git.chise.org Git - chise/xemacs-chise.git.1/blob - lisp/utf-2000/char-db-util.el

   1 ;;; char-db-util.el --- Character Database utility -*- coding: utf-8-er; -*-
   2
   3 ;; Copyright (C) 1998,1999,2000,2001,2002,2003,2004,2005,2006,2007,
   4 ;;   2008,2009,2010,2011,2012,2013,2014,2015,2016,2017 MORIOKA Tomohiko.
   5
   6 ;; Author: MORIOKA Tomohiko <tomo@kanji.zinbun.kyoto-u.ac.jp>
   7 ;; Keywords: CHISE, Character Database, ISO/IEC 10646, UCS, Unicode, MULE.
   8
   9 ;; This file is part of XEmacs CHISE.
  10
  11 ;; XEmacs CHISE is free software; you can redistribute it and/or
  12 ;; modify it under the terms of the GNU General Public License as
  13 ;; published by the Free Software Foundation; either version 2, or (at
  14 ;; your option) any later version.
  15
  16 ;; XEmacs CHISE is distributed in the hope that it will be useful,
  17 ;; but WITHOUT ANY WARRANTY; without even the implied warranty of
  18 ;; MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
  19 ;; General Public License for more details.
  20
  21 ;; You should have received a copy of the GNU General Public License
  22 ;; along with XEmacs CHISE; see the file COPYING.  If not, write to
  23 ;; the Free Software Foundation, Inc., 59 Temple Place - Suite 330,
  24 ;; Boston, MA 02111-1307, USA.
  25
  26 ;;; Code:
  27
  28 (require 'chise-subr)
  29 (require 'ideograph-subr)
  30
  31 (defconst unidata-normative-category-alist
  32   '(("Lu" letter        uppercase)
  33     ("Ll" letter        lowercase)
  34     ("Lt" letter        titlecase)
  35     ("Mn" mark          non-spacing)
  36     ("Mc" mark          spacing-combining)
  37     ("Me" mark          enclosing)
  38     ("Nd" number        decimal-digit)
  39     ("Nl" number        letter)
  40     ("No" number        other)
  41     ("Zs" separator     space)
  42     ("Zl" separator     line)
  43     ("Zp" separator     paragraph)
  44     ("Cc" other         control)
  45     ("Cf" other         format)
  46     ("Cs" other         surrogate)
  47     ("Co" other         private-use)
  48     ("Cn" other         not-assigned)))
  49
  50 (defconst unidata-informative-category-alist
  51   '(("Lm" letter        modifier)
  52     ("Lo" letter        other)
  53     ("Pc" punctuation   connector)
  54     ("Pd" punctuation   dash)
  55     ("Ps" punctuation   open)
  56     ("Pe" punctuation   close)
  57     ("Pi" punctuation   initial-quote)
  58     ("Pf" punctuation   final-quote)
  59     ("Po" punctuation   other)
  60     ("Sm" symbol        math)
  61     ("Sc" symbol        currency)
  62     ("Sk" symbol        modifier)
  63     ("So" symbol        other)
  64     ))
  65
  66 (defconst shuowen-radicals
  67   [?一 ?上 ?示 ?三 ?王 ?玉 ?玨 ?气 ?士 ?丨 ; 010
  68    ?屮 ?艸 ?蓐 ?茻 ?小 ?八 ?釆 ?半 ?牛 ?犛 ; 020
  69    ?告 ?口 ?凵 ?吅 ?哭 ?走 ?止 ?癶 ?步 ?此 ; 030
  70    ?正 ?是 ?辵 ?彳 ?廴 ?㢟 ?行 ?齒 ?牙 ?足 ; 040
  71    ?疋 ?品 ?龠 ?冊 ?㗊 ?舌 ?干 ?谷 ?只 ?㕯 ; 050
  72    ?句 ?丩 ?古 ?十 ?卅 ?言 ?誩 ?音 ?䇂 ?丵 ; 060
  73    ?菐 ?𠬞 ?𠬜 ?共 ?異 ?舁 ?𦥑 ?䢅 ?爨 ?革 ; 070
  74    ?鬲 ?䰜 ?爪 ?𠃨 ?鬥 ?又 ?𠂇 ?㕜 ?支 ?𦘒 ; 080
  75    ?聿 ?畫 ?隶 ?臤 ?臣 ?殳 ?殺 ?𠘧 ?寸 ?皮 ; 090
  76    ?㼱 ?攴 ?敎 ?卜 ?用 ?爻 ?㸚 ?𥄎 ?目 ?䀠 ; 100
  77    ?眉 ?盾 ?自 ?白 ?鼻 ?皕 ?習 ?羽 ?隹 ?奞 ; 110
  78    ?萑 ?𦫳 ?苜 ?羊 ?羴 ?瞿 ?雔 ?雥 ?鳥 ?烏 ; 120
  79    ?𠦒 ?冓 ?幺 ?𢆶 ?叀 ?玄 ?予 ?放 ?𠬪 ?𣦼 ; 130
  80    ?歺 ?死 ?冎 ?骨 ?肉 ?筋 ?刀 ?刃 ?㓞 ?丰 ; 140
  81    ?耒 ?𧢲 ?竹 ?箕 ?丌 ?左 ?工 ?㠭 ?巫 ?甘 ; 150
  82    ?曰 ?乃 ?丂 ?可 ?兮 ?号 ?亏 ?旨 ?喜 ?壴 ; 160
  83 ;  ?旨 ?曰 ?乃 ?丂 ?可 ?兮 ?号 ?亏 ?喜 ?壴 ; 160
  84    ?鼓 ?豈 ?豆 ?豊 ?豐 ?䖒 ?虍 ?虎 ?虤 ?皿 ; 170
  85    ?𠙴 ?去 ?血 ?丶 ?丹 ?青 ?井 ?皀 ?鬯 ?食 ; 180
  86    ?亼 ?會 ?倉 ?入 ?缶 ?矢 ?高 ?冂 ?𩫏 ?京 ; 190
  87    ?亯 ?𣆪 ?畗 ?㐭 ?嗇 ?來 ?麥 ?夊 ?舛 ?䑞 ; 200
  88    ?韋 ?弟 ?夂 ?久 ?桀 ?木 ?東 ?林 ?才 ?叒 ; 210
  89    ?之 ?帀 ?出 ?𣎵 ?生 ?乇 ?𠂹 ?𠌶 ?華 ?𥝌 ; 220
  90    ?稽 ?巢 ?桼 ?束 ?㯻 ?囗 ?員 ?貝 ?邑 ?𨛜 ; 230
  91    ?日 ?旦 ?倝 ?㫃 ?冥 ?晶 ?月 ?有 ?明 ?囧 ; 240
  92    ?夕 ?多 ?毌 ?𢎘 ?𣐺 ?卣 ?齊 ?朿 ?片 ?鼎 ; 250
  93    ?克 ?彔 ?禾 ?秝 ?黍 ?香 ?米 ?毇 ?臼 ?凶 ; 260
  94    ?𣎳 ?林 ?麻 ?尗 ?耑 ?韭 ?瓜 ?瓠 ?宀 ?宮 ; 270
  95    ?呂 ?穴 ?㝱 ?𤕫 ?冖 ?𠔼 ?冃 ?㒳 ?网 ?襾 ; 280
  96    ?巾 ?巿 ?帛 ?白 ?㡀 ?黹 ?人 ?𠤎 ?匕 ?从 ; 290
  97    ?比 ?北 ?丘 ?㐺 ?𡈼 ?重 ?臥 ?身 ?㐆 ?衣 ; 300
  98    ?裘 ?老 ?毛 ?毳 ?尸 ?尺 ?尾 ?履 ?舟 ?方 ; 310
  99    ?儿 ?兄 ?兂 ?皃 ?𠑹 ?先 ?秃 ?見 ?覞 ?欠 ; 320
 100    ?㱃 ?㳄 ?旡 ?頁 ?𦣻 ?面 ?丏 ?首 ?𥄉 ?須 ; 330
 101    ?彡 ?彣 ?文 ?髟 ?后 ?司 ?卮 ?卩 ?印 ?色 ; 340
 102    ?𠨍 ?辟 ?勹 ?包 ?茍 ?鬼 ?甶 ?厶 ?嵬 ?山 ; 350
 103    ?屾 ?屵 ?广 ?厂 ?丸 ?危 ?石 ?長 ?勿 ?冄 ; 360
 104    ?而 ?豕 ?㣇 ?彑 ?豚 ?豸 ?𤉡 ?易 ?象 ?馬 ; 370
 105    ?𢊁 ?鹿 ?麤 ?㲋 ?兔 ?萈 ?犬 ?㹜 ?鼠 ?能 ; 380
 106    ?熊 ?火 ?炎 ?黑 ?囪 ?焱 ?炙 ?赤 ?大 ?亦 ; 390
 107    ])
 108
 109 (defun shuowen-radical (number)
 110   (aref shuowen-radicals (1- number)))
 111
 112 (defvar char-db-file-coding-system 'utf-8-mcs-er)
 113
 114 (defvar char-db-ignored-attributes '(ideographic-products))
 115
 116 (defvar char-db-coded-charset-priority-list
 117   '(ascii
 118     control-1
 119     latin-iso8859-1
 120     latin-iso8859-2
 121     latin-iso8859-3
 122     latin-iso8859-4
 123     latin-iso8859-9
 124     latin-jisx0201
 125     cyrillic-iso8859-5
 126     greek-iso8859-7
 127     thai-tis620
 128     ;; =mj
 129     ;; =adobe-japan1-0
 130     ;; =adobe-japan1-1
 131     ;; =adobe-japan1-2
 132     ;; =adobe-japan1-3
 133     ;; =adobe-japan1-4
 134     ;; =adobe-japan1-5
 135     ;; =adobe-japan1-6
 136     =jis-x0208
 137     =jis-x0208@1978
 138     =jis-x0208@1983
 139     japanese-jisx0212
 140     chinese-gb2312
 141     =jis-x0208@1990
 142     chinese-cns11643-1
 143     chinese-cns11643-2
 144     chinese-cns11643-3
 145     chinese-cns11643-4
 146     chinese-cns11643-5
 147     chinese-cns11643-6
 148     chinese-cns11643-7
 149     =jis-x0213-1
 150     =jis-x0213-1@2000
 151     =jis-x0213-1@2004
 152     =jis-x0213-2
 153     korean-ksc5601
 154     chinese-isoir165
 155     katakana-jisx0201
 156     hebrew-iso8859-8
 157     chinese-gb12345
 158     latin-viscii
 159     ethiopic-ucs
 160     =big5-cdp
 161     =hanyo-denshi/ja
 162     =hanyo-denshi/jb
 163     =hanyo-denshi/jc
 164     =hanyo-denshi/jd
 165     =hanyo-denshi/ft
 166     =hanyo-denshi/ia
 167     =hanyo-denshi/ib
 168     =hanyo-denshi/hg
 169     =hanyo-denshi/jt
 170     =hanyo-denshi/ks
 171     =hanyo-denshi/tk
 172     ;; ==mj
 173     ;; ==adobe-japan1-0
 174     ;; ==adobe-japan1-1
 175     ;; ==adobe-japan1-2
 176     ;; ==adobe-japan1-3
 177     ;; ==adobe-japan1-4
 178     ;; ==adobe-japan1-5
 179     ;; ==adobe-japan1-6
 180     ==jis-x0208
 181     ==jis-x0213-1
 182     ==jis-x0213-2
 183     ==jis-x0212
 184     ==hanyo-denshi/ja
