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 MORIOKA Tomohiko.
   4
   5 ;; Author: MORIOKA Tomohiko <tomo@kanji.zinbun.kyoto-u.ac.jp>
   6 ;; Keywords: CHISE, Character Database, ISO/IEC 10646, UCS, Unicode, MULE.
   7
   8 ;; This file is part of XEmacs CHISE.
   9
  10 ;; XEmacs CHISE is free software; you can redistribute it and/or
  11 ;; modify it under the terms of the GNU General Public License as
  12 ;; published by the Free Software Foundation; either version 2, or (at
  13 ;; your option) any later version.
  14
  15 ;; XEmacs CHISE is distributed in the hope that it will be useful,
  16 ;; but WITHOUT ANY WARRANTY; without even the implied warranty of
  17 ;; MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
  18 ;; General Public License for more details.
  19
  20 ;; You should have received a copy of the GNU General Public License
  21 ;; along with XEmacs CHISE; see the file COPYING.  If not, write to
  22 ;; the Free Software Foundation, Inc., 59 Temple Place - Suite 330,
  23 ;; Boston, MA 02111-1307, USA.
  24
  25 ;;; Code:
  26
  27 (require 'alist)
  28
  29 (defconst unidata-normative-category-alist
  30   '(("Lu" letter        uppercase)
  31     ("Ll" letter        lowercase)
  32     ("Lt" letter        titlecase)
  33     ("Mn" mark          non-spacing)
  34     ("Mc" mark          spacing-combining)
  35     ("Me" mark          enclosing)
  36     ("Nd" number        decimal-digit)
  37     ("Nl" number        letter)
  38     ("No" number        other)
  39     ("Zs" separator     space)
  40     ("Zl" separator     line)
  41     ("Zp" separator     paragraph)
  42     ("Cc" other         control)
  43     ("Cf" other         format)
  44     ("Cs" other         surrogate)
  45     ("Co" other         private-use)
  46     ("Cn" other         not-assigned)))
  47
  48 (defconst unidata-informative-category-alist
  49   '(("Lm" letter        modifier)
  50     ("Lo" letter        other)
  51     ("Pc" punctuation   connector)
  52     ("Pd" punctuation   dash)
  53     ("Ps" punctuation   open)
  54     ("Pe" punctuation   close)
  55     ("Pi" punctuation   initial-quote)
  56     ("Pf" punctuation   final-quote)
  57     ("Po" punctuation   other)
  58     ("Sm" symbol        math)
  59     ("Sc" symbol        currency)
  60     ("Sk" symbol        modifier)
  61     ("So" symbol        other)
  62     ))
  63
  64 (defconst ideographic-radicals
  65   (let ((v (make-vector 215 nil))
  66         (i 1))
  67     (while (< i 215)
  68       (aset v i (decode-char '=ucs (+ #x2EFF i)))
  69       (setq i (1+ i)))
  70     v))
  71
  72 (defun ideographic-radical (number)
  73   (aref ideographic-radicals number))
  74
  75 (defconst shuowen-radicals
  76   [?一 ?上 ?示 ?三 ?王 ?玉 ?玨 ?气 ?士 ?丨 ?屮 ?艸 ?茻])
  77
  78 (defun shuowen-radical (number)
  79   (aref shuowen-radicals (1- number)))
  80
  81 (defvar char-db-file-coding-system 'utf-8-mcs-er)
  82
  83 (defvar char-db-feature-domains
  84   '(ucs daikanwa cns gt jis jis/alt jis/a jis/b
  85         jis-x0212 jis-x0213 cdp shinjigen misc unknown))
  86
  87 (defvar char-db-ignored-attributes '(ideographic-products))
  88
  89 (defun char-attribute-name< (ka kb)
  90   (cond
  91    ((eq '->denotational kb)
  92     t)
  93    ((eq '->subsumptive kb)
  94     (not (eq '->denotational ka)))
  95    ((eq '->denotational ka)
  96     nil)
  97    ((eq '->subsumptive ka)
  98     nil)
  99    ((find-charset ka)
 100     (if (find-charset kb)
 101         (if (<= (charset-id ka) 1)
 102             (if (<= (charset-id kb) 1)
 103                 (cond
 104                  ((= (charset-dimension ka)
 105                      (charset-dimension kb))
 106                   (> (charset-id ka)(charset-id kb)))
 107                  (t
 108                   (> (charset-dimension ka)
 109                      (charset-dimension kb))
 110                   ))
 111               t)
 112           (if (<= (charset-id kb) 1)
 113               nil
 114             (< (charset-id ka)(charset-id kb))))
 115       nil))
 116    ((find-charset kb)
 117     t)
 118    ((symbolp ka)
 119     (cond ((symbolp kb)
 120            (string< (symbol-name ka)
 121                     (symbol-name kb)))
 122           (t)))
 123    ((symbolp kb)
 124     nil)))
 125
 126 (defvar char-db-coded-charset-priority-list
