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,
   4 ;;   2007, 2008, 2009, 2010, 2011, 2012 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    ?旨 ?曰 ?乃 ?丂 ?可 ?兮 ?号 ?亏 ?喜 ?壴 ; 160
  83    ?鼓 ?豈 ?豆 ?豊 ?豐 ?䖒 ?虍 ?虎 ?虤 ?皿 ; 170
  84    ?𠙴 ?去 ?血 ?丶 ?丹 ?青 ?井 ?皀 ?鬯 ?食 ; 180
  85    ?亼 ?會 ?倉 ?入 ?缶 ?矢 ?高 ?冂 ?𩫏 ?京 ; 190
  86    ?亯 ?𣆪 ?畗 ?㐭 ?嗇 ?來 ?麥 ?夊 ?舛 ?䑞 ; 200
  87    ?韋 ?弟 ?夂 ?久 ?桀 ?木 ?東 ?林 ?才 ?叒 ; 210
  88    ?之 ?帀 ?出 ?𣎵 ?生 ?乇 ?𠂹 ?𠌶 ?華 ?𥝌 ; 220
  89    ?稽 ?巢 ?桼 ?束 ?㯻 ?囗 ?員 ?貝 ?邑 ?𨛜 ; 230
  90    ?日 ?旦 ?倝 ?㫃 ?冥 ?晶 ?月 ?有 ?明 ?囧 ; 240
  91    ?夕 ?多 ?毌 ?𢎘 ?𣐺 ?卣 ?齊 ?朿 ?片 ?鼎 ; 250
  92    ?克 ?彔 ?禾 ?秝 ?黍 ?香 ?米 ?毇 ?臼 ?凶 ; 260
  93    ?𣎳 ?林 ?麻 ?尗 ?耑 ?韭 ?瓜 ?瓠 ?宀 ?宮 ; 270
  94    ?呂 ?穴 ?㝱 ?𤕫 ?冖 ?𠔼 ?冃 ?㒳 ?网 ?襾 ; 280
  95    ?巾 ?巿 ?帛 ?白 ?㡀 ?黹 ?人 ?𠤎 ?匕 ?从 ; 290
  96    ])
  97
  98 (defun shuowen-radical (number)
  99   (aref shuowen-radicals (1- number)))
 100
 101 (defvar char-db-file-coding-system 'utf-8-mcs-er)
 102
 103 (defvar char-db-ignored-attributes '(ideographic-products))
 104
 105 (defvar char-db-coded-charset-priority-list
 106   '(ascii
 107     control-1
 108     latin-iso8859-1
 109     latin-iso8859-2
 110     latin-iso8859-3
 111     latin-iso8859-4
 112     latin-iso8859-9
 113     latin-jisx0201
 114     cyrillic-iso8859-5
 115     greek-iso8859-7
 116     thai-tis620
 117     =jis-x0208
 118     =jis-x0208@1978
 119     =jis-x0208@1983
 120     japanese-jisx0212
 121     chinese-gb2312
 122     =jis-x0208@1990
 123     chinese-cns11643-1
 124     chinese-cns11643-2
 125     chinese-cns11643-3
 126     chinese-cns11643-4
 127     chinese-cns11643-5
 128     chinese-cns11643-6
 129     chinese-cns11643-7
 130     =jis-x0213-1
 131     =jis-x0213-1@2000
 132     =jis-x0213-1@2004
 133     =jis-x0213-2
 134     korean-ksc5601
 135     chinese-isoir165
 136     katakana-jisx0201
 137     hebrew-iso8859-8
 138     chinese-gb12345
 139     latin-viscii
 140     ethiopic-ucs
 141     =big5-cdp
 142     =gt
 143     =adobe-japan1-0
 144     =adobe-japan1-1
 145     =adobe-japan1-2
 146     =adobe-japan1-3
 147     =adobe-japan1-4
 148     =adobe-japan1-5
 149     =adobe-japan1-6
 150     =hanyo-denshi/ja
 151     =hanyo-denshi/jb
 152     =hanyo-denshi/jc
 153     =hanyo-denshi/jd
 154     =hanyo-denshi/ft
 155     =hanyo-denshi/ia
 156     =hanyo-denshi/ib
 157     =hanyo-denshi/hg
 158     =hanyo-denshi/jt
 159     =hanyo-denshi/ks
 160     =daikanwa
 161     =daikanwa@rev2
 162     =daikanwa@rev1
 163     =cbeta
 164     =gt-k
 165     ideograph-hanziku-1
 166     ideograph-hanziku-2
 167     ideograph-hanziku-3
 168     ideograph-hanziku-4
 169     ideograph-hanziku-5
 170     ideograph-hanziku-6
 171     ideograph-hanziku-7
 172     ideograph-hanziku-8
 173     ideograph-hanziku-9
 174     ideograph-hanziku-10
 175     ideograph-hanziku-11
 176     ideograph-hanziku-12
 177     =>>>jis-x0208
 178     =>>>jis-x0213-1
 179     =>>>jis-x0213-2
 180     =>>>gt
 181     =>>>adobe-japan1
 182     =>>jis-x0208
 183     =>>jis-x0213-1
 184     =>>jis-x0213-1@2000
 185     =>>jis-x0213-1@2004
 186     =>>jis-x0213-2
 187     =>>jis-x0208@1978
 188     =>>hanyo-denshi/ft
 189     =>>hanyo-denshi/jt
 190     =>>hanyo-denshi/ks
 191     =>>gt
 192     =>>daikanwa
 193     =>>adobe-japan1
 194     =+>jis-x0208
 195     =+>jis-x0213-1
 196     =+>jis-x0213-2
 197     =+>jis-x0208@1978
 198     =+>adobe-japan1
 199     =>jis-x0208
 200     =>jis-x0208@1997
 201     =>jis-x0213-1
 202     =>jis-x0213-1@2000
 203     =>jis-x0213-1@2004
 204     =>jis-x0213-2
 205     ==>ucs@bucs
 206     =>ucs@iso
 207     =>ucs@unicode
 208     =>ucs@jis
 209     =>ucs@cns
 210     =>ucs@ks
 211     =+>ucs@iso
 212     =+>ucs@unicode
 213     =+>ucs@jis
 214     =+>ucs@jis/1990
 215     =+>ucs@cns
 216     =+>ucs@ks
 217     =>>ucs@iso
 218     =>>ucs@unicode
 219     =>>ucs@jis
 220     =>>ucs@cns
 221     =>>>ucs@iso
 222     =>>>ucs@unicode
 223     =ucs@iso
 224     =ucs@unicode
 225     =>>big5-cdp
 226     =>>gt-k
 227     =+>gt
 228     =>gt
 229     =>big5-cdp
 230     =>daikanwa
 231     =>daikanwa/ho
 232     =>cns11643-7
 233     =big5
 234     =big5-eten
 235     =>gt-k
 236     =zinbun-oracle
 237     =>zinbun-oracle
 238     =ruimoku-v6
 239     =>>ruimoku-v6
 240     =jef-china3
 241     =shinjigen))
 242
 243
 244 ;;; @ char-db formatters
 245 ;;;
 246
 247 (defun char-db-make-char-spec (char)
 248   (let (ret char-spec)
 249     (cond ((characterp char)
 250            (cond ((and (setq ret (encode-char char '=ucs 'defined-only))
 251                        (not (and (<= #xE000 ret)(<= ret #xF8FF))))
 252                   (setq char-spec (list (cons '=ucs ret)))
 253                   (cond ((setq ret (get-char-attribute char 'name))
 254                          (setq char-spec (cons (cons 'name ret) char-spec))
 255                          )
 256                         ((setq ret (get-char-attribute char 'name*))
 257                          (setq char-spec (cons (cons 'name* ret) char-spec))
 258                          ))
 259                   )
 260                  ((setq ret
 261                         (catch 'tag
 262                           (let ((rest char-db-coded-charset-priority-list)
 263                                 ccs)
 264                             (while rest
 265                               (setq ccs (charset-name
 266                                          (find-charset (car rest))))
 267                               (if (setq ret
 268                                         (encode-char char ccs
 269                                                      'defined-only))
 270                                   (throw 'tag (cons ccs ret)))
 271                               (setq rest (cdr rest))))))
 272                   (setq char-spec (list ret))
 273                   (dolist (ccs (delq (car ret) (charset-list)))
 274                     (if (and (or (charset-iso-final-char ccs)
 275                                  (memq ccs
 276                                        '(=daikanwa
 277                                          =daikanwa@rev2
 278                                          ;; =gt-k
 279                                          =jis-x0208@1997
 280                                          ))
 281                                  (string-match "=ucs@" (symbol-name ccs)))
 282                              (setq ccs (charset-name ccs))
 283                              (null (assq ccs char-spec))
 284                              (setq ret (encode-char char ccs 'defined-only)))
 285                         (setq char-spec (cons (cons ccs ret) char-spec))))
 286                   (if (null char-spec)
 287                       (setq char-spec (split-char char)))
 288                   (cond ((setq ret (get-char-attribute char 'name))
 289                          (setq char-spec (cons (cons 'name ret) char-spec))
 290                          )
 291                         ((setq ret (get-char-attribute char 'name*))
 292                          (setq char-spec (cons (cons 'name* ret) char-spec))
 293                          ))
 294                   )
 295                  ((setq ret (get-char-attribute
