X-Git-Url: http://git.chise.org/gitweb/?a=blobdiff_plain;f=lisp%2Futf-2000%2Fchar-db-util.el;h=232904a7a9e175cd106d9736696cfb91216d5b6b;hb=9524e80e571eeb14d2bc62b0e87f798e61f9718d;hp=5c9a31782bb9672b9ee54194a402c8fbf3f3285d;hpb=5c2e90ba7433c98f4ee33dbec9e91d81deb6a5d5;p=chise%2Fxemacs-chise.git.1

diff --git a/lisp/utf-2000/char-db-util.el b/lisp/utf-2000/char-db-util.el
index 5c9a317..232904a 100644
--- a/lisp/utf-2000/char-db-util.el
+++ b/lisp/utf-2000/char-db-util.el
@@ -1,6 +1,7 @@
 ;;; char-db-util.el --- Character Database utility -*- coding: utf-8-er; -*-
 
-;; Copyright (C) 1998,1999,2000,2001,2002,2003,2004,2005 MORIOKA Tomohiko.
+;; Copyright (C) 1998, 1999, 2000, 2001, 2002, 2003, 2004, 2005, 2006,
+;;   2007, 2008, 2009 MORIOKA Tomohiko.
 
 ;; Author: MORIOKA Tomohiko <tomo@kanji.zinbun.kyoto-u.ac.jp>
 ;; Keywords: CHISE, Character Database, ISO/IEC 10646, UCS, Unicode, MULE.
@@ -73,7 +74,36 @@
   (aref ideographic-radicals number))
 
 (defconst shuowen-radicals
-  [?一 ?上 ?示 ?三 ?王 ?玉 ?玨 ?气 ?士 ?丨 ?屮 ?艸 ?茻])
+  [?一 ?上 ?示 ?三 ?王 ?玉 ?玨 ?气 ?士 ?丨 ; 010
+   ?屮 ?艸 ?蓐 ?茻 ?小 ?八 ?釆 ?半 ?牛 ?犛 ; 020
+   ?告 ?口 ?凵 ?吅 ?哭 ?走 ?止 ?癶 ?步 ?此 ; 030
+   ?正 ?是 ?辵 ?彳 ?廴 ?㢟 ?行 ?齒 ?牙 ?足 ; 040
+   ?疋 ?品 ?龠 ?冊 ?㗊 ?舌 ?干 ?谷 ?只 ?㕯 ; 050
+   ?句 ?丩 ?古 ?十 ?卅 ?言 ?誩 ?音 ?䇂 ?丵 ; 060
+   ?菐 ?𠬞 ?廾 ?共 ?異 ?舁 ?𦥑 ?䢅 ?爨 ?革 ; 070
+   ?鬲 ?䰜 ?爪 ?𠃨 ?鬥 ?又 ?𠂇 ?㕜 ?支 ?𦘒 ; 080
+   ?聿 ?畫 ?隶 ?臤 ?臣 ?殳 ?殺 ?𠘧 ?寸 ?皮 ; 090
+   ?㼱 ?攴 ?敎 ?卜 ?用 ?爻 ?㸚 ?𥄎 ?目 ?䀠 ; 100
+   ?眉 ?盾 ?自 ?白 ?鼻 ?皕 ?習 ?羽 ?隹 ?奞 ; 110
+   ?萑 ?𦫳 ?苜 ?羊 ?羴 ?瞿 ?雔 ?雥 ?鳥 ?烏 ; 120
+   ?𠦒 ?冓 ?幺 ?𢆶 ?叀 ?玄 ?予 ?放 ?𠬪 ?𣦼 ; 130
+   ?歺 ?死 ?冎 ?骨 ?肉 ?筋 ?刀 ?刃 ?㓞 ?丰 ; 140
+   ?耒 ?𧢲 ?竹 ?箕 ?丌 ?左 ?工 ?㠭 ?巫 ?甘 ; 150
+   ?曰 ?乃 ?丂 ?可 ?兮 ?号 ?亏 ?旨 ?喜 ?壴 ; 160
+   ?鼓 ?豈 ?豆 ?豊 ?豐 ?䖒 ?虍 ?虎 ?虤 ?皿 ; 170
+   ?𠙴 ?去 ?血 ?丶 ?丹 ?青 ?井 ?皀 ?鬯 ?食 ; 180
+   ?亼 ?會 ?倉 ?入 ?缶 ?矢 ?高 ?冂 ?𩫏 ?京 ; 190
+   ?亯 ?𣆪 ?畗 ?㐭 ?嗇 ?來 ?麥 ?夊 ?舛 ?䑞 ; 200
+   ?韋 ?弟 ?夂 ?久 ?桀 ?木 ?東 ?林 ?才 ?叒 ; 210
+   ?之 ?帀 ?出 ?𣎵 ?生 ?乇 ?𠂹 ?𠌶 ?華 ?𥝌 ; 220
+   ?稽 ?巢 ?桼 ?束 ?㯻 ?囗 ?員 ?貝 ?邑 ?𨛜 ; 230
+   ?日 ?旦 ?倝 ?㫃 ?冥 ?晶 ?月 ?有 ?明 ?囧 ; 240
+   ?夕 ?多 ?毌 ?𢎘 ?𣐺 ?卣 ?齊 ?朿 ?片 ?鼎 ; 250
+   ?克 ?彔 ?禾 ?秝 ?黍 ?香 ?米 ?毇 ?臼 ?凶 ; 260
+   ?𣎳 ?林 ?麻 ?尗 ?耑 ?韭 ?瓜 ?瓠 ?宀 ?宮 ; 270
+   ?呂 ?穴 ?㝱 ?𤕫 ?冖 ?𠔼 ?冃 ?㒳 ?网 ?襾 ; 280
+   ?巾 ?巿 ?帛 ?白 ?㡀 ?黹 ?人 ?𠤎 ?匕 ?从 ; 290
+   ])
 