 185     ==hanyo-denshi/jb
 186     ==hanyo-denshi/jc
 187     ==hanyo-denshi/ft
 188     ==hanyo-denshi/ia
 189     ==hanyo-denshi/ib
 190     ==hanyo-denshi/hg
 191     ==hanyo-denshi/jt
 192     ==hanyo-denshi/ks
 193     ==hanyo-denshi/tk
 194     =gt
 195     =gt-k
 196     =daikanwa
 197     =daikanwa@rev2
 198     =daikanwa@rev1
 199     =daikanwa/+p
 200     ==daikanwa
 201     =cbeta
 202     ideograph-hanziku-1
 203     ideograph-hanziku-2
 204     ideograph-hanziku-3
 205     ideograph-hanziku-4
 206     ideograph-hanziku-5
 207     ideograph-hanziku-6
 208     ideograph-hanziku-7
 209     ideograph-hanziku-8
 210     ideograph-hanziku-9
 211     ideograph-hanziku-10
 212     ideograph-hanziku-11
 213     ideograph-hanziku-12
 214     ;; =>>>adobe-japan1-0
 215     ;; =>>>adobe-japan1-1
 216     ;; =>>>adobe-japan1-2
 217     ;; =>>>adobe-japan1-3
 218     ;; =>>>adobe-japan1-4
 219     ;; =>>>adobe-japan1-5
 220     ;; =>>>adobe-japan1-6
 221     ;; =>>>jis-x0208
 222     ;; =>>>jis-x0213-1
 223     ;; =>>>jis-x0213-2
 224     ;; =>>>hanyo-denshi/ja
 225     ;; =>>>hanyo-denshi/jb
 226     ;; =>>>hanyo-denshi/jc
 227     ;; =>>>hanyo-denshi/ft
 228     ;; =>>>hanyo-denshi/ib
 229     ;; =>>>hanyo-denshi/hg
 230     ;; =>>>hanyo-denshi/jt
 231     ;; =>>>hanyo-denshi/ks
 232     ;; =>>>gt
 233     =>>adobe-japan1-0
 234     =>>adobe-japan1-1
 235     =>>adobe-japan1-2
 236     =>>adobe-japan1-3
 237     =>>adobe-japan1-4
 238     =>>adobe-japan1-5
 239     =>>adobe-japan1-6
 240     =>>jis-x0208
 241     =>>jis-x0213-1
 242     =>>jis-x0213-1@2000
 243     =>>jis-x0213-1@2004
 244     =>>jis-x0213-2
 245     =>>jis-x0208@1978
 246     =>>hanyo-denshi/ft
 247     =>>hanyo-denshi/jt
 248     =>>hanyo-denshi/ks
 249     =>>gt
 250     =>>daikanwa
 251     =>>cbeta
 252     =+>jis-x0208
 253     =+>jis-x0213-1
 254     =+>jis-x0213-2
 255     =+>adobe-japan1-0
 256     =+>adobe-japan1-1
 257     =+>adobe-japan1-2
 258     =+>adobe-japan1-3
 259     =+>adobe-japan1-4
 260     =+>adobe-japan1-5
 261     =+>adobe-japan1-6
 262     =+>jis-x0208@1978
 263     =>jis-x0208
 264     =>jis-x0208@1997
 265     =>jis-x0213-1
 266     =>jis-x0213-1@2000
 267     =>jis-x0213-1@2004
 268     =>jis-x0213-2
 269     ==>ucs@bucs
 270     =>iwds-1
 271     ;; =>ucs@hanyo-denshi
 272     =>ucs@iso
 273     =>ucs@unicode
 274     =>ucs@jis
 275     =>ucs@cns
 276     =>ucs@ks
 277     =+>ucs@iso
 278     =+>ucs@unicode
 279     =+>ucs@jis
 280     =+>ucs@jis/1990
 281     =+>ucs@cns
 282     =+>ucs@ks
 283     =>>ucs@iso
 284     =>>ucs@unicode
 285     =>>ucs@jis
 286     =>>ucs@cns
 287     =>>>ucs@iso
 288     =>>>ucs@unicode
 289     ==ucs@iso
 290     ==ucs@unicode
 291     ;; ==ucs@cns
 292     ==gb2312
 293     ==ks-x1001
 294     ==cns11643-1
 295     ==cns11643-2
 296     ==cns11643-3
 297     ==cns11643-4
 298     ==cns11643-5
 299     ==cns11643-6
 300     ==cns11643-7
 301     ==gt
 302     ==jis-x0208@1990
 303     ;; ==jis-x0208@1983
 304     ==jis-x0208@1978
 305     ==gt-k
 306     =ucs@iso
 307     =ucs@unicode
 308     =ucs@cns
 309     ==big5-cdp
 310     ==cbeta
 311     =>>big5-cdp
 312     =>>gt-k
 313     =+>gt
 314     =>gt
 315     =>mj
 316     =>big5-cdp
 317     =>daikanwa
 318     =>daikanwa/ho
 319     =>cns11643-5
 320     =>cns11643-7
 321     =big5
 322     =big5-eten
 323     =>gt-k
 324     =zinbun-oracle
 325     =>zinbun-oracle
 326     =ruimoku-v6
 327     =>>ruimoku-v6
 328     ==ruimoku-v6
 329     =jef-china3
 330     =>cbeta
 331     =shinjigen
 332     =ucs-var-001
 333     =ucs-var-002
 334     =ucs-var-003
 335     =ucs-var-004
 336     =ucs-var-005
 337     =ucs-var-006
 338     =ucs-var-008
 339     =ucs-var-010
 340     =ucs-itaiji-001
 341     =ucs-itaiji-002
 342     =ucs-itaiji-003
 343     =ucs-itaiji-004
 344     =ucs-itaiji-005
 345     =ucs-itaiji-006
 346     =ucs-itaiji-007
 347     =ucs-itaiji-008
 348     =ucs-itaiji-009
 349     =ucs-itaiji-010
 350     =ucs-itaiji-011
 351     =>ucs-itaiji-001
 352     =>ucs-itaiji-002
 353     =>ucs-itaiji-003
 354     =>ucs-itaiji-004
 355     =>ucs-itaiji-005
 356     =>ucs-itaiji-006
 357     =>ucs-itaiji-007
 358     =big5-cdp-var-001
 359     =big5-cdp-var-002
 360     =big5-cdp-var-003
 361     =big5-cdp-var-004
 362     =big5-cdp-var-005
 363     =big5-cdp-itaiji-001
 364     =big5-cdp-itaiji-002
 365     =big5-cdp-itaiji-003
 366     =>big5-cdp-itaiji-001
 367     =>ucs@iwds-1
 368     =>ucs@cognate
 369     =>ucs@component
 370     ==ucs@gb
 371     =ucs@gb
 372     ==ucs-var-002
 373     =ucs@JP/hanazono))
 374
 375
 376 ;;; @ char-db formatters
 377 ;;;
 378
 379 (defun char-db-make-char-spec (char)
 380   (let (ret char-spec)
 381     (cond ((characterp char)
 382            (cond ((and (setq ret (encode-char char '=ucs 'defined-only))
 383                        (not (and (<= #xE000 ret)(<= ret #xF8FF))))
 384                   (setq char-spec (list (cons '=ucs ret)))
 385                   (cond ((setq ret (get-char-attribute char 'name))
 386                          (setq char-spec (cons (cons 'name ret) char-spec))
 387                          )
 388                         ((setq ret (get-char-attribute char 'name*))
 389                          (setq char-spec (cons (cons 'name* ret) char-spec))
 390                          ))
 391                   )
 392                  ((encode-char char '=mj 'defined-only)
 393                   (setq char-spec nil)
 394                   (dolist (ccs (charset-list))
 395                     (if (and (or (eq ccs '=mj)
 396                                  ;; (eq (charset-property ccs 'iso-ir) 177)
 397                                  (string-match "=ucs@" (symbol-name ccs))
 398                                  )
 399                              (setq ccs (charset-name ccs))
 400                              (null (assq ccs char-spec))
 401                              (setq ret (encode-char char ccs 'defined-only)))
 402                         (setq char-spec (cons (cons ccs ret) char-spec))))
 403                   )
 404                  ((encode-char char '==mj 'defined-only)
 405                   (setq char-spec nil)
 406                   (dolist (ccs (charset-list))
 407                     (if (and (or (eq ccs '==mj)
 408                                  ;; (eq (charset-property ccs 'iso-ir) 177)
 409                                  (string-match "=ucs@" (symbol-name ccs))
 410                                  )