 296                              char 'ideographic-combination))
 297                   (setq char-spec
 298                         (cons (cons 'ideographic-combination ret)
 299                               char-spec))
 300                   ))
 301            char-spec)
 302           ((consp char)
 303            char))))
 304
 305 (defun char-db-insert-char-spec (char &optional readable column
 306                                       required-features)
 307   (unless column
 308     (setq column (current-column)))
 309   (let (char-spec temp-char)
 310     (setq char-spec (char-db-make-char-spec char))
 311     (unless (or (characterp char) ; char
 312                 (condition-case nil
 313                     (setq char (find-char char-spec))
 314                   (error nil)))
 315       ;; define temporary character
 316       ;;   Current implementation is dirty.
 317       (setq temp-char (define-char (cons '(ideograph-daikanwa . 0)
 318                                          char-spec)))
 319       (remove-char-attribute temp-char 'ideograph-daikanwa)
 320       (setq char temp-char))
 321     (insert-char-attributes char
 322                             readable
 323                             (union (mapcar #'car char-spec)
 324                                    required-features)
 325                             nil 'for-sub-node)
 326     (when temp-char
 327       ;; undefine temporary character
 328       ;;   Current implementation is dirty.
 329       (setq char-spec (char-attribute-alist temp-char))
 330       (while char-spec
 331         (remove-char-attribute temp-char (car (car char-spec)))
 332         (setq char-spec (cdr char-spec))))))
 333
 334 (defun char-db-insert-alist (alist &optional readable column)
 335   (unless column
 336     (setq column (current-column)))
 337   (let ((line-breaking
 338          (concat "\n" (make-string (1+ column) ?\ )))
 339         name value
 340         ret al ; cal
 341         key
 342         lbs cell rest separator)
 343     (insert "(")
 344     (while alist
 345       (setq name (car (car alist))
 346             value (cdr (car alist)))
 347       (cond ((eq name 'char)
 348              (insert "(char . ")
 349              (if (setq ret (condition-case nil
 350                                (find-char value)
 351                              (error nil)))
 352                  (progn
 353                    (setq al nil
 354                          ;; cal nil
 355                          )
 356                    (while value
 357                      (setq key (car (car value)))
 358                      ;; (if (find-charset key)
 359                      ;;     (setq cal (cons key cal))
 360                      (setq al (cons key al))
 361                      ;; )
 362                      (setq value (cdr value)))
 363                    (insert-char-attributes ret
 364                                            readable
 365                                            (or al 'none) ; cal
 366                                            nil 'for-sub-node))
 367                (insert (prin1-to-string value)))
 368              (insert ")")
 369              (insert line-breaking))
 370             ((consp value)
 371              (insert (format "(%-18s " name))
 372              (setq lbs (concat "\n" (make-string (current-column) ?\ )))
 373              (while (consp value)
 374                (setq cell (car value))
 375                (if (and (consp cell)
 376                         (consp (car cell))
 377                         (setq ret (condition-case nil
 378                                       (find-char cell)
 379                                     (error nil)))
 380                         )
 381                    (progn
 382                      (setq rest cell
 383                            al nil
 384                            ;; cal nil
 385                            )
 386                      (while rest
 387                        (setq key (car (car rest)))
 388                        ;; (if (find-charset key)
 389                        ;;     (setq cal (cons key cal))
 390                        (setq al (cons key al))
 391                        ;; )
 392                        (setq rest (cdr rest)))
 393                      (if separator
 394                          (insert lbs))
 395                      (insert-char-attributes ret
 396                                              readable
 397                                              al ; cal
 398                                              nil 'for-sub-node)
 399                      (setq separator lbs))
 400                  (if separator
 401                      (insert separator))
 402                  (insert (prin1-to-string cell))
 403                  (setq separator " "))
 404                (setq value (cdr value)))
 405              (insert ")")
 406              (insert line-breaking))
 407             (t
 408              (insert (format "(%-18s . %S)%s"
 409                              name value
 410                              line-breaking))))
 411       (setq alist (cdr alist))))
 412   (insert ")"))
 413
 414 (defun char-db-insert-char-reference (plist &optional readable column)
 415   (unless column
 416     (setq column (current-column)))
 417   (let ((line-breaking
 418          (concat "\n" (make-string (1+ column) ?\ )))
 419         (separator "")
 420         name value)
 421     (insert "(")
 422     (while plist
 423       (setq name (pop plist))
 424       (setq value (pop plist))
 425       (cond ((eq name :char)
 426              (insert separator)
 427              (insert ":char\t")
 428              (cond ((numberp value)
 429                     (setq value (decode-char '=ucs value)))
 430                    ;; ((consp value)
 431                    ;;  (setq value (or (find-char value)
 432                    ;;                  value)))
 433                    )
 434              (char-db-insert-char-spec value readable)
 435              (insert line-breaking)
 436              (setq separator ""))
 437             ((eq name :radical)
 438              (insert (format "%s%s\t%d ; %c%s"
 439                              separator
 440                              name value
 441                              (ideographic-radical value)
 442                              line-breaking))
 443              (setq separator ""))
 444             (t
 445              (insert (format "%s%s\t%S" separator name value))
 446              (setq separator line-breaking)))
 447       ))
 448   (insert ")"))
 449
 450 (defun char-db-decode-isolated-char (ccs code-point)
 451   (let (ret)
 452     (setq ret
 453           (cond ((eq ccs 'arabic-iso8859-6)
 454                  (decode-char ccs code-point))
 455                 ((and (memq ccs '(=gt-pj-1
 456                                   =gt-pj-2
 457                                   =gt-pj-3
 458                                   =gt-pj-4
 459                                   =gt-pj-5
 460                                   =gt-pj-6
 461                                   =gt-pj-7