 (defun shuowen-radical (number)
   (aref shuowen-radicals (1- number)))
@@ -81,7 +111,7 @@
 (defvar char-db-file-coding-system 'utf-8-mcs-er)
 
 (defvar char-db-feature-domains
-  '(ucs daikanwa cns gt jis jis/alt jis/a jis/b
+  '(ucs ucs/compat daikanwa cns gt jis jis/alt jis/a jis/b
 	jis-x0212 jis-x0213 cdp shinjigen misc unknown))
 
 (defvar char-db-ignored-attributes '(ideographic-products))
@@ -96,20 +126,85 @@
     nil)
    ((eq '->subsumptive ka)
     nil)
+   ((and (symbolp ka)
+	 (string-match "^->" (symbol-name ka)))
+    (cond ((and (symbolp kb)
+		(string-match "^->" (symbol-name kb)))
+	   (string< (symbol-name ka)
+		    (symbol-name kb))
+	   ))
+    )
+   ((and (symbolp kb)
+	 (string-match "^->" (symbol-name kb)))
+    t)
+   ((and (symbolp ka)
+	 (string-match "^<-" (symbol-name ka)))
+    (cond ((symbolp kb)
+	   (cond ((string-match "^<-" (symbol-name kb))
+		  (string< (symbol-name ka)
+			   (symbol-name kb))
+		  )
+                 ;; ((string-match "^->" (symbol-name kb))
+                 ;;  t)
+		 )))
+    )
+   ((and (symbolp kb)
+	 (string-match "^<-" (symbol-name kb)))
+    t
+    ;; (not (string-match "^->" (symbol-name ka)))
+    )
    ((find-charset ka)
     (if (find-charset kb)
-	(if (<= (charset-id ka) 1)
-	    (if (<= (charset-id kb) 1)
+	(if (or (<= (charset-id ka) 1)
+		(and (charset-final ka)
+		     (>= (charset-final ka) ?@)))
+	    (if (or (<= (charset-id kb) 1)
+		    (and (charset-final kb)
+			 (>= (charset-final kb) ?@)))
 		(cond
 		 ((= (charset-dimension ka)
 		     (charset-dimension kb))
-		  (> (charset-id ka)(charset-id kb)))
+		  (if (and (<= (charset-id ka) 1)
+			   (<= (charset-id kb) 1))
+		      (> (charset-id ka)(charset-id kb))
+		    (if (and (charset-final ka)
+			     (charset-final kb))
+			(cond
+			 ((= (charset-final ka)(charset-final kb))
+			  (if (<= (charset-id ka) 1)
+			      (if (<= (charset-id kb) 1)
+				  (> (charset-id ka)(charset-id kb))
+				t)
+			    (if (<= (charset-id kb) 1)
+				nil
+			      (< (charset-id ka)(charset-id kb))))
+			  )
+			 ((< (charset-final ka)(charset-final kb))
+			  ))
+		      (if (<= (charset-id ka) 1)
+			  (if (<= (charset-id kb) 1)
+			      (> (charset-id ka)(charset-id kb))
+			    t)
+			(if (<= (charset-id kb) 1)
+			    nil
+			  (cond ((and (charset-final ka)
+				      (>= (charset-final ka) ?@))
+				 t)
+				((and (charset-final kb)
+				      (>= (charset-final kb) ?@))
+				 nil)
+				(t
+				 (< (charset-id ka)(charset-id kb))
+				 ))))))
+		  )
 		 (t
 		  (> (charset-dimension ka)
 		     (charset-dimension kb))
 		  ))
 	      t)
-	  (if (<= (charset-id kb) 1)
+	  (if (or (<= (charset-id kb) 1)
+		  (and (charset-final kb)
+		       (>= (charset-final kb) ?@)))
 	      nil
 	    (< (charset-id ka)(charset-id kb))))
       nil))
@@ -140,6 +235,7 @@
     =jis-x0208@1983
     japanese-jisx0212
     chinese-gb2312
+    =jis-x0208@1990
     chinese-cns11643-1
     chinese-cns11643-2
     chinese-cns11643-3
@@ -147,7 +243,6 @@
     chinese-cns11643-5
     chinese-cns11643-6
     chinese-cns11643-7
-    =jis-x0208@1990
     =jis-x0213-1-2000
     =jis-x0213-2-2000
     korean-ksc5601
@@ -181,6 +276,7 @@
     =big5-eten
     =jis-x0208@1997
     =zinbun-oracle
+    =ruimoku-v6
     =jef-china3))
 