 127   '(ascii
 128     control-1
 129     latin-iso8859-1
 130     latin-iso8859-2
 131     latin-iso8859-3
 132     latin-iso8859-4
 133     latin-iso8859-9
 134     latin-jisx0201
 135     cyrillic-iso8859-5
 136     greek-iso8859-7
 137     thai-tis620
 138     =jis-x0208
 139     =jis-x0208@1978
 140     =jis-x0208@1983
 141     japanese-jisx0212
 142     chinese-gb2312
 143     chinese-cns11643-1
 144     chinese-cns11643-2
 145     chinese-cns11643-3
 146     chinese-cns11643-4
 147     chinese-cns11643-5
 148     chinese-cns11643-6
 149     chinese-cns11643-7
 150     =jis-x0208@1990
 151     =jis-x0213-1-2000
 152     =jis-x0213-2-2000
 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     =gt
 162     ideograph-daikanwa-2
 163     ideograph-daikanwa
 164     =cbeta
 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     =gt-k
 178     =ucs@iso
 179     =ucs@unicode
 180     =big5
 181     =big5-eten
 182     =jis-x0208@1997
 183     =zinbun-oracle
 184     =jef-china3))
 185
 186 (defun char-db-make-char-spec (char)
 187   (let (ret char-spec)
 188     (cond ((characterp char)
 189            (cond ((and (setq ret (encode-char char '=ucs 'defined-only))
 190                        (not (and (<= #xE000 ret)(<= ret #xF8FF))))
 191                   (setq char-spec (list (cons '=ucs ret)))
 192                   (cond ((setq ret (get-char-attribute char 'name))
 193                          (setq char-spec (cons (cons 'name ret) char-spec))
 194                          )
 195                         ((setq ret (get-char-attribute char 'name*))
 196                          (setq char-spec (cons (cons 'name* ret) char-spec))
 197                          ))
 198                   )
 199                  ((setq ret
 200                         (catch 'tag
 201                           (let ((rest char-db-coded-charset-priority-list)
 202                                 ccs)
 203                             (while rest
 204                               (setq ccs (charset-name
 205                                          (find-charset (car rest))))
 206                               (if (setq ret
 207                                         (encode-char char ccs
 208                                                      'defined-only))
 209                                   (throw 'tag (cons ccs ret)))
 210                               (setq rest (cdr rest))))))
 211                   (setq char-spec (list ret))
 212                   (dolist (ccs (delq (car ret) (charset-list)))
 213                     (if (and (or (charset-iso-final-char ccs)
 214                                  (memq ccs
 215                                        '(=daikanwa
 216                                          =daikanwa@rev2
 217                                          ;; =gt-k
 218                                          )))
 219                              (setq ccs (charset-name ccs))
 220                              (null (assq ccs char-spec))
 221                              (setq ret (encode-char char ccs 'defined-only)))
 222                         (setq char-spec (cons (cons ccs ret) char-spec))))
 223                   (if (null char-spec)
 224                       (setq char-spec (split-char char)))
 225                   (cond ((setq ret (get-char-attribute char 'name))
 226                          (setq char-spec (cons (cons 'name ret) char-spec))
 227                          )
 228                         ((setq ret (get-char-attribute char 'name*))
 229                          (setq char-spec (cons (cons 'name* ret) char-spec))
 230                          ))
 231                   ))
 232            char-spec)
 233           ((consp char)
 234            char))))
 235
 236 (defun char-db-insert-char-spec (char &optional readable column
 237                                       required-features)
 238   (unless column
 239     (setq column (current-column)))
 240   (let (char-spec temp-char)
 241     (setq char-spec (char-db-make-char-spec char))
 242     (unless (or (characterp char) ; char
 243                 (condition-case nil
 244                     (setq char (find-char char-spec))
 245                   (error nil)))
 246       ;; define temporary character
 247       ;;   Current implementation is dirty.