 462                                   =gt-pj-8
 463                                   =gt-pj-9
 464                                   =gt-pj-10
 465                                   =gt-pj-11))
 466                       (setq ret (decode-char ccs code-point))
 467                       (setq ret (encode-char ret '=gt 'defined-only)))
 468                  (decode-builtin-char '=gt ret))
 469                 (t
 470                  (decode-builtin-char ccs code-point))))
 471     (cond ((and (<= 0 (char-int ret))
 472                 (<= (char-int ret) #x1F))
 473            (decode-char '=ucs (+ #x2400 (char-int ret))))
 474           ((= (char-int ret) #x7F)
 475            ?\u2421)
 476           (t ret))))
 477
 478 (defvar char-db-convert-obsolete-format t)
 479
 480 (defun char-db-insert-ccs-feature (name value line-breaking)
 481   (cond
 482    ((integerp value)
 483     (insert
 484      (format
 485       (cond
 486        ((memq name '(=shinjigen
 487                      =shinjigen@1ed
 488                      =shinjigen@rev =shinjigen/+p@rev
 489                      =daikanwa/ho =>>daikanwa/ho =>daikanwa/ho))
 490         "(%-18s .  %04d)\t; %c")
 491        ((eq name '=shinjigen@1ed/24pr)
 492         "(%-18s . %04d)\t; %c")
 493        ((or (memq name '(=daikanwa =>>daikanwa =>daikanwa
 494                          =daikanwa@rev1 =daikanwa@rev2
 495                          =daikanwa/+p =>>daikanwa/+p
 496                          =daikanwa/+2p
 497                          =gt =>>>gt =>>gt =+>gt =>gt
 498                          =gt-k =>>gt-k =>gt-k
 499                          =adobe-japan1-0 =>>>adobe-japan1-0 =>>adobe-japan1-0
 500                          =adobe-japan1-1 =>>>adobe-japan1-1 =>>adobe-japan1-1
 501                          =adobe-japan1-2 =>>>adobe-japan1-2 =>>adobe-japan1-2
 502                          =adobe-japan1-3 =>>>adobe-japan1-3 =>>adobe-japan1-3
 503                          =adobe-japan1-4 =>>>adobe-japan1-4 =>>adobe-japan1-4
 504                          =adobe-japan1-5 =>>>adobe-japan1-5 =>>adobe-japan1-5
 505                          =adobe-japan1-6 =>>>adobe-japan1-6 =>>adobe-japan1-6
 506                          =+>adobe-japan1
 507                          =cbeta =>>cbeta
 508                          =zinbun-oracle =>zinbun-oracle))
 509             ;; (string-match "^=adobe-" (symbol-name name))
 510             )
 511         "(%-18s . %05d)\t; %c")
 512        ((memq name '(=hanyo-denshi/ks =>>hanyo-denshi/ks mojikyo))
 513         "(%-18s . %06d)\t; %c")
 514        ((>= (charset-dimension name) 2)
 515         "(%-18s . #x%04X)\t; %c")
 516        (t
 517         "(%-18s . #x%02X)\t; %c"))
 518       name
 519       (if (= (charset-iso-graphic-plane name) 1)
 520           (logior value
 521                   (cond ((= (charset-dimension name) 1)
 522                          #x80)
 523                         ((= (charset-dimension name) 2)
 524                          #x8080)
 525                         ((= (charset-dimension name) 3)
 526                          #x808080)
 527                         (t 0)))
 528         value)
 529       (char-db-decode-isolated-char name value)))
 530     (if (and (= (charset-chars name) 94)
 531              (= (charset-dimension name) 2))
 532         (insert (format " [%02d-%02d]"
 533                         (- (lsh value -8) 32)
 534                         (- (logand value 255) 32))))
 535     )
 536    (t
 537     (insert (format "(%-18s . %s)" name value))
 538     ))
 539   (insert line-breaking))
 540
 541 (defun char-db-insert-relation-feature (char name value line-breaking
 542                                              ccss readable)
 543   (insert (format "(%-18s%s " name line-breaking))
 544   (let ((lbs (concat "\n" (make-string (current-column) ?\ )))
 545         separator cell sources required-features
 546         ret)
 547     (while (consp value)
 548       (setq cell (car value))
 549       (if (integerp cell)
 550           (setq cell (decode-char '=ucs cell)))
 551       (cond
 552        ((eq name '->subsumptive)
 553         (when (or (not (some (lambda (atr)
 554                                (get-char-attribute cell atr))
 555                              char-db-ignored-attributes))
 556                   (some (lambda (ccs)
 557                           (encode-char cell ccs 'defined-only))
 558                         ccss))
 559           (if separator
 560               (insert lbs))
 561           (let ((char-db-ignored-attributes
 562                  (cons '<-subsumptive
 563                        char-db-ignored-attributes)))
 564             (insert-char-attributes cell readable nil nil 'for-sub-node))
 565           (setq separator lbs))
 566         )
 567        ((characterp cell)
 568         (setq sources
 569               (get-char-attribute
 570                char (intern (format "%s*sources" name))))
 571         (setq required-features nil)
 572         (dolist (source sources)
 573           (cond
 574            ((memq source '(JP
 575                            JP/Jouyou
 576                            shinjigen shinjigen@1ed shinjigen@rev))
 577             (setq required-features
 578                   (union required-features
 579                          '(=jis-x0208
 580                            =jis-x0208@1990
 581                            =jis-x0213-1@2000
 582                            =jis-x0213-1@2004
 583                            =jis-x0213-2
 584                            =jis-x0212
 585                            =jis-x0208@1983
 586                            =jis-x0208@1978
 587                            =shinjigen))))
 588            ((eq source 'CN)
 589             (setq required-features
 590                   (union required-features
 591                          '(=gb2312
 592                            =gb12345
 593                            =iso-ir165)))))
 594           (cond
 595            ((find-charset (setq ret (intern (format "=%s" source))))
 596             (setq required-features
 597                   (cons ret required-features)))
 598            (t (setq required-features
 599                     (cons source required-features)))))
 600         (cond ((string-match "@JP" (symbol-name name))
 601                (setq required-features
 602                      (union required-features
 603                             '(=jis-x0208
 604                               =jis-x0208@1990
 605                               =jis-x0213-1-2000
 606                               =jis-x0213-2-2000
 607                               =jis-x0212
 608                               =jis-x0208@1983
 609                               =jis-x0208@1978))))
 610               ((string-match "@CN" (symbol-name name))
 611                (setq required-features
 612                      (union required-features
 613                             '(=gb2312
 614                               =gb12345
 615                               =iso-ir165)))))
 616         (if separator
 617             (insert lbs))
 618         (if readable
