thai-tis620
     =jis-x0208
-    japanese-jisx0208
+    =jis-x0208@1978
+    =jis-x0208@1983
     japanese-jisx0212
-    japanese-jisx0208-1978
     chinese-gb2312
     chinese-cns11643-1
     chinese-cns11643-2
@@ -138,7 +138,7 @@
     chinese-cns11643-5
     chinese-cns11643-6
     chinese-cns11643-7
-    =jis-x0208-1990
+    =jis-x0208@1990
     =jis-x0213-1-2000
     =jis-x0213-2-2000
     korean-ksc5601
@@ -165,9 +165,11 @@
     ideograph-hanziku-10
     ideograph-hanziku-11
     ideograph-hanziku-12
+    =gt-k
+    =ucs@iso
+    =ucs@unicode
     =big5
     =big5-eten
-    =gt-k
     =jis-x0208@1997
     =jef-china3))
 
@@ -204,6 +206,8 @@
 					 =daikanwa@rev2
 					 ;; =gt-k
 					 )))
+			     (setq ccs (charset-name ccs))
+			     (null (assq ccs char-spec))
 			     (setq ret (encode-char char ccs 'defined-only)))
 			(setq char-spec (cons (cons ccs ret) char-spec))))
 		  (if (null char-spec)
@@ -620,7 +624,19 @@
       (setq attributes (delq 'ideographic-radical attributes))
       )
     (let (key)
-      (dolist (domain char-db-feature-domains)
+      (dolist (domain
+	       (append
+		char-db-feature-domains
+		(let (dest domain)
+		  (dolist (feature (char-attribute-list))
+		    (setq feature (symbol-name feature))
+		    (when (string-match
+			   "\\(radical\\|strokes\\)@\\([^@*]+\\)\\(\\*\\|$\\)"
+			   feature)
+		      (setq domain (intern (match-string 2 feature)))
+		     (unless (memq domain dest)
+		       (setq dest (cons domain dest)))))
+		  (sort dest #'string<))))
 	(setq key (intern (format "%s@%s" 'ideographic-radical domain)))
 	(when (and (memq key attributes)
 		   (setq value (get-char-attribute char key)))
@@ -878,12 +894,18 @@
 			     (intern (format "%s*sources" name))))
 		      (not (string-match "\\*sources$" (symbol-name name)))
 		      (or (eq name '<-identical)
+			  (string-match "^->halfwidth" (symbol-name name))
+			  (and
+			   (string-match "^->fullwidth" (symbol-name name))
+			   (not
+			    (and (consp value)
+				 (characterp (car value))
+				 (encode-char
+				  (car value) '=ucs 'defined-only))))
 			  (string-match "^->simplified" (symbol-name name))
-                          ;; (string-match "^<-same" (symbol-name name))
-			  (string-match "^->same" (symbol-name name))
-                          ;; (string-match "^->ideographic-same" (symbol-name name))
 			  (string-match "^->vulgar" (symbol-name name))
 			  (string-match "^->wrong" (symbol-name name))
+			  (string-match "^->same" (symbol-name name))
 			  (string-match "^->original" (symbol-name name))
 			  (string-match "^->ancient" (symbol-name name))
 			  ))
@@ -1088,24 +1110,30 @@
 					   no-ucs-unified
 					   script excluded-script)
   (insert-char-data char printable)
-  (let ((variants (or (char-variants char)
-		      (let ((ucs (get-char-attribute char '->ucs)))
-			(if ucs
-			    (delete char (char-variants (int-char ucs)))))))
-	variant vs)
+  (let ((variants (char-variants char))
+	rest
+	variant vs ret)
     (setq variants (sort variants #'<))
-    (while variants
-      (setq variant (car variants))
-      (if (and (or (null script)
-		   (null (setq vs (get-char-attribute variant 'script)))
-		   (memq script vs))
-	       (or (null excluded-script)
-		   (null (setq vs (get-char-attribute variant 'script)))
-		   (not (memq excluded-script vs))))
-	  (or (and no-ucs-unified (get-char-attribute variant '=ucs))
-	      (insert-char-data variant printable)))
-      (setq variants (cdr variants))
-      )))
+    (setq rest variants)
+    (setq variants (cons char variants))
+    (while rest
+      (setq variant (car rest))
+      (unless (get-char-attribute variant '<-subsumptive)
+	(if (and (or (null script)
+		     (null (setq vs (get-char-attribute variant 'script)))
+		     (memq script vs))
+		 (or (null excluded-script)
+		     (null (setq vs (get-char-attribute variant 'script)))
+		     (not (memq excluded-script vs))))
+	    (unless (and no-ucs-unified (get-char-attribute variant '=ucs))
+	      (insert-char-data variant printable)
+	      (if (setq ret (char-variants variant))
+		  (while ret
+		    (or (memq (car ret) variants)
+                        ;; (get-char-attribute (car ret) '<-subsumptive)
+			(setq rest (nconc rest (list (car ret)))))
+		    (setq ret (cdr ret)))))))
+      (setq rest (cdr rest)))))
 
 (defun insert-char-range-data (min max &optional script excluded-script)
   (let ((code min)