X-Git-Url: http://git.chise.org/gitweb/?a=blobdiff_plain;f=lisp%2Futf-2000%2Fideograph-util.el;h=adc7564ebd6389296b828c9cfed2fb881798c632;hb=60394efc79f8db484482b85ae5ab793b68b7048b;hp=477c053592bb9958842f67bcea45c6a1b7e0169d;hpb=163a939693b22fcf952bda9e04b76ce4d0093794;p=chise%2Fxemacs-chise.git

diff --git a/lisp/utf-2000/ideograph-util.el b/lisp/utf-2000/ideograph-util.el
index 477c053..adc7564 100644
--- a/lisp/utf-2000/ideograph-util.el
+++ b/lisp/utf-2000/ideograph-util.el
@@ -1,6 +1,6 @@
 ;;; ideograph-util.el --- Ideographic Character Database utility
 
-;; Copyright (C) 1999,2000,2001 MORIOKA Tomohiko.
+;; Copyright (C) 1999,2000,2001,2002,2003 MORIOKA Tomohiko.
 
 ;; Author: MORIOKA Tomohiko <tomo@kanji.zinbun.kyoto-u.ac.jp>
 ;; Keywords: UTF-2000, ISO/IEC 10646, Unicode, UCS-4, MULE.
@@ -29,16 +29,34 @@
 (defvar ideograph-radical-chars-vector
   (make-vector 215 nil))
 
-(defun char-ideographic-radical (char)
-  (or (get-char-attribute char 'ideographic-radical)
-      (let ((radical
-	     (or (get-char-attribute char 'daikanwa-radical)
-		 (get-char-attribute char 'kangxi-radical)
-		 (get-char-attribute char 'japanese-radical)
-		 (get-char-attribute char 'korean-radical))))
-	(when radical
-	  (put-char-attribute char 'ideographic-radical radical)
-	  radical))))
+(defun char-ideographic-radical (char &optional radical)
+  (let (ret)
+    (or (catch 'tag
+	  (dolist (domain '(ucs daikanwa cns))
+	    (if (and (setq ret (get-char-attribute
+				char
+				(intern
+				 (format "%s@%s"
+					 'ideographic-radical domain))))
+		     (or (eq ret radical)
+			 (null radical)))
+		(throw 'tag ret))))
+	(catch 'tag
+	  (dolist (cell (get-char-attribute char 'ideographic-))
+	    (if (and (setq ret (plist-get cell :radical))
+		     (or (eq ret radical)
+			 (null radical)))
+		(throw 'tag ret))))
+	(get-char-attribute char 'ideographic-radical)
+	(progn
+	  (setq ret
+		(or (get-char-attribute char 'daikanwa-radical)
+		    (get-char-attribute char 'kangxi-radical)
+		    (get-char-attribute char 'japanese-radical)
+		    (get-char-attribute char 'korean-radical)))
+	  (when ret
+	    (put-char-attribute char 'ideographic-radical ret)
+	    ret)))))
 
 (defvar ideograph-radical-strokes-vector
   ;;0  1  2  3  4  5  6  7  8  9
@@ -67,26 +85,62 @@
    11 12 12 12 12 13 13 13 13 14
    14 15 16 16 17])
 
-(defun char-ideographic-strokes (char)
-  (or (get-char-attribute char 'daikanwa-strokes)
-      (get-char-attribute char 'ideographic-strokes)
-      (let ((strokes
-	     (or (get-char-attribute char 'kangxi-strokes)
-		 (get-char-attribute char 'japanese-strokes)
-		 (get-char-attribute char 'korean-strokes)
-		 (let ((r (char-ideographic-radical char))
-		       (ts (get-char-attribute char 'total-strokes)))
-		   (if (and r ts)
-		       (- ts (aref ideograph-radical-strokes-vector r))))
-		 )))
-	(when strokes
-	  (put-char-attribute char 'ideographic-strokes strokes)
-	  strokes))))
+(defun char-ideographic-strokes (char &optional radical)
+  (let (ret)
+    (or (catch 'tag
+	  (dolist (domain '(ucs daikanwa cns))
+	    (if (and (setq ret (get-char-attribute
+				char
+				(intern
+				 (format "%s@%s"
+					 'ideographic-radical domain))))
+		     (or (eq ret radical)
+			 (null radical)))
+		(throw 'tag
+		       (get-char-attribute
+			char
+			(intern
+			 (format "%s@%s"
+				 'ideographic-strokes domain)))))))
+	(catch 'tag
+	  (dolist (cell (get-char-attribute char 'ideographic-))
+	    (if (and (setq ret (plist-get cell :radical))
+		     (or (eq ret radical)
+			 (null radical)))
+		(throw 'tag (plist-get cell :strokes)))))
+	(get-char-attribute char 'daikanwa-strokes)
+	(get-char-attribute char 'ideographic-strokes)
+	(let ((strokes
+	       (or (get-char-attribute char 'kangxi-strokes)
+		   (get-char-attribute char 'japanese-strokes)
+		   (get-char-attribute char 'korean-strokes)
+		   (let ((r (char-ideographic-radical char))
+			 (ts (get-char-attribute char 'total-strokes)))
+		     (if (and r ts)
+			 (- ts (aref ideograph-radical-strokes-vector r))))
+		   )))
+	  (when strokes
+	    (put-char-attribute char 'ideographic-strokes strokes)
+	    strokes)))))
 