 619             (insert (format "%S" cell))
 620           (char-db-insert-char-spec cell readable
 621                                     nil
 622                                     required-features))
 623         (setq separator lbs))
 624        ((consp cell)
 625         (if separator
 626             (insert lbs))
 627         (if (consp (car cell))
 628             (char-db-insert-char-spec cell readable)
 629           (char-db-insert-char-reference cell readable))
 630         (setq separator lbs))
 631        (t
 632         (if separator
 633             (insert separator))
 634         (insert (prin1-to-string cell))
 635         (setq separator " ")))
 636       (setq value (cdr value)))
 637     (insert ")")
 638     (insert line-breaking)))
 639
 640 (defun insert-char-attributes (char &optional readable attributes column
 641                                     for-sub-node)
 642   (unless column
 643     (setq column (current-column)))
 644   (let (name value ; has-long-ccs-name
 645         rest
 646         radical strokes
 647         (line-breaking
 648          (concat "\n" (make-string (1+ column) ?\ )))
 649         lbs cell separator ret
 650         key al cal
 651         dest-ccss ; sources required-features
 652         ccss)
 653     (let (atr-d)
 654       (setq attributes
 655             (sort (if attributes
 656                       (if (consp attributes)
 657                           (progn
 658                             (dolist (name attributes)
 659                               (unless (memq name char-db-ignored-attributes)
 660                                 (if (find-charset name)
 661                                     (push name ccss))
 662                                 (push name atr-d)))
 663                             atr-d))
 664                     (dolist (name (char-attribute-list))
 665                       (unless (memq name char-db-ignored-attributes)
 666                         (if (find-charset name)
 667                             (push name ccss))
 668                         (push name atr-d)))
 669                     atr-d)
 670                   #'char-attribute-name<)))
 671     (insert "(")
 672     (when (memq '<-subsumptive attributes)
 673       (when (or readable (not for-sub-node))
 674         (when (setq value (get-char-attribute char '<-subsumptive))
 675           (char-db-insert-relation-feature char '<-subsumptive value
 676                                            line-breaking
 677                                            ccss readable)))
 678       (setq attributes (delq '<-subsumptive attributes)))
 679     (when (and (memq '<-denotational attributes)
 680                (setq value (get-char-attribute char '<-denotational)))
 681       (char-db-insert-relation-feature char '<-denotational value
 682                                        line-breaking
 683                                        ccss readable)
 684       (setq attributes (delq '<-denotational attributes)))
 685     (when (and (memq 'name attributes)
 686                (setq value (get-char-attribute char 'name)))
 687       (insert (format
 688                (if (> (+ (current-column) (length value)) 48)
 689                    "(name . %S)%s"
 690                  "(name               . %S)%s")
 691                value line-breaking))
 692       (setq attributes (delq 'name attributes))
 693       )
 694     (when (and (memq 'name* attributes)
 695                (setq value (get-char-attribute char 'name*)))
 696       (insert (format
 697                (if (> (+ (current-column) (length value)) 48)
 698                    "(name* . %S)%s"
 699                  "(name*              . %S)%s")
 700                value line-breaking))
 701       (setq attributes (delq 'name* attributes))
 702       )
 703     (when (and (memq 'script attributes)
 704                (setq value (get-char-attribute char 'script)))
 705       (insert (format "(script\t\t%s)%s"
 706                       (mapconcat (function prin1-to-string)
 707                                  value " ")
 708                       line-breaking))
 709       (setq attributes (delq 'script attributes))
 710       )
 711     (dolist (name '(=>ucs =>ucs*))
 712       (when (and (memq name attributes)
 713                  (setq value (get-char-attribute char name)))
 714         (insert (format "(%-18s . #x%04X)\t; %c%s"
 715                         name value (decode-char '=ucs value)
 716                         line-breaking))
 717         (setq attributes (delq name attributes))))
 718     (dolist (name '(=>ucs@gb =>ucs@big5))
 719       (when (and (memq name attributes)
 720                  (setq value (get-char-attribute char name)))
 721         (insert (format "(%-18s . #x%04X)\t; %c%s"
 722                         name value
 723                         (decode-char (intern
 724                                       (concat "="
 725                                               (substring
 726                                                (symbol-name name) 2)))
 727                                      value)
 728                         line-breaking))
 729         (setq attributes (delq name attributes))
 730         ))
 731     ;; (dolist (name '(=>daikanwa))
 732     ;;   (when (and (memq name attributes)
 733     ;;              (setq value (get-char-attribute char name)))
 734     ;;     (insert
 735     ;;      (if (integerp value)
 736     ;;          (format "(%-18s . %05d)\t; %c%s"
 737     ;;                  name value (decode-char '=daikanwa value)
 738     ;;                  line-breaking)
 739     ;;        (format "(%-18s %s)\t; %c%s"
 740     ;;                name
 741     ;;                (mapconcat (function prin1-to-string)
 742     ;;                           value " ")
 743     ;;                (char-representative-of-daikanwa char)
 744     ;;                line-breaking)))
 745     ;;     (setq attributes (delq name attributes))))
 746     (when (and (memq 'general-category attributes)
 747                (setq value (get-char-attribute char 'general-category)))
 748       (insert (format
 749                "(general-category\t%s) ; %s%s"
 750                (mapconcat (lambda (cell)
 751                             (format "%S" cell))
 752                           value " ")
 753                (cond ((rassoc value unidata-normative-category-alist)
 754                       "Normative Category")
 755                      ((rassoc value unidata-informative-category-alist)
 756                       "Informative Category")
 757                      (t
 758                       "Unknown Category"))
 759                line-breaking))
 760       (setq attributes (delq 'general-category attributes))
 761       )
 762     (when (and (memq 'bidi-category attributes)
 763                (setq value (get-char-attribute char 'bidi-category)))
 764       (insert (format "(bidi-category\t. %S)%s"
 765                       value
 766                       line-breaking))
 767       (setq attributes (delq 'bidi-category attributes))
 768       )
 769     (unless (or (not (memq 'mirrored attributes))
