;;; 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, 2011 MORIOKA Tomohiko.
;; Author: MORIOKA Tomohiko <tomo@kanji.zinbun.kyoto-u.ac.jp>
;; Keywords: CHISE, Character Database, ISO/IEC 10646, UCS, Unicode, MULE.
;;; Code:
-(require 'alist)
+(require 'chise-subr)
+(require 'ideograph-subr)
(defconst unidata-normative-category-alist
'(("Lu" letter uppercase)
("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)
- ((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
=jis-x0208@1983
japanese-jisx0212
chinese-gb2312
+ =jis-x0208@1990
chinese-cns11643-1
chinese-cns11643-2
chinese-cns11643-3
chinese-cns11643-5
chinese-cns11643-6
chinese-cns11643-7
- =jis-x0208@1990
- =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
ethiopic-ucs
=big5-cdp
=gt
+ =adobe-japan1-0
+ =adobe-japan1-1
+ =adobe-japan1-2
+ =adobe-japan1-3
+ =adobe-japan1-4
+ =adobe-japan1-5
+ =adobe-japan1-6
+ =hanyo-denshi/ja
+ =hanyo-denshi/jb
+ =hanyo-denshi/jc
+ =hanyo-denshi/jd
+ =hanyo-denshi/ft
+ =hanyo-denshi/ia
+ =hanyo-denshi/ib
+ =hanyo-denshi/hg
ideograph-daikanwa-2
ideograph-daikanwa
=cbeta
+ =gt-k
ideograph-hanziku-1
ideograph-hanziku-2
ideograph-hanziku-3
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
+ =>>hanyo-denshi/ft
+ =>>gt
+ =>jis-x0208@usual
+ =>jis-x0208
+ =>jis-x0208@1997
+ =>jis-x0213-1
+ =>jis-x0213-1@2000
+ =>jis-x0213-1@2004
+ =>jis-x0213-2@usual
+ =>jis-x0213-2
+ ==>ucs@bucs
+ =>ucs@iso
+ =>ucs@unicode
+ =>ucs@jis
+ =>ucs@JP
+ =>ucs@cns
+ =>ucs@ks
+ =>>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
- =jef-china3))
+ =>zinbun-oracle
+ =ruimoku-v6
+ =>>ruimoku-v6
+ =jef-china3
+ =shinjigen))
+
+
+;;; @ char-db formatters
+;;;
(defun char-db-make-char-spec (char)
(let (ret char-spec)
'(=daikanwa
=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)))
((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)
(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
+ =daikanwa/ho))
+ "(%-18s . %04d)\t; %c")
+ ((eq name '=shinjigen@1ed/24pr)
+ "(%-18s . %04d)\t; %c")
+ ((or (memq name '(=daikanwa
+ =daikanwa@rev1 =daikanwa@rev2
+ =daikanwa/+p =daikanwa/+2p
+ =>>daikanwa =>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)
+ ((memq name '(=hanyo-denshi/ks mojikyo))
"(%-18s . %06d)\t; %c")
((>= (charset-dimension name) 2)
"(%-18s . #x%04X)\t; %c")
(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
=jis-x0208@1990
- =jis-x0213-1-2000
- =jis-x0213-2-2000
+ =jis-x0213-1@2000
+ =jis-x0213-1@2004
+ =jis-x0213-2
=jis-x0212
=jis-x0208@1983
- =jis-x0208@1978))))
+ =jis-x0208@1978
+ =shinjigen))))
((eq source 'CN)
(setq required-features
(union required-features
atr-d)
#'char-attribute-name<)))
(insert "(")
+ (when (memq '<-subsumptive attributes)
+ (when readable
+ (when (setq value (get-char-attribute char '<-subsumptive))
+ (char-db-insert-relation-feature char '<-subsumptive value
+ line-breaking
+ ccss readable)))
+ (setq attributes (delq '<-subsumptive attributes)))
+ (when (and (memq '<-denotational attributes)
+ (setq value (get-char-attribute char '<-denotational)))
+ (char-db-insert-relation-feature char '<-denotational value
+ line-breaking
+ ccss readable)
+ (setq attributes (delq '<-denotational attributes)))
(when (and (memq 'name attributes)
(setq value (get-char-attribute char 'name)))
(insert (format
line-breaking))
(setq attributes (delq 'script attributes))
)
- ;; (when (and (memq '<-denotational attributes)
- ;; (setq value (get-char-attribute char '<-denotational))
- ;; (null (cdr value))
- ;; (setq value (encode-char (car value) 'ucs 'defined-only)))
- ;; (insert (format "(%-18s . #x%04X)\t; %c%s"
- ;; '=>ucs value (decode-char 'ucs value)
- ;; line-breaking))
- ;; (setq attributes (delq '<-denotational attributes)))
(dolist (name '(=>ucs =>ucs*))
(when (and (memq name attributes)
(setq value (get-char-attribute char name)))
name value (decode-char '=ucs value)
line-breaking))
(setq attributes (delq name attributes))))