 (defun char-db-make-char-spec (char)
@@ -215,6 +311,7 @@
 				       '(=daikanwa
 					 =daikanwa@rev2
 					 ;; =gt-k
+					 =jis-x0208@1997
 					 )))
 			     (setq ccs (charset-name ccs))
 			     (null (assq ccs char-spec))
@@ -228,6 +325,12 @@
 			((setq ret (get-char-attribute char 'name*))
 			 (setq char-spec (cons (cons 'name* ret) char-spec))
 			 ))
+		  )
+		 ((setq ret (get-char-attribute
+			     char 'ideographic-combination))
+		  (setq char-spec
+			(cons (cons 'ideographic-combination ret)
+			      char-spec))
 		  ))
 	   char-spec)
 	  ((consp char)
@@ -410,9 +513,16 @@
 (defun char-db-insert-ccs-feature (name value line-breaking)
   (insert
    (format
-    (cond ((memq name '(=daikanwa
-			=daikanwa@rev1 =daikanwa@rev2
-			=gt =gt-k =cbeta =zinbun-oracle))
+    (cond ((memq name '(=shinjigen
+			=shinjigen@1ed
+			=shinjigen@rev =shinjigen/+p@rev))
+	   "(%-18s .  %04d)\t; %c")
+	  ((eq name '=shinjigen@1ed/24pr)
+	   "(%-18s . %04d)\t; %c")
+	  ((or (memq name '(=daikanwa
+			    =daikanwa@rev1 =daikanwa@rev2
+			    =gt =gt-k =cbeta =zinbun-oracle))
+	       (string-match "^=adobe-" (symbol-name name)))
 	   "(%-18s . %05d)\t; %c")
 	  ((eq name 'mojikyo)
 	   "(%-18s . %06d)\t; %c")
@@ -472,7 +582,9 @@
 	(setq required-features nil)
 	(dolist (source sources)
 	  (cond
-	   ((memq source '(JP JP/Jouyou shinjigen-1))
+	   ((memq source '(JP
+			   JP/Jouyou
+			   shinjigen shinjigen@1ed shinjigen@rev))
 	    (setq required-features
 		  (union required-features
 			 '(=jis-x0208
@@ -481,7 +593,8 @@
 			   =jis-x0213-2-2000
 			   =jis-x0212
 			   =jis-x0208@1983
-			   =jis-x0208@1978))))
+			   =jis-x0208@1978
+			   =shinjigen))))
 	   ((eq source 'CN)
 	    (setq required-features
 		  (union required-features
@@ -566,7 +679,7 @@
 		  #'char-attribute-name<)))
     (insert "(")
     (when (memq '<-subsumptive attributes)
-      (unless readable
+      (when readable
 	(when (setq value (get-char-attribute char '<-subsumptive))
 	  (char-db-insert-relation-feature char '<-subsumptive value
 					   line-breaking
@@ -1005,37 +1118,51 @@
 				 name value
 				 line-breaking))
 		 )
-		((and (not readable)
-		      (null (get-char-attribute
-			     char
-			     (intern (format "%s*sources" name))))
-		      (not (string-match "\\*sources$" (symbol-name name)))
-		      (or (eq name '<-identical)
-			  (string-match "^->compat" (symbol-name name))
-			  (string-match "^->halfwidth" (symbol-name name))
-			  (and
-			   (string-match "^->fullwidth" (symbol-name name))
-			   (not
-			    (and (consp value)
-				 (characterp (car value))
-				 (encode-char
-				  (car value) '=ucs 'defined-only))))
-			  (string-match "^->simplified" (symbol-name name))
