;;; char-db-util.el --- Character Database utility -*- coding: utf-8-er; -*-
-;; Copyright (C) 1998,1999,2000,2001,2002,2003,2004,2005,2006,2007,
-;; 2008,2009,2010,2011,2012,2013,2014,2015,2016,2017 MORIOKA Tomohiko.
+;; Copyright (C) 1998, 1999, 2000, 2001, 2002, 2003, 2004, 2005, 2006, 2007,
+;; 2008, 2009, 2010, 2011, 2012, 2013, 2014, 2015, 2016, 2017, 2018, 2019,
+;; 2020, 2021, 2022, 2023, 2024 MORIOKA Tomohiko.
;; Author: MORIOKA Tomohiko <tomo@kanji.zinbun.kyoto-u.ac.jp>
;; Keywords: CHISE, Character Database, ISO/IEC 10646, UCS, Unicode, MULE.
?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
+ ?疋 ?品 ?龠 ?冊 ?㗊 ?舌 ?干 ?𧮫 ?只 ?㕯 ; 050
?句 ?丩 ?古 ?十 ?卅 ?言 ?誩 ?音 ?䇂 ?丵 ; 060
?菐 ?𠬞 ?𠬜 ?共 ?異 ?舁 ?𦥑 ?䢅 ?爨 ?革 ; 070
- ?鬲 ?䰜 ?爪 ?𠃨 ?鬥 ?又 ?𠂇 ?㕜 ?支 ?𦘒 ; 080
+ ?鬲 ?䰜 ?爪 ?𠃨 ?鬥 ?又 ?𠂇 ?史 ?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
+ ?雈 ?𦫳 ?𥄕 ?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
+ ?歺 ?死 ?冎 ?骨 ?肉 ?筋 ?刀 ?刃 ?㓞 ?丯 ; 140
+ ?耒 ?角 ?竹 ?箕 ?丌 ?左 ?工 ?㠭 ?巫 ?甘 ; 150
?曰 ?乃 ?丂 ?可 ?兮 ?号 ?亏 ?旨 ?喜 ?壴 ; 160
; ?旨 ?曰 ?乃 ?丂 ?可 ?兮 ?号 ?亏 ?喜 ?壴 ; 160
?鼓 ?豈 ?豆 ?豊 ?豐 ?䖒 ?虍 ?虎 ?虤 ?皿 ; 170
?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
+ ?𣎳 ?𣏟 ?麻 ?尗 ?耑 ?韭 ?瓜 ?瓠 ?宀 ?宮 ; 270
+ ?呂 ?穴 ?㝱 ?疒 ?冖 ?𠔼 ?冃 ?㒳 ?网 ?襾 ; 280
?巾 ?巿 ?帛 ?白 ?㡀 ?黹 ?人 ?𠤎 ?匕 ?从 ; 290
?比 ?北 ?丘 ?㐺 ?𡈼 ?重 ?臥 ?身 ?㐆 ?衣 ; 300
?裘 ?老 ?毛 ?毳 ?尸 ?尺 ?尾 ?履 ?舟 ?方 ; 310
?而 ?豕 ?㣇 ?彑 ?豚 ?豸 ?𤉡 ?易 ?象 ?馬 ; 370
?𢊁 ?鹿 ?麤 ?㲋 ?兔 ?萈 ?犬 ?㹜 ?鼠 ?能 ; 380
?熊 ?火 ?炎 ?黑 ?囪 ?焱 ?炙 ?赤 ?大 ?亦 ; 390
+ ?夨 ?夭 ?交 ?尣 ?壺 ?壹 ?㚔 ?奢 ?亢 ?夲 ; 400
+ ?夰 ?亣 ?夫 ?立 ?竝 ?囟 ?思 ?心 ?惢 ?水 ; 410
+ ?沝 ?瀕 ?𡿨 ?巜 ?川 ?泉 ?灥 ?永 ?𠂢 ?谷 ; 420
+ ?仌 ?雨 ?雲 ?魚 ?𩺰 ?燕 ?龍 ?飛 ?非 ?卂 ; 430
+ ?𠃉 ?不 ?至 ?西 ?鹵 ?鹽 ?戶 ?門 ?耳 ?𦣝 ; 440
+ ?手 ?𠦬 ?女 ?毋 ?民 ?丿 ?𠂆 ?乁 ?氏 ?氐 ; 450
+ ?戈 ?戉 ?我 ?亅 ?琴 ?𠃊 ?亡 ?匸 ?匚 ?曲 ; 460
+ ?甾 ?瓦 ?弓 ?弜 ?弦 ?系 ?糸 ?素 ?絲 ?率 ; 470
+ ?虫 ?䖵 ?蟲 ?風 ?它 ?龜 ?黽 ?卵 ?二 ?土 ; 480
+ ?垚 ?堇 ?里 ?田 ?畕 ?黃 ?男 ?力 ?劦 ?金 ; 490
+ ?幵 ?勺 ?几 ?且 ?斤 ?斗 ?矛 ?車 ?𠂤 ?𨸏 ; 500
+ ?𨺅 ?厽 ?四 ?宁 ?叕 ?亞 ?五 ?六 ?七 ?九 ; 510
+ ?禸 ?嘼 ?甲 ?乙 ?丙 ?丁 ?戊 ?己 ?巴 ?庚 ; 520
+ ?辛 ?辡 ?壬 ?癸 ?子 ?了 ?孨 ?𠫓 ?丑 ?寅 ; 530
+ ?卯 ?辰 ?巳 ?午 ?未 ?申 ?酉 ?酋 ?戌 ?亥 ; 540
])
(defun shuowen-radical (number)
=hanyo-denshi/ia
=hanyo-denshi/ib
=hanyo-denshi/hg
+ =hanyo-denshi/ip
=hanyo-denshi/jt
=hanyo-denshi/ks
=hanyo-denshi/tk
=gt
=gt-k
=daikanwa
+ =daijiten
=daikanwa@rev2
=daikanwa@rev1