 411                              (setq ccs (charset-name ccs))
 412                              (null (assq ccs char-spec))
 413                              (setq ret (encode-char char ccs 'defined-only)))
 414                         (setq char-spec (cons (cons ccs ret) char-spec))))
 415                   )
 416                  ((encode-char char '=adobe-japan1 'defined-only)
 417                   (setq char-spec nil)
 418                   (dolist (ccs (charset-list))
 419                     (if (and (or (memq ccs
 420                                        '(=adobe-japan1-0
 421                                          =adobe-japan1-1
 422                                          =adobe-japan1-2
 423                                          =adobe-japan1-3
 424                                          =adobe-japan1-4
 425                                          =adobe-japan1-5
 426                                          =adobe-japan1-6
 427                                          ))
 428                                  ;; (eq (charset-property ccs 'iso-ir) 177)
 429                                  (string-match "=ucs@" (symbol-name ccs))
 430                                  )
 431                              (setq ccs (charset-name ccs))
 432                              (null (assq ccs char-spec))
 433                              (setq ret (encode-char char ccs 'defined-only)))
 434                         (setq char-spec (cons (cons ccs ret) char-spec))))
 435                   )
 436                  ((encode-char char '==adobe-japan1 'defined-only)
 437                   (setq char-spec nil)
 438                   (dolist (ccs (charset-list))
 439                     (if (and (or (memq ccs
 440                                        '(==adobe-japan1-0
 441                                          ==adobe-japan1-1
 442                                          ==adobe-japan1-2
 443                                          ==adobe-japan1-3
 444                                          ==adobe-japan1-4
 445                                          ==adobe-japan1-5
 446                                          ==adobe-japan1-6
 447                                          ))
 448                                  ;; (eq (charset-property ccs 'iso-ir) 177)
 449                                  (string-match "=ucs@" (symbol-name ccs))
 450                                  )
 451                              (setq ccs (charset-name ccs))
 452                              (null (assq ccs char-spec))
 453                              (setq ret (encode-char char ccs 'defined-only)))
 454                         (setq char-spec (cons (cons ccs ret) char-spec))))
 455                   )
 456                  ((setq ret
 457                         (catch 'tag
 458                           (let ((rest char-db-coded-charset-priority-list)
 459                                 ccs)
 460                             (while rest
 461                               (setq ccs (charset-name
 462                                          (find-charset (car rest))))
 463                               (if (setq ret
 464                                         (encode-char char ccs
 465                                                      'defined-only))
 466                                   (throw 'tag (cons ccs ret)))
 467                               (setq rest (cdr rest))))))
 468                   (setq char-spec (list ret))
 469                   (dolist (ccs (delq (car ret) (charset-list)))
 470                     (if (and (or (charset-iso-final-char ccs)
 471                                  (memq ccs
 472                                        '(=daikanwa
 473                                          =daikanwa@rev2
 474                                          ;; =gt-k
 475                                          =jis-x0208@1997
 476                                          ))
 477                                  (eq (charset-property ccs 'iso-ir) 177)
 478                                  ;; (string-match "=ucs@" (symbol-name ccs))
 479                                  )
 480                              (setq ccs (charset-name ccs))
 481                              (null (assq ccs char-spec))
 482                              (setq ret (encode-char char ccs 'defined-only)))
 483                         (setq char-spec (cons (cons ccs ret) char-spec))))
 484                   (if (null char-spec)
 485                       (setq char-spec (split-char char)))
 486                   (cond ((setq ret (get-char-attribute char 'name))
 487                          (setq char-spec (cons (cons 'name ret) char-spec))
 488                          )
 489                         ((setq ret (get-char-attribute char 'name*))
 490                          (setq char-spec (cons (cons 'name* ret) char-spec))
 491                          ))
 492                   )
 493                  ((setq ret (get-char-attribute
 494                              char 'ideographic-combination))
 495                   (setq char-spec
 496                         (cons (cons 'ideographic-combination ret)
 497                               char-spec))
 498                   ))
 499            char-spec)
 500           ((consp char)
 501            char))))
 502
 503 (defun char-db-insert-char-spec (char &optional readable column
 504                                       required-features)
 505   (unless column
 506     (setq column (current-column)))
 507   (let (char-spec temp-char)
 508     (setq char-spec (char-db-make-char-spec char))
 509     (unless (or (characterp char) ; char
 510                 (condition-case nil
 511                     (setq char (find-char char-spec))
 512                   (error nil)))
 513       ;; define temporary character
 514       ;;   Current implementation is dirty.
 515       (setq temp-char (define-char (cons '(ideograph-daikanwa . 0)
 516                                          char-spec)))
 517       (remove-char-attribute temp-char 'ideograph-daikanwa)
 518       (setq char temp-char))
 519     (insert-char-attributes char
 520                             readable
 521                             (union (mapcar #'car char-spec)
 522                                    required-features)
 523                             nil 'for-sub-node)
 524     (when temp-char
 525       ;; undefine temporary character
 526       ;;   Current implementation is dirty.