 770                 (eq (setq value (get-char-attribute char 'mirrored 'empty))
 771                     'empty))
 772       (insert (format "(mirrored\t\t. %S)%s"
 773                       value
 774                       line-breaking))
 775       (setq attributes (delq 'mirrored attributes))
 776       )
 777     (cond
 778      ((and (memq 'decimal-digit-value attributes)
 779            (setq value (get-char-attribute char 'decimal-digit-value)))
 780       (insert (format "(decimal-digit-value . %S)%s"
 781                       value
 782                       line-breaking))
 783       (setq attributes (delq 'decimal-digit-value attributes))
 784       (when (and (memq 'digit-value attributes)
 785                  (setq value (get-char-attribute char 'digit-value)))
 786         (insert (format "(digit-value\t . %S)%s"
 787                         value
 788                         line-breaking))
 789         (setq attributes (delq 'digit-value attributes))
 790         )
 791       (when (and (memq 'numeric-value attributes)
 792                  (setq value (get-char-attribute char 'numeric-value)))
 793         (insert (format "(numeric-value\t . %S)%s"
 794                         value
 795                         line-breaking))
 796         (setq attributes (delq 'numeric-value attributes))
 797         )
 798       )
 799      (t
 800       (when (and (memq 'digit-value attributes)
 801                  (setq value (get-char-attribute char 'digit-value)))
 802         (insert (format "(digit-value\t. %S)%s"
 803                         value
 804                         line-breaking))
 805         (setq attributes (delq 'digit-value attributes))
 806         )
 807       (when (and (memq 'numeric-value attributes)
 808                  (setq value (get-char-attribute char 'numeric-value)))
 809         (insert (format "(numeric-value\t. %S)%s"
 810                         value
 811                         line-breaking))
 812         (setq attributes (delq 'numeric-value attributes))
 813         )))
 814     (when (and (memq 'iso-10646-comment attributes)
 815                (setq value (get-char-attribute char 'iso-10646-comment)))
 816       (insert (format "(iso-10646-comment\t. %S)%s"
 817                       value
 818                       line-breaking))
 819       (setq attributes (delq 'iso-10646-comment attributes))
 820       )
 821     (when (and (memq 'morohashi-daikanwa attributes)
 822                (setq value (get-char-attribute char 'morohashi-daikanwa)))
 823       (insert (format "(morohashi-daikanwa\t%s)%s"
 824                       (mapconcat (function prin1-to-string) value " ")
 825                       line-breaking))
 826       (setq attributes (delq 'morohashi-daikanwa attributes))
 827       )
 828     (setq radical nil
 829           strokes nil)
 830     (when (and (memq 'ideographic-radical attributes)
 831                (setq value (get-char-attribute char 'ideographic-radical)))
 832       (setq radical value)
 833       (insert (format "(ideographic-radical . %S)\t; %c%s"
 834                       radical
 835                       (ideographic-radical radical)
 836                       line-breaking))
 837       (setq attributes (delq 'ideographic-radical attributes))
 838       )
 839     (when (and (memq 'shuowen-radical attributes)
 840                (setq value (get-char-attribute char 'shuowen-radical)))
 841       (insert (format "(shuowen-radical\t. %S)\t; %c%s"
 842                       value
 843                       (shuowen-radical value)
 844                       line-breaking))
 845       (setq attributes (delq 'shuowen-radical attributes))
 846       )
 847     (let (key)
 848       (dolist (domain
 849                (append
 850                 char-db-feature-domains
 851                 (let (dest domain)
 852                   (dolist (feature (char-attribute-list))
 853                     (setq feature (symbol-name feature))
 854                     (when (string-match
 855                            "\\(radical\\|strokes\\)@\\([^@*]+\\)\\(\\*\\|$\\)"
 856                            feature)
 857                       (setq domain (intern (match-string 2 feature)))
 858                      (unless (memq domain dest)
 859                        (setq dest (cons domain dest)))))
 860                   (sort dest #'string<))))
 861         (setq key (intern (format "%s@%s" 'ideographic-radical domain)))
 862         (when (and (memq key attributes)
 863                    (setq value (get-char-attribute char key)))
 864           (setq radical value)
 865           (insert (format "(%s . %S)\t; %c%s"
 866                           key
 867                           radical
 868                           (ideographic-radical radical)
 869                           line-breaking))
 870           (setq attributes (delq key attributes))
 871           )
 872         (setq key (intern (format "%s@%s" 'ideographic-strokes domain)))
 873         (when (and (memq key attributes)
 874                    (setq value (get-char-attribute char key)))
 875           (setq strokes value)
 876           (insert (format "(%s . %S)%s"
 877                           key
 878                           strokes
 879                           line-breaking))
 880           (setq attributes (delq key attributes))
 881           )
 882         (setq key (intern (format "%s@%s" 'total-strokes domain)))
 883         (when (and (memq key attributes)
 884                    (setq value (get-char-attribute char key)))
 885           (insert (format "(%s       . %S)%s"
 886                           key
 887                           value
 888                           line-breaking))
 889           (setq attributes (delq key attributes))
 890           )
 891         (dolist (feature '(ideographic-radical
 892                            ideographic-strokes
 893                            total-strokes))
 894           (setq key (intern (format "%s@%s*sources" feature domain)))
 895           (when (and (memq key attributes)
 896                      (setq value (get-char-attribute char key)))
 897             (insert (format "(%s%s" key line-breaking))
 898             (dolist (cell value)
 899               (insert (format " %s" cell)))
 900             (insert ")")
 901             (insert line-breaking)
 902             (setq attributes (delq key attributes))
 903             ))
 904         ))
 905     (when (and (memq 'ideographic-strokes attributes)
 906                (setq value (get-char-attribute char 'ideographic-strokes)))
 907       (setq strokes value)
 908       (insert (format "(ideographic-strokes . %S)%s"
 909                       strokes
 910                       line-breaking))
 911       (setq attributes (delq 'ideographic-strokes attributes))
 912       )
 913     (when (and (memq 'kangxi-radical attributes)
 914                (setq value (get-char-attribute char 'kangxi-radical)))
 915       (unless (eq value radical)
 916         (insert (format "(kangxi-radical\t . %S)\t; %c%s"
 917                         value
 918                         (ideographic-radical value)