=daikanwa/+p
+ =daikanwa/ho
==daikanwa
+ ==daijiten
=cbeta
ideograph-hanziku-1
ideograph-hanziku-2
;; =>>>hanyo-denshi/jt
;; =>>>hanyo-denshi/ks
;; =>>>gt
+ =>>mj
=>>adobe-japan1-0
=>>adobe-japan1-1
=>>adobe-japan1-2
=>jis-x0213-1@2004
=>jis-x0213-2
==>ucs@bucs
- =>iwds-1
- =>ucs@hanyo-denshi
+ ==>daijiten
+ ;; =>ucs@hanyo-denshi
=>ucs@iso
=>ucs@unicode
=>ucs@jis
=>>ucs@iso
=>>ucs@unicode
=>>ucs@jis
+ =>>ucs@JP
=>>ucs@cns
=>>>ucs@iso
=>>>ucs@unicode
=>>big5-cdp
=>>gt-k
=+>gt
- =>gt
- =>mj
+ =+>big5-cdp
=>big5-cdp
- =>daikanwa
- =>daikanwa/ho
- =>cns11643-5
- =>cns11643-7
+ =>big5-cdp@iwds-1
+ =>mj
+ =>hanyo-denshi/tk
+ =>iwds-1
=big5
=big5-eten
- =>gt-k
=zinbun-oracle
=>zinbun-oracle
=ruimoku-v6
=>>ruimoku-v6
==ruimoku-v6
+ =>ruimoku-v6
=jef-china3
- =>cbeta
=shinjigen
+ =>cns11643-5
+ =>cns11643-7
+ =>gt
+ =>gt-k
+ =>daikanwa
+ =>daikanwa/ho
+ =>cbeta
=ucs-var-001
=ucs-var-002
=ucs-var-003
=ucs-var-004
=ucs-var-005
=ucs-var-006
+ =ucs-var-007
=ucs-var-008
=ucs-var-010
=ucs-itaiji-001
=ucs-itaiji-004
=ucs-itaiji-005
=ucs-itaiji-006
+ =ucs-itaiji-007
=ucs-itaiji-008
=ucs-itaiji-009
+ =ucs-itaiji-010
+ =ucs-itaiji-011
+ =ucs-itaiji-012
+ =ucs-itaiji-013
+ =ucs-itaiji-014
+ =ucs-itaiji-015
+ =ucs-itaiji-016
+ =ucs-itaiji-018
+ =+>ucs-var-003
=>ucs-itaiji-001
=>ucs-itaiji-002
+ =>ucs-itaiji-003
+ =>ucs-itaiji-004
=>ucs-itaiji-005
+ =>ucs-itaiji-006
+ =>ucs-itaiji-007
+ =>ucs-itaiji-008
+ =>ucs-itaiji-009
+ =>ucs-itaiji-010
+ =>ucs-itaiji-011
+ =>ucs-itaiji-012
+ =>ucs-itaiji-014
+ =>ucs-itaiji-015
+ =>ucs-itaiji-016
+ =>ucs-var-001
+ =big5-cdp-var-001
+ =big5-cdp-var-002
=big5-cdp-var-003
=big5-cdp-var-004
=big5-cdp-var-005
+ =big5-cdp-var-007
+ =big5-cdp-var-008
+ =big5-cdp-var-010
=big5-cdp-itaiji-001
=big5-cdp-itaiji-002
=big5-cdp-itaiji-003
+ =big5-cdp-itaiji-004
+ =big5-cdp-itaiji-005
=>big5-cdp-itaiji-001
+ =>big5-cdp-itaiji-002
+ =>big5-cdp-itaiji-003
+ =>big5-cdp-itaiji-004
+ =>big5-cdp-itaiji-005
+ =>big5-cdp-itaiji-006
+ =>big5-cdp-var-001
=>ucs@iwds-1
+ =>ucs@cognate
=>ucs@component
+ =>ucs@iwds-1/normalized
+ =>ucs-itaiji-001@iwds-1
+ =>ucs-itaiji-002@iwds-1
+ =>ucs-itaiji-003@iwds-1
+ =>ucs-itaiji-006@iwds-1
+ =>big5-cdp@component
+ =>big5-cdp@cognate
==ucs@gb
- =ucs@JP/hanazono))