 527       (setq char-spec (char-attribute-alist temp-char))
 528       (while char-spec
 529         (remove-char-attribute temp-char (car (car char-spec)))
 530         (setq char-spec (cdr char-spec))))))
 531
 532 (defun char-db-insert-alist (alist &optional readable column)
 533   (unless column
 534     (setq column (current-column)))
 535   (let ((line-breaking
 536          (concat "\n" (make-string (1+ column) ?\ )))
 537         name value
 538         ret al ; cal
 539         key
 540         lbs cell rest separator)
 541     (insert "(")
 542     (while alist
 543       (setq name (car (car alist))
 544             value (cdr (car alist)))
 545       (cond ((eq name 'char)
 546              (insert "(char . ")
 547              (if (setq ret (condition-case nil
 548                                (find-char value)
 549                              (error nil)))
 550                  (progn
 551                    (setq al nil
 552                          ;; cal nil
 553                          )
 554                    (while value
 555                      (setq key (car (car value)))
 556                      ;; (if (find-charset key)
 557                      ;;     (setq cal (cons key cal))
 558                      (setq al (cons key al))
 559                      ;; )
 560                      (setq value (cdr value)))
 561                    (insert-char-attributes ret
 562                                            readable
 563                                            (or al 'none) ; cal
 564                                            nil 'for-sub-node))
 565                (insert (prin1-to-string value)))
 566              (insert ")")
 567              (insert line-breaking))
 568             ((consp value)
 569              (insert (format "(%-18s " name))
 570              (setq lbs (concat "\n" (make-string (current-column) ?\ )))
 571              (while (consp value)
 572                (setq cell (car value))
 573                (if (and (consp cell)
 574                         (consp (car cell))
 575                         (setq ret (condition-case nil
 576                                       (find-char cell)
 577                                     (error nil)))
 578                         )
 579                    (progn
 580                      (setq rest cell
 581                            al nil
 582                            ;; cal nil
 583                            )
 584                      (while rest
 585                        (setq key (car (car rest)))
 586                        ;; (if (find-charset key)
 587                        ;;     (setq cal (cons key cal))
 588                        (setq al (cons key al))
 589                        ;; )
 590                        (setq rest (cdr rest)))
 591                      (if separator
 592                          (insert lbs))
 593                      (insert-char-attributes ret
 594                                              readable
 595                                              al ; cal
 596                                              nil 'for-sub-node)
 597                      (setq separator lbs))
 598                  (if separator
 599                      (insert separator))
 600                  (insert (prin1-to-string cell))
 601                  (setq separator " "))
 602                (setq value (cdr value)))
 603              (insert ")")
 604              (insert line-breaking))
 605             (t
 606              (insert (format "(%-18s . %S)%s"
 607                              name value
 608                              line-breaking))))
 609       (setq alist (cdr alist))))
 610   (insert ")"))
 611
 612 (defun char-db-insert-char-reference (plist &optional readable column)
 613   (unless column
 614     (setq column (current-column)))
 615   (let ((line-breaking
 616          (concat "\n" (make-string (1+ column) ?\ )))
 617         (separator "")
 618         name value)
 619     (insert "(")
 620     (while plist
 621       (setq name (pop plist))
 622       (setq value (pop plist))
 623       (cond ((eq name :char)
 624              (insert separator)
 625              (insert ":char\t")
 626              (cond ((numberp value)
 627                     (setq value (decode-char '=ucs value)))
 628                    ;; ((consp value)
 629                    ;;  (setq value (or (find-char value)
 630                    ;;                  value)))
 631                    )
 632              (char-db-insert-char-spec value readable)
 633              (insert line-breaking)
 634              (setq separator ""))
 635             ((eq name :radical)
 636              (insert (format "%s%s\t%d ; %c%s"
 637                              separator
 638                              name value
 639                              (ideographic-radical value)
 640                              line-breaking))
 641              (setq separator ""))
 642             (t
 643              (insert (format "%s%s\t%S" separator name value))
 644              (setq separator line-breaking)))
 645       ))
 646   (insert ")"))
 647
 648 (defun char-db-decode-isolated-char (ccs code-point)
 649   (let (ret)
 650     (setq ret
 651           (cond ((eq ccs 'arabic-iso8859-6)
 652                  (decode-char ccs code-point))
 653                 ;; ((eq ccs '=gt)
 654                 ;;  (decode-builtin-char '==gt code-point))
 655                 ((and (memq ccs '(=gt-pj-1
 656                                   =gt-pj-2
 657                                   =gt-pj-3
 658                                   =gt-pj-4
 659                                   =gt-pj-5
 660                                   =gt-pj-6
 661                                   =gt-pj-7
 662                                   =gt-pj-8
 663                                   =gt-pj-9
 664                                   =gt-pj-10
 665                                   =gt-pj-11))
 666                       (setq ret (decode-char ccs code-point))
 667                       (setq ret (encode-char ret '=gt 'defined-only)))
 668                  (decode-builtin-char '=gt ret))
 669                 (t
 670                  (decode-builtin-char ccs code-point))))
 671     (cond ((null ret)
 672            (or (decode-char ccs code-point)
 673                (define-char (list (cons ccs code-point)))))
 674           ((and (<= 0 (char-int ret))
 675                 (<= (char-int ret) #x1F))
 676            (decode-char '=ucs (+ #x2400 (char-int ret))))
 677           ((= (char-int ret) #x7F)
 678            ?\u2421)
 679           (t ret))))
 680
 681 (defvar char-db-convert-obsolete-format t)
 682
 683 (defun char-db-insert-ccs-feature (name value line-breaking)
 684   (cond
 685    ((integerp value)
 686     (insert
 687      (format
 688       (cond
 689        ((memq name '(=>iwds-1
 690                      ==shinjigen
 691                      =shinjigen
 692                      =shinjigen@1ed ==shinjigen@1ed
 693                      =shinjigen@rev ==shinjigen@rev
 694                      =shinjigen/+p@rev ==shinjigen/+p@rev
 695                      ===daikanwa/ho ==daikanwa/ho
 696                      =daikanwa/ho =>>daikanwa/ho =>daikanwa/ho))
 697         "(%-18s .  %04d)\t; %c")
 698        ((eq name '=shinjigen@1ed/24pr)
 699         "(%-18s . %04d)\t; %c")
 700        ((or
 701          (memq name
 702                '(===daikanwa
 703                  ==daikanwa =daikanwa =>>daikanwa =>daikanwa
 704                  =daikanwa@rev1 =daikanwa@rev2
 705                  =daikanwa/+p ==daikanwa/+p ===daikanwa/+p
 706                  =>>daikanwa/+p
 707                  =daikanwa/+2p =>>daikanwa/+2p
 708                  =gt ==gt ===gt
 709                  =>>gt =+>gt =>gt
 710                  =gt-k ==gt-k ===gt-k
 711                  =>>gt-k =>gt-k
 712                  =adobe-japan1-0 ==adobe-japan1-0 ===adobe-japan1-0
 713                  =adobe-japan1-1 ==adobe-japan1-1 ===adobe-japan1-1
 714                  =adobe-japan1-2 ==adobe-japan1-2 ===adobe-japan1-2
 715                  =adobe-japan1-3 ==adobe-japan1-3 ===adobe-japan1-3
 716                  =adobe-japan1-4 ==adobe-japan1-4 ===adobe-japan1-4
 717                  =adobe-japan1-5 ==adobe-japan1-5 ===adobe-japan1-5
 718                  =adobe-japan1-6 ==adobe-japan1-6 ===adobe-japan1-6
 719                  =>>adobe-japan1-0 =+>adobe-japan1-0
 720                  =>>adobe-japan1-1 =+>adobe-japan1-1
 721                  =>>adobe-japan1-2 =+>adobe-japan1-2
 722                  =>>adobe-japan1-3 =+>adobe-japan1-3
 723                  =>>adobe-japan1-4 =+>adobe-japan1-4
 724                  =>>adobe-japan1-5 =+>adobe-japan1-5
 725                  =>>adobe-japan1-6 =+>adobe-japan1-6
 726                  =>cbeta =cbeta =>>cbeta ==cbeta ===cbeta
 727                  =zinbun-oracle =>zinbun-oracle
 728                  ===hng-jou ===hng-keg ===hng-dng ===hng-mam
 729                  ===hng-drt ===hng-kgk ===hng-myz ===hng-kda
 730                  ===hng-khi ===hng-khm ===hng-hok ===hng-kyd ===hng-sok
 731                  ===hng-yhk ===hng-kak ===hng-kar ===hng-kae
 732                  ===hng-sys ===hng-tsu ===hng-tzj
 733                  ===hng-hos ===hng-nak ===hng-jhk
 734                  ===hng-hod ===hng-gok ===hng-ink ===hng-nto
 735                  ===hng-nkm ===hng-k24 ===hng-nkk
 736                  ===hng-kcc ===hng-kcj ===hng-kbk ===hng-sik
 737                  ===hng-skk ===hng-kyu ===hng-ksk ===hng-wan
 738                  ===hng-okd ===hng-wad ===hng-kmi
 739                  ===hng-zkd ===hng-doh ===hng-jyu
 740                  ===hng-tzs ===hng-kss ===hng-kyo
 741                  ===hng-smk))
 742          ;; (string-match "^=adobe-" (symbol-name name))
 743          )
 744         "(%-18s . %05d)\t; %c")
 745        ((memq name '(=hanyo-denshi/ks
 746                      ==hanyo-denshi/ks ===hanyo-denshi/ks
 747                      =>>hanyo-denshi/ks
 748                      =koseki ==koseki
 749                      =mj ==mj ===mj =>>mj =>mj
 750                      =zihai mojikyo))
 751         "(%-18s . %06d)\t; %c")
 752        ((memq name '(=hanyo-denshi/tk ==hanyo-denshi/tk))
 753         "(%-18s . %08d)\t; %c")
 754        ((>= (charset-dimension name) 2)