 919                         line-breaking))
 920         (or radical
 921             (setq radical value)))
 922       (setq attributes (delq 'kangxi-radical attributes))
 923       )
 924     (when (and (memq 'kangxi-strokes attributes)
 925                (setq value (get-char-attribute char 'kangxi-strokes)))
 926       (unless (eq value strokes)
 927         (insert (format "(kangxi-strokes\t . %S)%s"
 928                         value
 929                         line-breaking))
 930         (or strokes
 931             (setq strokes value)))
 932       (setq attributes (delq 'kangxi-strokes attributes))
 933       )
 934     (when (and (memq 'japanese-radical attributes)
 935                (setq value (get-char-attribute char 'japanese-radical)))
 936       (unless (eq value radical)
 937         (insert (format "(japanese-radical\t . %S)\t; %c%s"
 938                         value
 939                         (ideographic-radical value)
 940                         line-breaking))
 941         (or radical
 942             (setq radical value)))
 943       (setq attributes (delq 'japanese-radical attributes))
 944       )
 945     (when (and (memq 'japanese-strokes attributes)
 946                (setq value (get-char-attribute char 'japanese-strokes)))
 947       (unless (eq value strokes)
 948         (insert (format "(japanese-strokes\t . %S)%s"
 949                         value
 950                         line-breaking))
 951         (or strokes
 952             (setq strokes value)))
 953       (setq attributes (delq 'japanese-strokes attributes))
 954       )
 955     (when (and (memq 'cns-radical attributes)
 956                (setq value (get-char-attribute char 'cns-radical)))
 957       (insert (format "(cns-radical\t . %S)\t; %c%s"
 958                       value
 959                       (ideographic-radical value)
 960                       line-breaking))
 961       (setq attributes (delq 'cns-radical attributes))
 962       )
 963     (when (and (memq 'cns-strokes attributes)
 964                (setq value (get-char-attribute char 'cns-strokes)))
 965       (unless (eq value strokes)
 966         (insert (format "(cns-strokes\t . %S)%s"
 967                         value
 968                         line-breaking))
 969         (or strokes
 970             (setq strokes value)))
 971       (setq attributes (delq 'cns-strokes attributes))
 972       )
 973     (when (and (memq 'shinjigen-1-radical attributes)
 974                (setq value (get-char-attribute char 'shinjigen-1-radical)))
 975       (unless (eq value radical)
 976         (insert (format "(shinjigen-1-radical . %S)\t; %c%s"
 977                         value
 978                         (ideographic-radical value)
 979                         line-breaking))
 980         (or radical
 981             (setq radical value)))
 982       (setq attributes (delq 'shinjigen-1-radical attributes))
 983       )
 984     (when (and (memq 'ideographic- attributes)
 985                (setq value (get-char-attribute char 'ideographic-)))
 986       (insert "(ideographic-       ")
 987       (setq lbs (concat "\n" (make-string (current-column) ?\ ))
 988             separator nil)
 989       (while (consp value)
 990         (setq cell (car value))
 991         (if (integerp cell)
 992             (setq cell (decode-char '=ucs cell)))
 993         (cond ((characterp cell)
 994                (if separator
 995                    (insert lbs))
 996                (if readable
 997                    (insert (format "%S" cell))
 998                  (char-db-insert-char-spec cell readable))
 999                (setq separator lbs))
1000               ((consp cell)
1001                (if separator
1002                    (insert lbs))
1003                (if (consp (car cell))
1004                    (char-db-insert-char-spec cell readable)
1005                  (char-db-insert-char-reference cell readable))
1006                (setq separator lbs))
1007               (t
1008                (if separator
1009                    (insert separator))
1010                (insert (prin1-to-string cell))
1011                (setq separator " ")))
1012         (setq value (cdr value)))
1013       (insert ")")
1014       (insert line-breaking)
1015       (setq attributes (delq 'ideographic- attributes)))
1016     (when (and (memq 'total-strokes attributes)
1017                (setq value (get-char-attribute char 'total-strokes)))
1018       (insert (format "(total-strokes       . %S)%s"
1019                       value
1020                       line-breaking))
1021       (setq attributes (delq 'total-strokes attributes))
1022       )
1023     (when (and (memq '->ideograph attributes)
1024                (setq value (get-char-attribute char '->ideograph)))
1025       (insert (format "(->ideograph\t%s)%s"
1026                       (mapconcat (lambda (code)
1027                                    (cond ((symbolp code)
1028                                           (symbol-name code))
1029                                          ((integerp code)
1030                                           (format "#x%04X" code))
1031                                          (t
1032                                           (format "%s %S"
1033                                                   line-breaking code))))
1034                                  value " ")
1035                       line-breaking))
1036       (setq attributes (delq '->ideograph attributes))
1037       )
1038     ;; (when (and (memq '->decomposition attributes)
1039     ;;            (setq value (get-char-attribute char '->decomposition)))
1040     ;;   (insert (format "(->decomposition\t%s)%s"
1041     ;;                   (mapconcat (lambda (code)
1042     ;;                                (cond ((symbolp code)
1043     ;;                                       (symbol-name code))
1044     ;;                                      ((characterp code)
1045     ;;                                       (if readable
1046     ;;                                           (format "%S" code)
1047     ;;                                         (format "#x%04X"
1048     ;;                                                 (char-int code))
1049     ;;                                         ))
1050     ;;                                      ((integerp code)
1051     ;;                                       (format "#x%04X" code))
1052     ;;                                      (t
1053     ;;                                       (format "%s%S" line-breaking code))))
1054     ;;                              value " ")
1055     ;;                   line-breaking))
1056     ;;   (setq attributes (delq '->decomposition attributes))
1057     ;;   )
1058     (if (equal (get-char-attribute char '->titlecase)
1059                (get-char-attribute char '->uppercase))
1060         (setq attributes (delq '->titlecase attributes)))
1061     (when (and (memq '->mojikyo attributes)
1062                (setq value (get-char-attribute char '->mojikyo)))
1063       (insert (format "(->mojikyo\t\t. %06d)\t; %c%s"