 ;;;###autoload
 (defun update-ideograph-radical-table ()
   (interactive)
-  (let (ret script)
+  (let (ret radical script)
+    (dolist (domain '(ucs daikanwa cns))
+      (map-char-attribute
+       (lambda (char radical)
+	 (when (and radical
+		    (or (null (setq script (get-char-attribute char 'script)))
+			(memq 'Ideograph script)))
+	   (unless (memq char
+			 (setq ret
+			       (aref ideograph-radical-chars-vector radical)))
+	     (char-ideographic-strokes char)
+	     (aset ideograph-radical-chars-vector radical
+		   (cons char ret))))
+	 nil)
+       (intern (format "%s@%s" 'ideographic-radical domain))))
     (map-char-attribute
      (lambda (char radical)
        (when (and radical
@@ -99,7 +153,21 @@
 	   (aset ideograph-radical-chars-vector radical
 		 (cons char ret))))
        nil)
-     'ideographic-radical)))
+     'ideographic-radical)
+    (map-char-attribute
+     (lambda (char data)
+       (dolist (cell data)
+	 (setq radical (plist-get cell :radical))
+	 (when (and radical
+		    (or (null (setq script (get-char-attribute char 'script)))
+			(memq 'Ideograph script)))
+	   (unless (memq char
+			 (setq ret
+			       (aref ideograph-radical-chars-vector radical)))
+	     (char-ideographic-strokes char)
+	     (aset ideograph-radical-chars-vector radical
+		   (cons char ret))))))
+     'ideographic-)))
 
 (defun int-list< (a b)
   (if (numberp (car a))
@@ -111,6 +179,10 @@
     (numberp (car b))))
 
 (defun morohashi-daikanwa< (a b)
+  (if (integerp a)
+      (setq a (list a)))
+  (if (integerp b)
+      (setq b (list b)))
   (cond ((eq (car a) 'ho)
 	 (if (eq (car b) 'ho)
 	     (int-list< (cdr a)(cdr b))
@@ -124,112 +196,125 @@
 	     t
 	   (int-list< a b)))))
 