 755         "(%-18s . #x%04X)\t; %c")
 756        (t
 757         "(%-18s . #x%02X)\t; %c"))
 758       name
 759       (if (= (charset-iso-graphic-plane name) 1)
 760           (logior value
 761                   (cond ((= (charset-dimension name) 1)
 762                          #x80)
 763                         ((= (charset-dimension name) 2)
 764                          #x8080)
 765                         ((= (charset-dimension name) 3)
 766                          #x808080)
 767                         (t 0)))
 768         value)
 769       (char-db-decode-isolated-char name value)))
 770     (if (and (= (charset-chars name) 94)
 771              (= (charset-dimension name) 2))
 772         (insert (format " [%02d-%02d]"
 773                         (- (lsh value -8) 32)
 774                         (- (logand value 255) 32))))
 775     )
 776    (t
 777     (insert (format "(%-18s . %s)" name value))
 778     ))
 779   (insert line-breaking))
 780
 781 (defun char-db-insert-relation-feature (char name value line-breaking
 782                                              ccss readable)
 783   (insert (format "(%-18s%s " name line-breaking))
 784   (let ((lbs (concat "\n" (make-string (current-column) ?\ )))
 785         separator cell sources required-features
 786         ret)
 787     (while (consp value)
 788       (setq cell (car value))
 789       (if (integerp cell)
 790           (setq cell (decode-char '=ucs cell)))
 791       (cond
 792        ((eq name '->subsumptive)
 793         (when (or (not (some (lambda (atr)
 794                                (get-char-attribute cell atr))
 795                              char-db-ignored-attributes))
 796                   (some (lambda (ccs)
 797                           (encode-char cell ccs 'defined-only))
 798                         ccss))
 799           (if separator
 800               (insert lbs))
 801           (let ((char-db-ignored-attributes
 802                  (cons '<-subsumptive
 803                        char-db-ignored-attributes)))
 804             (insert-char-attributes cell readable nil nil 'for-sub-node))
 805           (setq separator lbs))
 806         )
 807        ((characterp cell)
 808         (setq sources
 809               (get-char-attribute
 810                char (intern (format "%s*sources" name))))
 811         (setq required-features nil)
 812         (dolist (source sources)
 813           (cond
 814            ((memq source '(JP
 815                            JP/Jouyou
 816                            shinjigen shinjigen@1ed shinjigen@rev))
 817             (setq required-features
 818                   (union required-features
 819                          '(=jis-x0208
 820                            =jis-x0208@1990
 821                            =jis-x0213-1@2000
 822                            =jis-x0213-1@2004
 823                            =jis-x0213-2
 824                            =jis-x0212
 825                            =jis-x0208@1983
 826                            =jis-x0208@1978
 827                            =shinjigen
 828                            =shinjigen@1ed
 829                            =shinjigen@rev
 830                            =shinjigen/+p@rev))))
 831            ((eq source 'CN)
 832             (setq required-features
 833                   (union required-features
 834                          '(=gb2312
 835                            =gb12345
 836                            =iso-ir165)))))
 837           (cond
 838            ((find-charset (setq ret (intern (format "=%s" source))))
 839             (setq required-features
 840                   (cons ret required-features)))
 841            (t (setq required-features
 842                     (cons source required-features)))))
 843         (cond ((string-match "@JP" (symbol-name name))
 844                (setq required-features
 845                      (union required-features
 846                             '(=jis-x0208
 847                               =jis-x0208@1990
 848                               =jis-x0213-1-2000
 849                               =jis-x0213-2-2000
 850                               =jis-x0212
 851                               =jis-x0208@1983
 852                               =jis-x0208@1978))))
 853               ((string-match "@CN" (symbol-name name))
 854                (setq required-features
 855                      (union required-features
 856                             '(=gb2312
 857                               =gb12345
 858                               =iso-ir165)))))
 859         (if separator
 860             (insert lbs))
 861         (if readable
 862             (insert (format "%S" cell))
 863           (char-db-insert-char-spec cell readable
 864                                     nil
 865                                     required-features))
 866         (setq separator lbs))
 867        ((consp cell)
 868         (if separator
 869             (insert lbs))
 870         (if (consp (car cell))
 871             (char-db-insert-char-spec cell readable)
 872           (char-db-insert-char-reference cell readable))
 873         (setq separator lbs))
 874        (t
 875         (if separator
 876             (insert separator))
 877         (insert (prin1-to-string cell))
 878         (setq separator " ")))
 879       (setq value (cdr value)))
 880     (insert ")")
 881     (insert line-breaking)))
 882
 883 (defun insert-char-attributes (char &optional readable attributes column
 884                                     for-sub-node)
 885   (unless column
 886     (setq column (current-column)))
 887   (let (name value ; has-long-ccs-name
 888         rest
 889         radical strokes
 890         (line-breaking
 891          (concat "\n" (make-string (1+ column) ?\ )))
 892         lbs cell separator ret
 893         key al cal
 894         dest-ccss ; sources required-features
 895         ccss)
 896     (let (atr-d)
 897       (setq attributes
 898             (sort (if attributes
 899                       (if (consp attributes)
 900                           (progn
 901                             (dolist (name attributes)
 902                               (unless (memq name char-db-ignored-attributes)
 903                                 (if (find-charset name)
 904                                     (push name ccss))
 905                                 (push name atr-d)))
 906                             atr-d))
 907                     (dolist (name (char-attribute-list))
 908                       (unless (memq name char-db-ignored-attributes)
 909                         (if (find-charset name)
 910                             (push name ccss))
 911                         (push name atr-d)))
 912                     atr-d)
 913                   #'char-attribute-name<)))
 914     (insert "(")
 915     (when (memq '<-subsumptive attributes)
 916       (when (or readable (not for-sub-node))
 917         (when (setq value (get-char-attribute char '<-subsumptive))
 918           (char-db-insert-relation-feature char '<-subsumptive value
 919                                            line-breaking
 920                                            ccss readable)))
 921       (setq attributes (delq '<-subsumptive attributes)))
 922     (when (and (memq '<-denotational attributes)
 923                (setq value (get-char-attribute char '<-denotational)))
 924       (char-db-insert-relation-feature char '<-denotational value
 925                                        line-breaking
 926                                        ccss readable)
 927       (setq attributes (delq '<-denotational attributes)))
 928     (when (and (memq '<-denotational@component attributes)
 929                (setq value (get-char-attribute char '<-denotational@component)))
 930       (char-db-insert-relation-feature char '<-denotational@component value
 931                                        line-breaking
 932                                        ccss readable)
 933       (setq attributes (delq '<-denotational@component attributes)))
 934     (when (and (memq 'name attributes)
 935                (setq value (get-char-attribute char 'name)))
 936       (insert (format
 937                (if (> (+ (current-column) (length value)) 48)
 938                    "(name . %S)%s"
 939                  "(name               . %S)%s")
 940                value line-breaking))
 941       (setq attributes (delq 'name attributes))
 942       )
 943     (when (and (memq 'name* attributes)
 944                (setq value (get-char-attribute char 'name*)))
 945       (insert (format
 946                (if (> (+ (current-column) (length value)) 48)
 947                    "(name* . %S)%s"
 948                  "(name*              . %S)%s")
 949                value line-breaking))
 950       (setq attributes (delq 'name* attributes))
 951       )
 952     (when (and (memq 'script attributes)
 953                (setq value (get-char-attribute char 'script)))
 954       (insert (format "(script\t\t%s)%s"
 955                       (mapconcat (function prin1-to-string)
 956                                  value " ")
 957                       line-breaking))
 958       (setq attributes (delq 'script attributes))
 959       )
 960     (dolist (name '(=>ucs =>ucs*))
 961       (when (and (memq name attributes)
 962                  (setq value (get-char-attribute char name)))
 963         (insert (format "(%-18s . #x%04X)\t; %c%s"
 964                         name value (decode-char '=ucs value)
 965                         line-breaking))
 966         (setq attributes (delq name attributes))))
 967     (dolist (name '(=>ucs@gb =>ucs@big5))
 968       (when (and (memq name attributes)
 969                  (setq value (get-char-attribute char name)))
 970         (insert (format "(%-18s . #x%04X)\t; %c%s"
 971                         name value
 972                         (decode-char (intern
 973                                       (concat "="
 974                                               (substring
 975                                                (symbol-name name) 2)))
 976                                      value)
 977                         line-breaking))
 978         (setq attributes (delq name attributes))