1064                       value (decode-char 'mojikyo value)
1065                       line-breaking))
1066       (setq attributes (delq '->mojikyo attributes))
1067       )
1068     (when (and (memq 'hanyu-dazidian-vol attributes)
1069                (setq value (get-char-attribute char 'hanyu-dazidian-vol)))
1070       (insert (format "(hanyu-dazidian-vol  . %d)%s"
1071                       value line-breaking))
1072       (setq attributes (delq 'hanyu-dazidian-vol attributes))
1073       )
1074     (when (and (memq 'hanyu-dazidian-page attributes)
1075                (setq value (get-char-attribute char 'hanyu-dazidian-page)))
1076       (insert (format "(hanyu-dazidian-page . %d)%s"
1077                       value line-breaking))
1078       (setq attributes (delq 'hanyu-dazidian-page attributes))
1079       )
1080     (when (and (memq 'hanyu-dazidian-char attributes)
1081                (setq value (get-char-attribute char 'hanyu-dazidian-char)))
1082       (insert (format "(hanyu-dazidian-char . %d)%s"
1083                       value line-breaking))
1084       (setq attributes (delq 'hanyu-dazidian-char attributes))
1085       )
1086     (unless readable
1087       (dolist (ignored '(composition
1088                          ->denotational <-subsumptive ->ucs-unified
1089                          ->ideographic-component-forms))
1090         (setq attributes (delq ignored attributes))))
1091     (while attributes
1092       (setq name (car attributes))
1093       (unless (eq (setq value (get-char-attribute char name 'value-is-empty))
1094                   'value-is-empty)
1095         (cond ((setq ret (find-charset name))
1096                (setq name (charset-name ret))
1097                (when (not (memq name dest-ccss))
1098                  (setq dest-ccss (cons name dest-ccss))
1099                  (char-db-insert-ccs-feature name value line-breaking))
1100                )
1101               ((string-match "^=>ucs@" (symbol-name name))
1102                (insert (format "(%-18s . #x%04X)\t; %c%s"
1103                                name value (decode-char '=ucs value)
1104                                line-breaking))
1105                )
1106               ((eq name 'jisx0208-1978/4X)
1107                (insert (format "(%-18s . #x%04X)%s"
1108                                name value
1109                                line-breaking))
1110                )
1111               ((and
1112                 (not readable)
1113                 (not (eq name '->subsumptive))
1114                 (not (eq name '->uppercase))
1115                 (not (eq name '->lowercase))
1116                 (not (eq name '->titlecase))
1117                 (not (eq name '->canonical))
1118                 (not (eq name '->Bopomofo))
1119                 (not (eq name '->mistakable))
1120                 (not (eq name '->ideographic-variants))
1121                 (null (get-char-attribute
1122                        char (intern (format "%s*sources" name))))
1123                 (not (string-match "\\*sources$" (symbol-name name)))
1124                 (null (get-char-attribute
1125                        char (intern (format "%s*note" name))))
1126                 (not (string-match "\\*note$" (symbol-name name)))
1127                 (or (eq name '<-identical)
1128                     (eq name '<-uppercase)
1129                     (eq name '<-lowercase)
1130                     (eq name '<-titlecase)
1131                     (eq name '<-canonical)
1132                     (eq name '<-ideographic-variants)
1133                     ;; (eq name '<-synonyms)
1134                     (string-match "^<-synonyms" (symbol-name name))
1135                     (eq name '<-mistakable)
1136                     (when (string-match "^->" (symbol-name name))
1137                       (cond
1138                        ((string-match "^->fullwidth" (symbol-name name))
1139                         (not (and (consp value)
1140                                   (characterp (car value))
1141                                   (encode-char
1142                                    (car value) '=ucs 'defined-only)))
1143                         )
1144                        (t)))
1145                     ))
1146                )
1147               ((or (eq name 'ideographic-structure)
1148                    (eq name 'ideographic-combination)
1149                    (eq name 'ideographic-)
1150                    (eq name '=decomposition)
1151                    (char-feature-base-name= '=decomposition name)
1152                    (char-feature-base-name= '=>decomposition name)
1153                    ;; (string-match "^=>*decomposition\\(@[^*]+\\)?$"
1154                    ;;               (symbol-name name))
1155                    (string-match "^\\(->\\|<-\\)[^*]*$" (symbol-name name))
1156                    (string-match "^\\(->\\|<-\\)[^*]*\\*sources$"
1157                                  (symbol-name name))
1158                    )
1159                (char-db-insert-relation-feature char name value
1160                                                 line-breaking
1161                                                 ccss readable))
1162               ((memq name '(ideograph=
1163                             original-ideograph-of
1164                             ancient-ideograph-of
1165                             vulgar-ideograph-of
1166                             wrong-ideograph-of
1167                             ;; simplified-ideograph-of
1168                             ideographic-variants
1169                             ;; ideographic-different-form-of
1170                             ))
1171                (insert (format "(%-18s%s " name line-breaking))
1172                (setq lbs (concat "\n" (make-string (current-column) ?\ ))
1173                      separator nil)
1174                (while (consp value)
1175                  (setq cell (car value))
1176                  (if (and (consp cell)
1177                           (consp (car cell)))
1178                      (progn
1179                        (if separator
1180                            (insert lbs))
1181                        (char-db-insert-alist cell readable)
1182                        (setq separator lbs))
1183                    (if separator
1184                        (insert separator))
1185                    (insert (prin1-to-string cell))
1186                    (setq separator " "))
1187                  (setq value (cdr value)))
1188                (insert ")")
1189                (insert line-breaking))
1190               ((consp value)
1191                (insert (format "(%-18s " name))
1192                (setq lbs (concat "\n" (make-string (current-column) ?\ ))
1193                      separator nil)
1194                (while (consp value)
1195                  (setq cell (car value))
1196                  (if (and (consp cell)
1197                           (consp (car cell))
1198                           (setq ret (condition-case nil
1199                                         (find-char cell)
1200                                       (error nil))))