-			  (string-match "^->vulgar" (symbol-name name))
-			  (string-match "^->wrong" (symbol-name name))
-			  (string-match "^->same" (symbol-name name))
-			  (string-match "^->formed" (symbol-name name))
-			  (string-match "^->original" (symbol-name name))
-			  (string-match "^->ancient" (symbol-name name))
-			  (string-match "^->Oracle-Bones" (symbol-name name))
-			  ))
+		((and
+		  (not readable)
+		  (not (eq name '->subsumptive))
+		  (not (eq name '->uppercase))
+		  (not (eq name '->lowercase))
+		  (not (eq name '->titlecase))
+		  (not (eq name '->canonical))
+		  (not (eq name '->Bopomofo))
+		  (not (eq name '->mistakable))
+		  (not (eq name '->ideographic-variants))
+		  (null (get-char-attribute
+			 char (intern (format "%s*sources" name))))
+		  (not (string-match "\\*sources$" (symbol-name name)))
+		  (null (get-char-attribute
+			 char (intern (format "%s*note" name))))
+		  (not (string-match "\\*note$" (symbol-name name)))
+		  (or (eq name '<-identical)
+		      (eq name '<-uppercase)
+		      (eq name '<-lowercase)
+		      (eq name '<-titlecase)
+		      (eq name '<-canonical)
+		      (eq name '<-ideographic-variants)
+                      ;; (eq name '<-synonyms)
+		      (string-match "^<-synonyms" (symbol-name name))
+		      (eq name '<-mistakable)
+		      (when (string-match "^->" (symbol-name name))
+			(cond
+			 ((string-match "^->fullwidth" (symbol-name name))
+			  (not (and (consp value)
+				    (characterp (car value))
+				    (encode-char
+				     (car value) '=ucs 'defined-only)))
+			  )
+			 (t)))
+		      ))
 		 )
 		((or (eq name 'ideographic-structure)
 		     (eq name 'ideographic-combination)
 		     (eq name 'ideographic-)
 		     (eq name '=decomposition)
 		     (string-match "^=>decomposition" (symbol-name name))
-		     (string-match "^\\(->\\|<-\\)" (symbol-name name)))
+		     (string-match "^\\(->\\|<-\\)[^*]*$" (symbol-name name))
+		     (string-match "^\\(->\\|<-\\)[^*]*\\*sources$"
+				   (symbol-name name))
+		     )
 		 (char-db-insert-relation-feature char name value
 						  line-breaking
 						  ccss readable))
@@ -1108,9 +1235,18 @@
 		 (insert ")")
 		 (insert line-breaking))
 		(t
-		 (insert (format "(%-18s . %S)%s"
-				 name value
-				 line-breaking)))
+                 (insert (format "(%-18s" name))
+		 (setq ret (prin1-to-string value))
+		 (unless (< (+ (current-column)
+			       (length ret)
+			       3)
+			    76)
+		   (insert line-breaking))
+		 (insert " . " ret ")" line-breaking)
+		 ;; (insert (format "(%-18s . %S)%s"
+                 ;;                 name value
+                 ;;                 line-breaking))
+		 )
 		))
       (setq attributes (cdr attributes)))
     (insert ")")))