 979         ))
 980     ;; (dolist (name '(=>daikanwa))
 981     ;;   (when (and (memq name attributes)
 982     ;;              (setq value (get-char-attribute char name)))
 983     ;;     (insert
 984     ;;      (if (integerp value)
 985     ;;          (format "(%-18s . %05d)\t; %c%s"
 986     ;;                  name value (decode-char '=daikanwa value)
 987     ;;                  line-breaking)
 988     ;;        (format "(%-18s %s)\t; %c%s"
 989     ;;                name
 990     ;;                (mapconcat (function prin1-to-string)
 991     ;;                           value " ")
 992     ;;                (char-representative-of-daikanwa char)
 993     ;;                line-breaking)))
 994     ;;     (setq attributes (delq name attributes))))
 995     (when (and (memq 'general-category attributes)
 996                (setq value (get-char-attribute char 'general-category)))
 997       (insert (format
 998                "(general-category\t%s) ; %s%s"
 999                (mapconcat (lambda (cell)
1000                             (format "%S" cell))
1001                           value " ")
1002                (cond ((rassoc value unidata-normative-category-alist)
1003                       "Normative Category")
1004                      ((rassoc value unidata-informative-category-alist)
1005                       "Informative Category")
1006                      (t
1007                       "Unknown Category"))
1008                line-breaking))
1009       (setq attributes (delq 'general-category attributes))
1010       )
1011     (when (and (memq 'bidi-category attributes)
1012                (setq value (get-char-attribute char 'bidi-category)))
1013       (insert (format "(bidi-category\t. %S)%s"
1014                       value
1015                       line-breaking))
1016       (setq attributes (delq 'bidi-category attributes))
1017       )
1018     (unless (or (not (memq 'mirrored attributes))
1019                 (eq (setq value (get-char-attribute char 'mirrored 'empty))
1020                     'empty))
1021       (insert (format "(mirrored\t\t. %S)%s"
1022                       value
1023                       line-breaking))
1024       (setq attributes (delq 'mirrored attributes))
1025       )
1026     (cond
1027      ((and (memq 'decimal-digit-value attributes)
1028            (setq value (get-char-attribute char 'decimal-digit-value)))
1029       (insert (format "(decimal-digit-value . %S)%s"
1030                       value
1031                       line-breaking))
1032       (setq attributes (delq 'decimal-digit-value attributes))
1033       (when (and (memq 'digit-value attributes)
1034                  (setq value (get-char-attribute char 'digit-value)))
1035         (insert (format "(digit-value\t . %S)%s"
1036                         value
1037                         line-breaking))
1038         (setq attributes (delq 'digit-value attributes))
1039         )
1040       (when (and (memq 'numeric-value attributes)
1041                  (setq value (get-char-attribute char 'numeric-value)))
1042         (insert (format "(numeric-value\t . %S)%s"
1043                         value
1044                         line-breaking))
1045         (setq attributes (delq 'numeric-value attributes))
1046         )
1047       )
1048      (t
1049       (when (and (memq 'digit-value attributes)
1050                  (setq value (get-char-attribute char 'digit-value)))
1051         (insert (format "(digit-value\t. %S)%s"
1052                         value
1053                         line-breaking))
1054         (setq attributes (delq 'digit-value attributes))
1055         )
1056       (when (and (memq 'numeric-value attributes)
1057                  (setq value (get-char-attribute char 'numeric-value)))
1058         (insert (format "(numeric-value\t. %S)%s"
1059                         value
1060                         line-breaking))
1061         (setq attributes (delq 'numeric-value attributes))
1062         )))
1063     (when (and (memq 'iso-10646-comment attributes)
1064                (setq value (get-char-attribute char 'iso-10646-comment)))
1065       (insert (format "(iso-10646-comment\t. %S)%s"
1066                       value
1067                       line-breaking))
1068       (setq attributes (delq 'iso-10646-comment attributes))
1069       )
1070     (when (and (memq 'morohashi-daikanwa attributes)
1071                (setq value (get-char-attribute char 'morohashi-daikanwa)))
1072       (insert (format "(morohashi-daikanwa\t%s)%s"
1073                       (mapconcat (function prin1-to-string) value " ")
1074                       line-breaking))
1075       (setq attributes (delq 'morohashi-daikanwa attributes))
1076       )
1077     (setq radical nil
1078           strokes nil)
1079     (when (and (memq 'ideographic-radical attributes)
1080                (setq value (get-char-attribute char 'ideographic-radical)))
1081       (setq radical value)
1082       (insert (format "(ideographic-radical . %S)\t; %c%s"
1083                       radical
1084                       (ideographic-radical radical)
1085                       line-breaking))
1086       (setq attributes (delq 'ideographic-radical attributes))
1087       )
1088     (when (and (memq 'shuowen-radical attributes)
1089                (setq value (get-char-attribute char 'shuowen-radical)))
1090       (insert (format "(shuowen-radical\t. %S)\t; %c%s"
1091                       value
1092                       (shuowen-radical value)
1093                       line-breaking))
1094       (setq attributes (delq 'shuowen-radical attributes))
1095       )
1096     (let (key)
1097       (dolist (domain
1098                (append
1099                 char-db-feature-domains
1100                 (let (dest domain)
1101                   (dolist (feature (char-attribute-list))
1102                     (setq feature (symbol-name feature))
1103                     (when (string-match
1104                            "\\(radical\\|strokes\\)@\\([^@*]+\\)\\(\\*\\|$\\)"
1105                            feature)
1106                       (setq domain (intern (match-string 2 feature)))
1107                      (unless (memq domain dest)
1108                        (setq dest (cons domain dest)))))
1109                   (sort dest #'string<))))
1110         (setq key (intern (format "%s@%s" 'ideographic-radical domain)))
1111         (when (and (memq key attributes)
1112                    (setq value (get-char-attribute char key)))
1113           (setq radical value)
1114           (insert (format "(%s . %S)\t; %c%s"
1115                           key
1116                           radical
1117                           (ideographic-radical radical)
1118                           line-breaking))
1119           (setq attributes (delq key attributes))
1120           )
1121         (setq key (intern (format "%s@%s" 'ideographic-strokes domain)))
1122         (when (and (memq key attributes)
1123                    (setq value (get-char-attribute char key)))
1124           (setq strokes value)
1125           (insert (format "(%s . %S)%s"
1126                           key
1127                           strokes
1128                           line-breaking))
1129           (setq attributes (delq key attributes))
1130           )
1131         (setq key (intern (format "%s@%s" 'total-strokes domain)))
1132         (when (and (memq key attributes)
1133                    (setq value (get-char-attribute char key)))
1134           (insert (format "(%s       . %S)%s"
1135                           key
1136                           value
1137                           line-breaking))
1138           (setq attributes (delq key attributes))
1139           )
1140         (dolist (feature '(ideographic-radical
1141                            ideographic-strokes
1142                            total-strokes))
1143           (setq key (intern (format "%s@%s*sources" feature domain)))
1144           (when (and (memq key attributes)
1145                      (setq value (get-char-attribute char key)))
1146             (insert (format "(%s%s" key line-breaking))
1147             (dolist (cell value)
1148               (insert (format " %s" cell)))
1149             (insert ")")
1150             (insert line-breaking)
1151             (setq attributes (delq key attributes))
1152             ))
1153         ))
1154     (when (and (memq 'ideographic-strokes attributes)
1155                (setq value (get-char-attribute char 'ideographic-strokes)))
1156       (setq strokes value)
1157       (insert (format "(ideographic-strokes . %S)%s"
1158                       strokes
1159                       line-breaking))
1160       (setq attributes (delq 'ideographic-strokes attributes))
1161       )
1162     (when (and (memq 'kangxi-radical attributes)
1163                (setq value (get-char-attribute char 'kangxi-radical)))
1164       (unless (eq value radical)
1165         (insert (format "(kangxi-radical\t . %S)\t; %c%s"
1166                         value
1167                         (ideographic-radical value)
1168                         line-breaking))
1169         (or radical
1170             (setq radical value)))
1171       (setq attributes (delq 'kangxi-radical attributes))
1172       )
1173     (when (and (memq 'kangxi-strokes attributes)
1174                (setq value (get-char-attribute char 'kangxi-strokes)))
1175       (unless (eq value strokes)
1176         (insert (format "(kangxi-strokes\t . %S)%s"
1177                         value
1178                         line-breaking))
1179         (or strokes
1180             (setq strokes value)))
1181       (setq attributes (delq 'kangxi-strokes attributes))
1182       )
1183     ;; (when (and (memq 'japanese-radical attributes)
1184     ;;            (setq value (get-char-attribute char 'japanese-radical)))
1185     ;;   (unless (eq value radical)
1186     ;;     (insert (format "(japanese-radical\t . %S)\t; %c%s"
1187     ;;                     value
1188     ;;                     (ideographic-radical value)
1189     ;;                     line-breaking))
1190     ;;     (or radical
1191     ;;         (setq radical value)))
1192     ;;   (setq attributes (delq 'japanese-radical attributes))
1193     ;;   )