-(defun ideograph-char< (a b)
-  (let ((a-m-m (get-char-attribute a 'ideograph-daikanwa))
-	(b-m-m (get-char-attribute b 'ideograph-daikanwa))
-	a-m-r b-m-r
-	a-s b-s
-	a-u b-u
-	ret pat)
-    (if a-m-m
-	(setq a-s (char-ideographic-strokes a))
-      (setq a-m-r (get-char-attribute a 'morohashi-daikanwa))
-      (if a-m-r
-	  (progn
-	    (setq a-m-m (car a-m-r)
-		  a-m-r (cdr a-m-r))
-	    (if (= (car a-m-r) 0)
-		(progn
-		  (setq ret (decode-char 'ideograph-daikanwa a-m-m))
-		  (if (= (get-char-attribute ret 'ideographic-radical)
-			 (get-char-attribute a 'ideographic-radical))
-		      (setq a-s (char-ideographic-strokes ret))
-		    (setq a-s (char-ideographic-strokes a))))
-              (setq a-s (char-ideographic-strokes
-			 (if (cdr a-m-r)
-			     (progn
-			       (setq pat (list a-m-m (car a-m-r)))
-			       (or (map-char-attribute (lambda (c v)
-							 (if (equal v pat)
-							     c))
-						       'morohashi-daikanwa)
-				   a))
-			   a)))
-	      ))
-	(setq a-s (char-ideographic-strokes a))))
-    (if b-m-m
-	(setq b-s (char-ideographic-strokes b))
-      (setq b-m-r (get-char-attribute b 'morohashi-daikanwa))
-      (if b-m-r
-	  (progn
-	    (setq b-m-m (car b-m-r)
-		  b-m-r (cdr b-m-r))
-	    (if (= (car b-m-r) 0)
-		(progn
-		  (setq ret (decode-char 'ideograph-daikanwa b-m-m))
-		  (if (= (get-char-attribute ret 'ideographic-radical)
-			 (get-char-attribute b 'ideographic-radical))
-		      (setq b-s (char-ideographic-strokes ret))
-		    (setq b-s (char-ideographic-strokes b))))
-              (setq b-s (char-ideographic-strokes
-			 (if (cdr b-m-r)
-			     (progn
-			       (setq pat (list b-m-m (car b-m-r)))
-			       (or (map-char-attribute (lambda (c v)
-							 (if (equal v pat)
-							     c))
-						       'morohashi-daikanwa)
-				   b))
-			   b)))
-	      ))
-	(setq b-s (char-ideographic-strokes b))))
-    (if a-s
-	(if b-s
-	    (if (= a-s b-s)
-		(if a-m-m
-		    (if b-m-m
-			(morohashi-daikanwa< (cons a-m-m a-m-r)
-					     (cons b-m-m b-m-r))
-		      t)
-		  (if b-m-m
-		      nil
-		    (setq a-u (get-char-attribute a 'ucs)
-			  b-u (get-char-attribute b 'ucs))
-		    (if a-u
-			(if b-u
-			    (< a-u b-u)
-			  (setq b-u (or (get-char-attribute b '=>ucs)
-					(get-char-attribute b '->ucs)))
-			  (if b-u
-			      (<= a-u b-u)
-			    t))
-		      (setq a-u (or (get-char-attribute a '=>ucs)
-				    (get-char-attribute a '->ucs)))
-		      (if a-u
-			  (if b-u
-			      (< a-u b-u)
-			    (setq b-u (or (get-char-attribute b '=>ucs)
-					  (get-char-attribute b '->ucs)))
-			    (if b-u
-				(< a-u b-u)
-			      t))
-			(if (or b-u (or (get-char-attribute b '=>ucs)
-					(get-char-attribute b '->ucs)))
-			    nil
-			  (< (char-int a)(char-int b)))))))
-	      (< a-s b-s))
-	  t))))
+;; (defun nil=-int< (a b)
+;;   (cond ((null a) nil)
+;;         ((null b) nil)
+;;         (t (< a b))))
+
+;; (defun nil>-int< (a b)
+;;   (cond ((null a) nil)
+;;         ((null b) t)
+;;         (t (< a b))))
+
+;;;###autoload
+(defun char-representative-of-daikanwa (char)
+  (if (or (encode-char char 'ideograph-daikanwa 'defined-only)
+	  (encode-char char '=daikanwa-rev2 'defined-only))
+      char
+    (let ((m (get-char-attribute char 'morohashi-daikanwa))
+	  m-m m-s pat)
+      (or (when m
+	    (setq m-m (pop m))
+	    (setq m-s (pop m))
+	    (if (= m-s 0)
+		(or (decode-char '=daikanwa-rev2 m-m 'defined-only)
+		    (decode-char 'ideograph-daikanwa m-m))
+	      (when m
+		(setq pat (list m-m m-s))
+		(map-char-attribute (lambda (c v)
+				      (if (equal pat v)
+					  c))
+				    'morohashi-daikanwa))))
+	  char))))
+
+(defun char-attributes-poly< (c1 c2 accessors testers defaulters)
+  (catch 'tag
+    (let (a1 a2 accessor tester dm)
+      (while (and accessors testers)
+	(setq accessor (car accessors)
+	      tester (car testers)
+	      dm (car defaulters))
+	(when (and accessor tester)
+	  (setq a1 (funcall accessor c1)
+		a2 (funcall accessor c2))
+	  (cond ((null a1)
+		 (if a2
+		     (cond ((eq dm '<)
+			    (throw 'tag t))
+			   ((eq dm '>)
+			    (throw 'tag nil)))))
+		((null a2)
+		 (cond ((eq dm '<)
+			(throw 'tag nil))
+		       ((eq dm '>)
+			(throw 'tag t))))
+		(t
+		 (cond ((funcall tester a1 a2)
+			(throw 'tag t))
+		       ((funcall tester a2 a1)
+			(throw 'tag nil))))))
+	(setq accessors (cdr accessors)
+	      testers (cdr testers)
+	      defaulters (cdr defaulters))))))
+
+(defvar ideographic-radical nil)
+
+(defun char-daikanwa-strokes (char &optional radical)
+  (unless radical
+    (setq radical ideographic-radical))
+  (let ((drc (char-representative-of-daikanwa char)))
+    (char-ideographic-strokes
+     (if (= (char-ideographic-radical drc radical)
+	    (char-ideographic-radical char radical))
+	 drc
+       char)
+     radical)))
+
+;;;###autoload
+(defun char-daikanwa (char)
+  (or (encode-char char 'ideograph-daikanwa 'defined-only)
+      (encode-char char '=daikanwa-rev2 'defined-only)
+      (get-char-attribute char 'morohashi-daikanwa)))
+
+;;;###autoload
+(defun char-ucs (char)
+  (or (encode-char char '=ucs 'defined-only)
+      (get-char-attribute char '=>ucs)))
+
+(defun char-id (char)
+  (logand (char-int char) #x3FFFFFFF))
+
+(defun ideograph-char< (a b &optional radical)
+  (let ((ideographic-radical (or radical
+				 ideographic-radical)))
+    (char-attributes-poly<
+     a b
+     '(char-daikanwa-strokes char-daikanwa char-ucs char-id)
+     '(< morohashi-daikanwa< < <)
+     '(> > > >))))
 