1201                      (progn
1202                        (setq rest cell
1203                              al nil
1204                              cal nil)
1205                        (while rest
1206                          (setq key (car (car rest)))
1207                          (if (find-charset key)
1208                              (setq cal (cons key cal))
1209                            (setq al (cons key al)))
1210                          (setq rest (cdr rest)))
1211                        (if separator
1212                            (insert lbs))
1213                        (insert-char-attributes ret
1214                                                readable
1215                                                al ; cal
1216                                                nil 'for-sub-node)
1217                        (setq separator lbs))
1218                    (setq ret (prin1-to-string cell))
1219                    (if separator
1220                        (if (< (+ (current-column)
1221                                  (length ret)
1222                                  (length separator))
1223                               76)
1224                            (insert separator)
1225                          (insert lbs)))
1226                    (insert ret)
1227                    (setq separator " "))
1228                  (setq value (cdr value)))
1229                (insert ")")
1230                (insert line-breaking))
1231               (t
1232                (insert (format "(%-18s" name))
1233                (setq ret (prin1-to-string value))
1234                (unless (< (+ (current-column)
1235                              (length ret)
1236                              3)
1237                           76)
1238                  (insert line-breaking))
1239                (insert " . " ret ")" line-breaking)
1240                ;; (insert (format "(%-18s . %S)%s"
1241                ;;                 name value
1242                ;;                 line-breaking))
1243                )
1244               ))
1245       (setq attributes (cdr attributes)))
1246     (insert ")")))
1247
1248 (defun insert-char-data (char &optional readable
1249                               attributes)
1250   (save-restriction
1251     (narrow-to-region (point)(point))
1252     (insert "(define-char
1253   '")
1254     (insert-char-attributes char readable attributes)
1255     (insert ")\n")
1256     (goto-char (point-min))
1257     (while (re-search-forward "[ \t]+$" nil t)
1258       (replace-match ""))
1259     ;; from tabify.
1260     (goto-char (point-min))
1261     (while (re-search-forward "[ \t][ \t][ \t]*" nil t)
1262       (let ((column (current-column))
1263             (indent-tabs-mode t))
1264         (delete-region (match-beginning 0) (point))
1265         (indent-to column)))
1266     (goto-char (point-max))
1267     ;; (tabify (point-min)(point-max))
1268     ))
1269
1270 (defun insert-char-data-with-variant (char &optional printable
1271                                            no-ucs-unified
1272                                            script excluded-script)
1273   (insert-char-data char printable)
1274   (let ((variants (char-variants char))
1275         rest
1276         variant vs ret)
1277     (setq variants (sort variants #'<))
1278     (setq rest variants)
1279     (setq variants (cons char variants))
1280     (while rest
1281       (setq variant (car rest))
1282       (unless (get-char-attribute variant '<-subsumptive)
1283         (if (and (or (null script)
1284                      (null (setq vs (get-char-attribute variant 'script)))
1285                      (memq script vs))
1286                  (or (null excluded-script)
1287                      (null (setq vs (get-char-attribute variant 'script)))
1288                      (not (memq excluded-script vs))))
1289             (unless (and no-ucs-unified (get-char-attribute variant '=ucs))
1290               (insert-char-data variant printable)
1291               (if (setq ret (char-variants variant))
1292                   (while ret
1293                     (or (memq (car ret) variants)
1294                         ;; (get-char-attribute (car ret) '<-subsumptive)
1295                         (setq rest (nconc rest (list (car ret)))))
1296                     (setq ret (cdr ret)))))))
1297       (setq rest (cdr rest)))))
1298
1299 (defun insert-char-range-data (min max &optional script excluded-script)
1300   (let ((code min)
1301         char)
1302     (while (<= code max)
1303       (setq char (decode-char '=ucs code))
1304       (if (encode-char char '=ucs 'defined-only)
1305           (insert-char-data-with-variant char nil 'no-ucs-unified
1306                                          script excluded-script))
1307       (setq code (1+ code)))))
1308
1309 (defun write-char-range-data-to-file (min max file
1310                                           &optional script excluded-script)
1311   (let ((coding-system-for-write char-db-file-coding-system))
1312     (with-temp-buffer
1313       (insert (format ";; -*- coding: %s -*-\n"
1314                       char-db-file-coding-system))
1315       (insert-char-range-data min max script excluded-script)
1316       (write-region (point-min)(point-max) file))))
1317
1318 (defvar what-character-original-window-configuration)
1319
1320 ;;;###autoload
1321 (defun what-char-definition (char)
1322   (interactive (list (char-after)))
1323   (let ((buf (get-buffer-create "*Character Description*"))
1324         (the-buf (current-buffer))
1325         (win-conf (current-window-configuration)))
1326     (pop-to-buffer buf)
1327     (make-local-variable 'what-character-original-window-configuration)
1328     (setq what-character-original-window-configuration win-conf)
1329     (setq buffer-read-only nil)
1330     (erase-buffer)
1331     (condition-case err
1332         (progn
1333           (insert-char-data-with-variant char 'printable)
1334           (unless (char-attribute-alist char)
1335             (insert (format ";; = %c\n"
1336                             (let* ((rest (split-char char))
1337                                    (ccs (pop rest))
1338                                    (code (pop rest)))
1339                               (while rest
1340                                 (setq code (logior (lsh code 8)
1341                                                    (pop rest))))
1342                               (decode-char ccs code)))))
1343           ;; (char-db-update-comment)
1344           (set-buffer-modified-p nil)
1345           (view-mode the-buf (lambda (buf)
1346                                (set-window-configuration
1347                                 what-character-original-window-configuration)
1348                                ))
1349           (goto-char (point-min)))
1350       (error (progn
1351                (set-window-configuration
1352                 what-character-original-window-configuration)
1353                (signal (car err) (cdr err)))))))
1354
1355
1356 ;;; @ end
1357 ;;;
1358
1359 (provide 'char-db-util)
1360
1361 ;;; char-db-util.el ends here