X-Git-Url: http://git.chise.org/gitweb/?a=blobdiff_plain;f=lisp%2Futf-2000%2Fchar-db-util.el;h=2cd174fd94362bdda1264b74001d50c3b25054f0;hb=8a686e75a8faa2aedcf3c062085b4c14a44754df;hp=9a519773de59fe0226a84b2daad5ecc24c2b8610;hpb=cad27acbcc0d333e59cd583d980925b8ef663348;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 9a51977..2cd174f 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, 2010 MORIOKA Tomohiko.
 
 ;; Author: MORIOKA Tomohiko <tomo@kanji.zinbun.kyoto-u.ac.jp>
 ;; Keywords: CHISE, Character Database, ISO/IEC 10646, UCS, Unicode, MULE.
@@ -24,7 +25,8 @@
 
 ;;; Code:
 
-(require 'alist)
+(require 'chise-subr)
+(require 'ideograph-subr)
 
 (defconst unidata-normative-category-alist
   '(("Lu" letter	uppercase)
@@ -61,95 +63,45 @@
     ("So" symbol	other)
     ))
 
-(defconst ideographic-radicals
-  (let ((v (make-vector 215 nil))
-	(i 1))
-    (while (< i 215)
-      (aset v i (decode-char '=ucs (+ #x2EFF i)))
-      (setq i (1+ i)))
-    v))
-
-(defun ideographic-radical (number)
-  (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)))
 
 (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
-	jis-x0212 jis-x0213 cdp shinjigen misc unknown))
-
 (defvar char-db-ignored-attributes '(ideographic-products))
 
-(defun char-attribute-name< (ka kb)
-  (cond
-   ((eq '->denotational kb)
-    t)
-   ((eq '->subsumptive kb)
-    (not (eq '->denotational ka)))
-   ((eq '->denotational ka)
-    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)
-		(cond
-		 ((= (charset-dimension ka)
-		     (charset-dimension kb))
-		  (> (charset-id ka)(charset-id kb)))
-		 (t
-		  (> (charset-dimension ka)
-		     (charset-dimension kb))
-		  ))
-	      t)
-	  (if (<= (charset-id kb) 1)
-	      nil
-	    (< (charset-id ka)(charset-id kb))))
-      nil))
-   ((find-charset kb)
-    t)
-   ((symbolp ka)
-    (cond ((symbolp kb)
-	   (string< (symbol-name ka)
-		    (symbol-name kb)))
-	  (t)))
-   ((symbolp kb)
-    nil)))
-
 (defvar char-db-coded-charset-priority-list
   '(ascii
     control-1
@@ -175,8 +127,10 @@
     chinese-cns11643-5
     chinese-cns11643-6
     chinese-cns11643-7
-    =jis-x0213-1-2000
-    =jis-x0213-2-2000
+    =jis-x0213-1
+    =jis-x0213-1@2000
+    =jis-x0213-1@2004
+    =jis-x0213-2
     korean-ksc5601
     chinese-isoir165
     katakana-jisx0201
@@ -189,6 +143,7 @@
     ideograph-daikanwa-2
     ideograph-daikanwa
     =cbeta
+    =gt-k
     ideograph-hanziku-1
     ideograph-hanziku-2
     ideograph-hanziku-3
@@ -201,15 +156,43 @@
     ideograph-hanziku-10
     ideograph-hanziku-11
     ideograph-hanziku-12
-    =gt-k
+    =>>jis-x0208
+    =>>jis-x0213-1
+    =>>jis-x0213-1@2000
+    =>>jis-x0213-1@2004
+    =>>jis-x0213-2
+    =>>jis-x0208@1978
+    =>>gt
+    =>jis-x0208
+    =>jis-x0208@1997
+    =>jis-x0213-1
+    =>jis-x0213-1@2000
+    =>jis-x0213-1@2004
+    =>jis-x0213-2
+    ==>ucs@bucs
+    =>ucs@iso
+    =>ucs@unicode
+    =>>ucs@unicode
+    =>>ucs@jis
+    =>>ucs@cns
     =ucs@iso
     =ucs@unicode
+    =>>big5-cdp
+    =>>gt-k
+    =>gt
+    =>big5-cdp
+    =>daikanwa
     =big5
     =big5-eten
-    =jis-x0208@1997
     =zinbun-oracle
+    =>zinbun-oracle
     =ruimoku-v6
-    =jef-china3))
+    =jef-china3
+    =shinjigen))
+
+
+;;; @ char-db formatters
+;;;
 
 (defun char-db-make-char-spec (char)
   (let (ret char-spec)
@@ -244,7 +227,8 @@
 					 =daikanwa@rev2
 					 ;; =gt-k
 					 =jis-x0208@1997
-					 )))
+					 ))
+				 (string-match "=ucs@" (symbol-name ccs)))
 			     (setq ccs (charset-name ccs))
 			     (null (assq ccs char-spec))
 			     (setq ret (encode-char char ccs 'defined-only)))
@@ -257,6 +241,12 @@
 			((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)
@@ -439,9 +429,17 @@
 (defun char-db-insert-ccs-feature (name value line-breaking)
   (insert
    (format
-    (cond ((or (memq name '(=daikanwa
+    (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))
+			    =>daikanwa
+			    =gt =>>gt =>gt =gt-k =>>gt-k =cbeta
+			    =zinbun-oracle =>zinbun-oracle))
 	       (string-match "^=adobe-" (symbol-name name)))
 	   "(%-18s . %05d)\t; %c")
 	  ((eq name 'mojikyo)
@@ -502,7 +500,9 @@
 	(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
@@ -511,7 +511,8 @@
 			   =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
@@ -654,21 +655,21 @@
 			line-breaking))
 	(setq attributes (delq name attributes))
 	))
-    (dolist (name '(=>daikanwa))
-      (when (and (memq name attributes)
-		 (setq value (get-char-attribute char name)))
-	(insert
-	 (if (integerp value)
-	     (format "(%-18s . %05d)\t; %c%s"
-		     name value (decode-char '=daikanwa value)
-		     line-breaking)
-	   (format "(%-18s %s)\t; %c%s"
-		   name
-		   (mapconcat (function prin1-to-string)
-			      value " ")
-		   (char-representative-of-daikanwa char)
-		   line-breaking)))
-	(setq attributes (delq name attributes))))
+    ;; (dolist (name '(=>daikanwa))
+    ;;   (when (and (memq name attributes)
+    ;;              (setq value (get-char-attribute char name)))
+    ;;     (insert
+    ;;      (if (integerp value)
+    ;;          (format "(%-18s . %05d)\t; %c%s"
+    ;;                  name value (decode-char '=daikanwa value)
+    ;;                  line-breaking)
+    ;;        (format "(%-18s %s)\t; %c%s"
+    ;;                name
+    ;;                (mapconcat (function prin1-to-string)
+    ;;                           value " ")
+    ;;                (char-representative-of-daikanwa char)
+    ;;                line-breaking)))
+    ;;     (setq attributes (delq name attributes))))
     (when (and (memq 'general-category attributes)
 	       (setq value (get-char-attribute char 'general-category)))
       (insert (format
@@ -1275,6 +1276,10 @@
 		what-character-original-window-configuration)
 	       (signal (car err) (cdr err)))))))
 
+
+;;; @ end
+;;;
+
 (provide 'char-db-util)
 
 ;;; char-db-util.el ends here