 (defun insert-ideograph-radical-char-data (radical)
   (let ((chars
 	 (sort (copy-list (aref ideograph-radical-chars-vector radical))
-	       (function ideograph-char<)))
-	(attributes (sort (char-attribute-list) #'char-attribute-name<))
-	(ccs (sort (charset-list) #'char-attribute-name<)))
+	       (lambda (a b)
+		 (ideograph-char< a b radical))))
+	attributes ccss)
+    (dolist (name (char-attribute-list))
+      (unless (memq name char-db-ignored-attributes)
+	(if (find-charset name)
+	    (push name ccss)
+	  (push name attributes))))
+    (setq attributes (sort attributes #'char-attribute-name<)
+	  ccss (sort ccss #'char-attribute-name<))
     (aset ideograph-radical-chars-vector radical chars)
-    (while chars
-      (insert-char-data (car chars) nil attributes ccs)
-      (setq chars (cdr chars)))))
+    (dolist (char chars)
+      (when (or (not (some (lambda (atr)
+			     (get-char-attribute char atr))
+			   char-db-ignored-attributes))
+		(some (lambda (ccs)
+			(encode-char char ccs 'defined-only))
+		      ccss))
+	(insert-char-data char nil attributes ccss)))))
 
 (defun write-ideograph-radical-char-data (radical file)
   (if (file-directory-p file)
@@ -246,11 +331,50 @@
 	       file))))
   (with-temp-buffer
     (insert-ideograph-radical-char-data radical)
-    (char-db-update-comment)
     (let ((coding-system-for-write 'utf-8))
       (write-region (point-min)(point-max) file)
       )))
 
+(defun ideographic-structure= (char1 char2)
+  (if (char-ref-p char1)
+      (setq char1 (plist-get char1 :char)))
+  (if (char-ref-p char2)
+      (setq char2 (plist-get char2 :char)))
+  (let ((s1 (if (characterp char1)
+		(get-char-attribute char1 'ideographic-structure)
+	      (cdr (assq 'ideographic-structure char1))))
+	(s2 (if (characterp char2)
+		(get-char-attribute char2 'ideographic-structure)
+	      (cdr (assq 'ideographic-structure char2))))
+	e1 e2)
+    (if (or (null s1)(null s2))
+	(char-spec= char1 char2)
+      (catch 'tag
+	(while (and s1 s2)
+	  (setq e1 (car s1)
+		e2 (car s2))
+	  (unless (ideographic-structure= e1 e2)
+	    (throw 'tag nil))
+	  (setq s1 (cdr s1)
+		s2 (cdr s2)))
+	(and (null s1)(null s2))))))
+
+;;;###autoload
+(defun ideographic-structure-find-char (structure)
+  (let (rest)
+    (map-char-attribute (lambda (char value)
+			  (setq rest structure)
+			  (catch 'tag
+			    (while (and rest value)
+			      (unless (ideographic-structure=
+				       (car rest)(car value))
+				(throw 'tag nil))
+			      (setq rest (cdr rest)
+				    value (cdr value)))
+			    (unless (or rest value)
+			      char)))
+			'ideographic-structure)))
+
 (provide 'ideograph-util)
 
 ;;; ideograph-util.el ends here