- (dolist (name '(=>ucs@gb =>ucs@cns =>ucs@jis =>ucs@ks =>ucs@big5))
+ (dolist (name '(=>ucs@gb =>ucs@big5))
(when (and (memq name attributes)
(setq value (get-char-attribute char name)))
(insert (format "(%-18s . #x%04X)\t; %c%s"
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
line-breaking))
(setq attributes (delq '->ideograph attributes))
)
- (when (and (memq '->decomposition attributes)
- (setq value (get-char-attribute char '->decomposition)))
- (insert (format "(->decomposition\t%s)%s"
- (mapconcat (lambda (code)
- (cond ((symbolp code)
- (symbol-name code))
- ((characterp code)
- (if readable
- (format "%S" code)
- (format "#x%04X"
- (char-int code))
- ))
- ((integerp code)
- (format "#x%04X" code))
- (t
- (format "%s%S" line-breaking code))))
- value " ")
- line-breaking))
- (setq attributes (delq '->decomposition attributes))
- )
+ ;; (when (and (memq '->decomposition attributes)
+ ;; (setq value (get-char-attribute char '->decomposition)))
+ ;; (insert (format "(->decomposition\t%s)%s"
+ ;; (mapconcat (lambda (code)
+ ;; (cond ((symbolp code)
+ ;; (symbol-name code))
+ ;; ((characterp code)
+ ;; (if readable
+ ;; (format "%S" code)
+ ;; (format "#x%04X"
+ ;; (char-int code))
+ ;; ))
+ ;; ((integerp code)
+ ;; (format "#x%04X" code))
+ ;; (t
+ ;; (format "%s%S" line-breaking code))))
+ ;; value " ")
+ ;; line-breaking))
+ ;; (setq attributes (delq '->decomposition attributes))
+ ;; )
(if (equal (get-char-attribute char '->titlecase)
(get-char-attribute char '->uppercase))
(setq attributes (delq '->titlecase attributes)))
name value
line-breaking))
)
- ((and (not readable)
- (null (get-char-attribute
- char
- (intern (format "%s*sources" name))))
- (not (string-match "\\*sources$" (symbol-name name)))
- (or (eq name '<-identical)
- (string-match "^->halfwidth" (symbol-name name))
- (and
- (string-match "^->fullwidth" (symbol-name name))
- (not
- (and (consp value)
- (characterp (car value))
- (encode-char
- (car value) '=ucs 'defined-only))))
- (string-match "^->simplified" (symbol-name name))
- (string-match "^->vulgar" (symbol-name name))
- (string-match "^->wrong" (symbol-name name))
- (string-match "^->same" (symbol-name name))
- (string-match "^->formed" (symbol-name name))
- (string-match "^->original" (symbol-name name))
- (string-match "^->ancient" (symbol-name name))
- (string-match "^->Oracle-Bones" (symbol-name name))
- ))
+ ((and
+ (not readable)
+ (not (eq name '->subsumptive))
+ (not (eq name '->uppercase))
+ (not (eq name '->lowercase))
+ (not (eq name '->titlecase))
+ (not (eq name '->canonical))
+ (not (eq name '->Bopomofo))
+ (not (eq name '->mistakable))
+ (not (eq name '->ideographic-variants))
+ (null (get-char-attribute
+ char (intern (format "%s*sources" name))))
+ (not (string-match "\\*sources$" (symbol-name name)))
+ (null (get-char-attribute
+ char (intern (format "%s*note" name))))
+ (not (string-match "\\*note$" (symbol-name name)))
+ (or (eq name '<-identical)
+ (eq name '<-uppercase)
+ (eq name '<-lowercase)
+ (eq name '<-titlecase)
+ (eq name '<-canonical)
+ (eq name '<-ideographic-variants)
+ ;; (eq name '<-synonyms)
+ (string-match "^<-synonyms" (symbol-name name))
+ (eq name '<-mistakable)
+ (when (string-match "^->" (symbol-name name))
+ (cond
+ ((string-match "^->fullwidth" (symbol-name name))
+ (not (and (consp value)
+ (characterp (car value))
+ (encode-char
+ (car value) '=ucs 'defined-only)))
+ )
+ (t)))
+ ))
)
((or (eq name 'ideographic-structure)
(eq name 'ideographic-combination)
(eq name 'ideographic-)
- (string-match "^\\(->\\|<-\\)" (symbol-name name)))
+ (eq name '=decomposition)
+ (string-match "^=>*decomposition\\(@[^*]+\\)?$"
+ (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))
(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 ")")))
what-character-original-window-configuration)
(signal (car err) (cdr err)))))))
+
+;;; @ end
+;;;
+
(provide 'char-db-util)
;;; char-db-util.el ends here
projects / chise / xemacs-chise.git.1 / blobdiff