1194     (when (and (memq 'japanese-strokes attributes)
1195                (setq value (get-char-attribute char 'japanese-strokes)))
1196       (unless (eq value strokes)
1197         (insert (format "(japanese-strokes\t . %S)%s"
1198                         value
1199                         line-breaking))
1200         (or strokes
1201             (setq strokes value)))
1202       (setq attributes (delq 'japanese-strokes attributes))
1203       )
1204     (when (and (memq 'cns-radical attributes)
1205                (setq value (get-char-attribute char 'cns-radical)))
1206       (insert (format "(cns-radical\t . %S)\t; %c%s"
1207                       value
1208                       (ideographic-radical value)
1209                       line-breaking))
1210       (setq attributes (delq 'cns-radical attributes))
1211       )
1212     (when (and (memq 'cns-strokes attributes)
1213                (setq value (get-char-attribute char 'cns-strokes)))
1214       (unless (eq value strokes)
1215         (insert (format "(cns-strokes\t . %S)%s"
1216                         value
1217                         line-breaking))
1218         (or strokes
1219             (setq strokes value)))
1220       (setq attributes (delq 'cns-strokes attributes))
1221       )
1222     ;; (when (and (memq 'shinjigen-1-radical attributes)
1223     ;;            (setq value (get-char-attribute char 'shinjigen-1-radical)))
1224     ;;   (unless (eq value radical)
1225     ;;     (insert (format "(shinjigen-1-radical . %S)\t; %c%s"
1226     ;;                     value
1227     ;;                     (ideographic-radical value)
1228     ;;                     line-breaking))
1229     ;;     (or radical
1230     ;;         (setq radical value)))
1231     ;;   (setq attributes (delq 'shinjigen-1-radical attributes))
1232     ;;   )
1233     ;; (when (and (memq 'ideographic- attributes)
1234     ;;            (setq value (get-char-attribute char 'ideographic-)))
1235     ;;   (insert "(ideographic-       ")
1236     ;;   (setq lbs (concat "\n" (make-string (current-column) ?\ ))
1237     ;;         separator nil)
1238     ;;   (while (consp value)
1239     ;;     (setq cell (car value))
1240     ;;     (if (integerp cell)
1241     ;;         (setq cell (decode-char '=ucs cell)))
1242     ;;     (cond ((characterp cell)
1243     ;;            (if separator
1244     ;;                (insert lbs))
1245     ;;            (if readable
1246     ;;                (insert (format "%S" cell))
1247     ;;              (char-db-insert-char-spec cell readable))
1248     ;;            (setq separator lbs))
1249     ;;           ((consp cell)
1250     ;;            (if separator
1251     ;;                (insert lbs))
1252     ;;            (if (consp (car cell))
1253     ;;                (char-db-insert-char-spec cell readable)
1254     ;;              (char-db-insert-char-reference cell readable))
1255     ;;            (setq separator lbs))
1256     ;;           (t
1257     ;;            (if separator
1258     ;;                (insert separator))
1259     ;;            (insert (prin1-to-string cell))
1260     ;;            (setq separator " ")))
1261     ;;     (setq value (cdr value)))
1262     ;;   (insert ")")
1263     ;;   (insert line-breaking)
1264     ;;   (setq attributes (delq 'ideographic- attributes)))
1265     (when (and (memq 'total-strokes attributes)
1266                (setq value (get-char-attribute char 'total-strokes)))
1267       (insert (format "(total-strokes       . %S)%s"
1268                       value
1269                       line-breaking))
1270       (setq attributes (delq 'total-strokes attributes))
1271       )
1272     ;; (when (and (memq '->ideograph attributes)
1273     ;;            (setq value (get-char-attribute char '->ideograph)))
1274     ;;   (insert (format "(->ideograph\t%s)%s"
1275     ;;                   (mapconcat (lambda (code)
1276     ;;                                (cond ((symbolp code)
1277     ;;                                       (symbol-name code))
1278     ;;                                      ((integerp code)
1279     ;;                                       (format "#x%04X" code))
1280     ;;                                      (t
1281     ;;                                       (format "%s %S"
1282     ;;                                               line-breaking code))))
1283     ;;                              value " ")
1284     ;;                   line-breaking))
1285     ;;   (setq attributes (delq '->ideograph attributes))
1286     ;;   )
1287     ;; (when (and (memq '->decomposition attributes)
1288     ;;            (setq value (get-char-attribute char '->decomposition)))
1289     ;;   (insert (format "(->decomposition\t%s)%s"
1290     ;;                   (mapconcat (lambda (code)
1291     ;;                                (cond ((symbolp code)
1292     ;;                                       (symbol-name code))
1293     ;;                                      ((characterp code)
1294     ;;                                       (if readable
1295     ;;                                           (format "%S" code)
1296     ;;                                         (format "#x%04X"
1297     ;;                                                 (char-int code))
1298     ;;                                         ))
1299     ;;                                      ((integerp code)
1300     ;;                                       (format "#x%04X" code))
1301     ;;                                      (t
1302     ;;                                       (format "%s%S" line-breaking code))))
1303     ;;                              value " ")
1304     ;;                   line-breaking))
1305     ;;   (setq attributes (delq '->decomposition attributes))
1306     ;;   )
1307     (if (equal (get-char-attribute char '->titlecase)
1308                (get-char-attribute char '->uppercase))
1309         (setq attributes (delq '->titlecase attributes)))
1310     (when (and (memq '->mojikyo attributes)
1311                (setq value (get-char-attribute char '->mojikyo)))
1312       (insert (format "(->mojikyo\t\t. %06d)\t; %c%s"
1313                       value (decode-char 'mojikyo value)
1314                       line-breaking))
1315       (setq attributes (delq '->mojikyo attributes))
1316       )
1317     ;; (when (and (memq 'hanyu-dazidian-vol attributes)
1318     ;;            (setq value (get-char-attribute char 'hanyu-dazidian-vol)))
1319     ;;   (insert (format "(hanyu-dazidian-vol  . %d)%s"
1320     ;;                   value line-breaking))
1321     ;;   (setq attributes (delq 'hanyu-dazidian-vol attributes))
1322     ;;   )
1323     ;; (when (and (memq 'hanyu-dazidian-page attributes)
1324     ;;            (setq value (get-char-attribute char 'hanyu-dazidian-page)))
1325     ;;   (insert (format "(hanyu-dazidian-page . %d)%s"
1326     ;;                   value line-breaking))
1327     ;;   (setq attributes (delq 'hanyu-dazidian-page attributes))
1328     ;;   )
1329     ;; (when (and (memq 'hanyu-dazidian-char attributes)
1330     ;;            (setq value (get-char-attribute char 'hanyu-dazidian-char)))
1331     ;;   (insert (format "(hanyu-dazidian-char . %d)%s"
1332     ;;                   value line-breaking))
1333     ;;   (setq attributes (delq 'hanyu-dazidian-char attributes))
1334     ;;   )
1335     (unless readable
1336       (dolist (ignored '(composition
1337                          ->denotational <-subsumptive ->ucs-unified
1338                          ->ideographic-component-forms))
1339         (setq attributes (delq ignored attributes))))
1340     (while attributes
1341       (setq name (car attributes))
1342       (unless (eq (setq value (get-char-attribute char name 'value-is-empty))
1343                   'value-is-empty)
1344         (cond ((setq ret (find-charset name))
1345                (setq name (charset-name ret))
1346                (when (not (memq name dest-ccss))
1347                  (setq dest-ccss (cons name dest-ccss))
1348                  (char-db-insert-ccs-feature name value line-breaking))
1349                )
1350               ((string-match "^=>ucs@" (symbol-name name))
1351                (insert (format "(%-18s . #x%04X)\t; %c%s"
1352                                name value (decode-char '=ucs value)
1353                                line-breaking))
1354                )
1355               ((eq name 'jisx0208-1978/4X)
1356                (insert (format "(%-18s . #x%04X)%s"
1357                                name value
1358                                line-breaking))
1359                )
1360               ((and
1361                 (not readable)
1362                 (not (eq name '->subsumptive))
1363                 (not (eq name '->uppercase))
1364                 (not (eq name '->lowercase))
1365                 (not (eq name '->titlecase))
1366                 (not (eq name '->canonical))
1367                 (not (eq name '->Bopomofo))
1368                 (not (eq name '->mistakable))
1369                 (not (eq name '->ideographic-variants))
1370                 (null (get-char-attribute
1371                        char (intern (format "%s*sources" name))))
1372                 (not (string-match "\\*sources$" (symbol-name name)))
1373                 (null (get-char-attribute
1374                        char (intern (format "%s*note" name))))
1375                 (not (string-match "\\*note$" (symbol-name name)))
1376                 (or (eq name '<-identical)
1377                     (eq name '<-uppercase)
1378                     (eq name '<-lowercase)
1379                     (eq name '<-titlecase)
1380                     (eq name '<-canonical)
1381                     (eq name '<-ideographic-variants)
1382                     ;; (eq name '<-synonyms)
1383                     (string-match "^<-synonyms" (symbol-name name))
1384                     (eq name '<-mistakable)
1385                     (when (string-match "^->" (symbol-name name))
1386                       (cond
1387                        ((string-match "^->fullwidth" (symbol-name name))
1388                         (not (and (consp value)
1389                                   (characterp (car value))
1390                                   (encode-char