+ =ucs@gb
+ ==ucs-var-002
+ ==ucs-itaiji-002
+ =ucs@JP/hanazono
+ =koseki
+ =shuowen-jiguge
+ =>shuowen-jiguge
+ ===shuowen-jiguge5
+ ===shuowen-jiguge4
+ ===chise-hdic-ktb-seal))
;;; @ char-db formatters
;; =gt-k
=jis-x0208@1997
))
- (eq (charset-property ccs 'iso-ir) 177)
+ (and (eq (charset-property ccs 'iso-ir) 177)
+ (not (eq (charset-name ccs) '=ucs@big5)))
;; (string-match "=ucs@" (symbol-name ccs))
)
(setq ccs (charset-name ccs))
=>>adobe-japan1-6 =+>adobe-japan1-6
=>cbeta =cbeta =>>cbeta ==cbeta ===cbeta
=zinbun-oracle =>zinbun-oracle
+ =daijiten ==daijiten ===daijiten ==>daijiten
===hng-jou ===hng-keg ===hng-dng ===hng-mam
===hng-drt ===hng-kgk ===hng-myz ===hng-kda
- ===hng-khi ===hng-khm ===hng-hok ===hng-kyd ===hng-sok
+ ===hng-khi ===hng-khm ===hng-fhs ===hng-hok
+ ===hng-kyd ===hng-sok
===hng-yhk ===hng-kak ===hng-kar ===hng-kae
===hng-sys ===hng-tsu ===hng-tzj
- ===hng-hos ===hng-nak ===hng-jhk
+ ===hng-hos ===hng-kkh ===hng-nak ===hng-jhk
===hng-hod ===hng-gok ===hng-ink ===hng-nto
- ===hng-nkm ===hng-k24 ===hng-nkk
+ ===hng-nkm ===hng-k24 ===hng-ini ===hng-nkk
===hng-kcc ===hng-kcj ===hng-kbk ===hng-sik
===hng-skk ===hng-kyu ===hng-ksk ===hng-wan
===hng-okd ===hng-wad ===hng-kmi
- ===hng-zkd ===hng-doh ===hng-jyu
- ===hng-tzs ===hng-kss ===hng-kyo
- ===hng-smk))
+ ===hng-zkd ===hng-doh ===hng-jyu ===hng-tzs
+ ===hng-sai ===hng-kad ===hng-kss
+ ===hng-kyo ===hng-ykk ===hng-saa
+ ===hng-sab ===hng-wks ===hng-wke ===hng-smk
+ ===hng-sgs ===hng-sts ===hng-kkd
+ =>shuowen-jiguge =shuowen-jiguge
+ ===shuowen-jiguge4 ===shuowen-jiguge5 ===shuowen-jiguge-A30))
;; (string-match "^=adobe-" (symbol-name name))
)
"(%-18s . %05d)\t; %c")
=mj ==mj ===mj =>>mj =>mj
=zihai mojikyo))
"(%-18s . %06d)\t; %c")
- ((memq name '(=hanyo-denshi/tk ==hanyo-denshi/tk))
+ ((memq name '(=hanyo-denshi/tk =>hanyo-denshi/tk ==hanyo-denshi/tk))
"(%-18s . %08d)\t; %c")
((>= (charset-dimension name) 2)
"(%-18s . #x%04X)\t; %c")
line-breaking
ccss readable)
(setq attributes (delq '<-denotational@component attributes)))
+ (when (and (memq '<-denotational@usage attributes)
+ (setq value (get-char-attribute char '<-denotational@usage)))
+ (char-db-insert-relation-feature char '<-denotational@usage value
+ line-breaking
+ ccss readable)
+ (setq attributes (delq '<-denotational@usage attributes)))
(when (and (memq 'name attributes)
(setq value (get-char-attribute char 'name)))
(insert (format
(setq strokes value)))
(setq attributes (delq 'cns-strokes attributes))
)
- ;; (when (and (memq 'shinjigen-1-radical attributes)
- ;; (setq value (get-char-attribute char 'shinjigen-1-radical)))
- ;; (unless (eq value radical)
- ;; (insert (format "(shinjigen-1-radical . %S)\t; %c%s"
- ;; value
- ;; (ideographic-radical value)
- ;; line-breaking))
- ;; (or radical
- ;; (setq radical value)))
- ;; (setq attributes (delq 'shinjigen-1-radical attributes))
- ;; )
;; (when (and (memq 'ideographic- attributes)
;; (setq value (get-char-attribute char 'ideographic-)))
;; (insert "(ideographic- ")
line-breaking))
(setq attributes (delq '->mojikyo attributes))
)
- ;; (when (and (memq 'hanyu-dazidian-vol attributes)
- ;; (setq value (get-char-attribute char 'hanyu-dazidian-vol)))
- ;; (insert (format "(hanyu-dazidian-vol . %d)%s"
- ;; value line-breaking))
- ;; (setq attributes (delq 'hanyu-dazidian-vol attributes))
- ;; )
- ;; (when (and (memq 'hanyu-dazidian-page attributes)
- ;; (setq value (get-char-attribute char 'hanyu-dazidian-page)))
- ;; (insert (format "(hanyu-dazidian-page . %d)%s"
- ;; value line-breaking))
- ;; (setq attributes (delq 'hanyu-dazidian-page attributes))
- ;; )
- ;; (when (and (memq 'hanyu-dazidian-char attributes)
- ;; (setq value (get-char-attribute char 'hanyu-dazidian-char)))
- ;; (insert (format "(hanyu-dazidian-char . %d)%s"
- ;; value line-breaking))
- ;; (setq attributes (delq 'hanyu-dazidian-char attributes))
- ;; )
(unless readable
(dolist (ignored '(composition
->denotational <-subsumptive ->ucs-unified
- ->ideographic-component-forms))
+ ;; ->ideographic-component-forms
+ ))
(setq attributes (delq ignored attributes))))
(while attributes
(setq name (car attributes))
))
)
((or (eq name 'ideographic-structure)
+ (eq name 'ideographic-structure@apparent)
+ (eq name 'ideographic-structure@apparent/leftmost)
+ (eq name 'ideographic-structure@apparent/rightmost)
(eq name 'ideographic-combination)
;; (eq name 'ideographic-)
(eq name '=decomposition)
projects / chise / xemacs-chise.git / blobdiff