1391                                    (car value) '=ucs 'defined-only)))
1392                         )
1393                        (t)))
1394                     ))
1395                )
1396               ((or (eq name 'ideographic-structure)
1397                    (eq name 'ideographic-combination)
1398                    ;; (eq name 'ideographic-)
1399                    (eq name '=decomposition)
1400                    (char-feature-base-name= '=decomposition name)
1401                    (char-feature-base-name= '=>decomposition name)
1402                    ;; (string-match "^=>*decomposition\\(@[^*]+\\)?$"
1403                    ;;               (symbol-name name))
1404                    (string-match "^\\(->\\|<-\\)[^*]*$" (symbol-name name))
1405                    (string-match "^\\(->\\|<-\\)[^*]*\\*sources$"
1406                                  (symbol-name name))
1407                    )
1408                (char-db-insert-relation-feature char name value
1409                                                 line-breaking
1410                                                 ccss readable))
1411               ;; ((memq name '(ideograph=
1412               ;;               original-ideograph-of
1413               ;;               ancient-ideograph-of
1414               ;;               vulgar-ideograph-of
1415               ;;               wrong-ideograph-of
1416               ;;               ;; simplified-ideograph-of
1417               ;;               ideographic-variants
1418               ;;               ;; ideographic-different-form-of
1419               ;;               ))
1420               ;;  (insert (format "(%-18s%s " name line-breaking))
1421               ;;  (setq lbs (concat "\n" (make-string (current-column) ?\ ))
1422               ;;        separator nil)
1423               ;;  (while (consp value)
1424               ;;    (setq cell (car value))
1425               ;;    (if (and (consp cell)
1426               ;;             (consp (car cell)))
1427               ;;        (progn
1428               ;;          (if separator
1429               ;;              (insert lbs))
1430               ;;          (char-db-insert-alist cell readable)
1431               ;;          (setq separator lbs))
1432               ;;      (if separator
1433               ;;          (insert separator))
1434               ;;      (insert (prin1-to-string cell))
1435               ;;      (setq separator " "))
1436               ;;    (setq value (cdr value)))
1437               ;;  (insert ")")
1438               ;;  (insert line-breaking))
1439               ((consp value)
1440                (insert (format "(%-18s " name))
1441                (setq lbs (concat "\n" (make-string (current-column) ?\ ))
1442                      separator nil)
1443                (while (consp value)
1444                  (setq cell (car value))
1445                  (if (and (consp cell)
1446                           (consp (car cell))
1447                           (setq ret (condition-case nil
1448                                         (find-char cell)
1449                                       (error nil))))
1450                      (progn
1451                        (setq rest cell
1452                              al nil
1453                              cal nil)
1454                        (while rest
1455                          (setq key (car (car rest)))
1456                          (if (find-charset key)
1457                              (setq cal (cons key cal))
1458                            (setq al (cons key al)))
1459                          (setq rest (cdr rest)))
1460                        (if separator
1461                            (insert lbs))
1462                        (insert-char-attributes ret
1463                                                readable
1464                                                al ; cal
1465                                                nil 'for-sub-node)
1466                        (setq separator lbs))
1467                    (setq ret (prin1-to-string cell))
1468                    (if separator
1469                        (if (< (+ (current-column)
1470                                  (length ret)
1471                                  (length separator))
1472                               76)
1473                            (insert separator)
1474                          (insert lbs)))
1475                    (insert ret)
1476                    (setq separator " "))
1477                  (setq value (cdr value)))
1478                (insert ")")
1479                (insert line-breaking))
1480               (t
1481                (insert (format "(%-18s" name))
1482                (setq ret (prin1-to-string value))
1483                (unless (< (+ (current-column)
1484                              (length ret)
1485                              3)
1486                           76)
1487                  (insert line-breaking))
1488                (insert " . " ret ")" line-breaking)
1489                ;; (insert (format "(%-18s . %S)%s"
1490                ;;                 name value
1491                ;;                 line-breaking))
1492                )
1493               ))
1494       (setq attributes (cdr attributes)))
1495     (insert ")")))
1496
1497 (defun insert-char-data (char &optional readable
1498                               attributes)
1499   (save-restriction
1500     (narrow-to-region (point)(point))
1501     (insert "(define-char
1502   '")
1503     (insert-char-attributes char readable attributes)
1504     (insert ")\n")
1505     (goto-char (point-min))
1506     (while (re-search-forward "[ \t]+$" nil t)
1507       (replace-match ""))
1508     ;; from tabify.
1509     (goto-char (point-min))
1510     (while (re-search-forward "[ \t][ \t][ \t]*" nil t)
1511       (let ((column (current-column))
1512             (indent-tabs-mode t))
1513         (delete-region (match-beginning 0) (point))
1514         (indent-to column)))
1515     (goto-char (point-max))
1516     ;; (tabify (point-min)(point-max))
1517     ))
1518
1519 (defun insert-char-data-with-variant (char &optional printable
1520                                            no-ucs-unified
1521                                            script excluded-script)
1522   (insert-char-data char printable)
1523   (let ((variants (char-variants char))
1524         rest
1525         variant vs ret)
1526     (setq variants (sort variants #'<))
1527     (setq rest variants)
1528     (setq variants (cons char variants))
1529     (while rest
1530       (setq variant (car rest))
1531       (unless (get-char-attribute variant '<-subsumptive)
1532         (if (and (or (null script)
1533                      (null (setq vs (get-char-attribute variant 'script)))
1534                      (memq script vs))
1535                  (or (null excluded-script)
1536                      (null (setq vs (get-char-attribute variant 'script)))
1537                      (not (memq excluded-script vs))))
1538             (unless (and no-ucs-unified (get-char-attribute variant '=ucs))
1539               (insert-char-data variant printable)
1540               (if (setq ret (char-variants variant))
1541                   (while ret
1542                     (or (memq (car ret) variants)
1543                         ;; (get-char-attribute (car ret) '<-subsumptive)
1544                         (setq rest (nconc rest (list (car ret)))))
1545                     (setq ret (cdr ret)))))))
1546       (setq rest (cdr rest)))))
1547
1548 (defun insert-char-range-data (min max &optional script excluded-script)
1549   (let ((code min)
1550         char)
1551     (while (<= code max)
1552       (setq char (decode-char '=ucs code))
1553       (if (encode-char char '=ucs 'defined-only)
1554           (insert-char-data-with-variant char nil 'no-ucs-unified
1555                                          script excluded-script))
1556       (setq code (1+ code)))))
1557
1558 (defun write-char-range-data-to-file (min max file
1559                                           &optional script excluded-script)
1560   (let ((coding-system-for-write char-db-file-coding-system))
1561     (with-temp-buffer
1562       (insert (format ";; -*- coding: %s -*-\n"
1563                       char-db-file-coding-system))
1564       (insert-char-range-data min max script excluded-script)
1565       (write-region (point-min)(point-max) file))))
1566
1567 (defvar what-character-original-window-configuration)
1568
1569 ;;;###autoload
1570 (defun what-char-definition (char)
1571   (interactive (list (char-after)))
1572   (let ((buf (get-buffer-create "*Character Description*"))
1573         (the-buf (current-buffer))
1574         (win-conf (current-window-configuration)))
1575     (pop-to-buffer buf)
1576     (make-local-variable 'what-character-original-window-configuration)
1577     (setq what-character-original-window-configuration win-conf)
1578     (setq buffer-read-only nil)
1579     (erase-buffer)
1580     (condition-case err
1581         (progn
1582           (insert-char-data-with-variant char 'printable)
1583           (unless (char-attribute-alist char)
1584             (insert (format ";; = %c\n"
1585                             (let* ((rest (split-char char))
1586                                    (ccs (pop rest))
1587                                    (code (pop rest)))
1588                               (while rest
1589                                 (setq code (logior (lsh code 8)
1590                                                    (pop rest))))
1591                               (decode-char ccs code)))))
1592           ;; (char-db-update-comment)
1593           (set-buffer-modified-p nil)
1594           (view-mode the-buf (lambda (buf)
1595                                (set-window-configuration
1596                                 what-character-original-window-configuration)
1597                                ))
1598           (goto-char (point-min)))
1599       (error (progn
1600                (set-window-configuration
1601                 what-character-original-window-configuration)
1602                (signal (car err) (cdr err)))))))
1603
1604
1605 ;;; @ end
1606 ;;;
1607
1608 (provide 'char-db-util)
1609
1610 ;;; char-db-util.el ends here