 248       (setq temp-char (define-char (cons '(ideograph-daikanwa . 0)
 249                                          char-spec)))
 250       (remove-char-attribute temp-char 'ideograph-daikanwa)
 251       (setq char temp-char))
 252     (insert-char-attributes char
 253                             readable
 254                             (union (mapcar #'car char-spec)
 255                                    required-features))
 256     (when temp-char
 257       ;; undefine temporary character
 258       ;;   Current implementation is dirty.
 259       (setq char-spec (char-attribute-alist temp-char))
 260       (while char-spec
 261         (remove-char-attribute temp-char (car (car char-spec)))
 262         (setq char-spec (cdr char-spec))))))
 263
 264 (defun char-db-insert-alist (alist &optional readable column)
 265   (unless column
 266     (setq column (current-column)))
 267   (let ((line-breaking
 268          (concat "\n" (make-string (1+ column) ?\ )))
 269         name value
 270         ret al ; cal
 271         key
 272         lbs cell rest separator)
 273     (insert "(")
 274     (while alist
 275       (setq name (car (car alist))
 276             value (cdr (car alist)))
 277       (cond ((eq name 'char)
 278              (insert "(char . ")
 279              (if (setq ret (condition-case nil
 280                                (find-char value)
 281                              (error nil)))
 282                  (progn
 283                    (setq al nil
 284                          ;; cal nil
 285                          )
 286                    (while value
 287                      (setq key (car (car value)))
 288                      ;; (if (find-charset key)
 289                      ;;     (setq cal (cons key cal))
 290                      (setq al (cons key al))
 291                      ;; )
 292                      (setq value (cdr value)))
 293                    (insert-char-attributes ret
 294                                            readable
 295                                            (or al 'none) ; cal
 296                                            ))
 297                (insert (prin1-to-string value)))
 298              (insert ")")
 299              (insert line-breaking))
 300             ((consp value)
 301              (insert (format "(%-18s " name))
 302              (setq lbs (concat "\n" (make-string (current-column) ?\ )))
 303              (while (consp value)
 304                (setq cell (car value))
 305                (if (and (consp cell)
 306                         (consp (car cell))
 307                         (setq ret (condition-case nil
 308                                       (find-char cell)
 309                                     (error nil)))
 310                         )
 311                    (progn
 312                      (setq rest cell
 313                            al nil
 314                            ;; cal nil
 315                            )
 316                      (while rest
 317                        (setq key (car (car rest)))
 318                        ;; (if (find-charset key)
 319                        ;;     (setq cal (cons key cal))
 320                        (setq al (cons key al))
 321                        ;; )
 322                        (setq rest (cdr rest)))
 323                      (if separator
 324                          (insert lbs))
 325                      (insert-char-attributes ret
 326                                              readable
 327                                              al ; cal
 328                                              )
 329                      (setq separator lbs))
 330                  (if separator
 331                      (insert separator))
 332                  (insert (prin1-to-string cell))
 333                  (setq separator " "))
 334                (setq value (cdr value)))
 335              (insert ")")
 336              (insert line-breaking))
 337             (t
 338              (insert (format "(%-18s . %S)%s"
 339                              name value
 340                              line-breaking))))
 341       (setq alist (cdr alist))))
 342   (insert ")"))
 343
 344 (defun char-db-insert-char-reference (plist &optional readable column)
 345   (unless column
 346     (setq column (current-column)))
 347   (let ((line-breaking
 348          (concat "\n" (make-string (1+ column) ?\ )))
 349         (separator "")
 350         name value)
 351     (insert "(")
 352     (while plist
 353       (setq name (pop plist))
 354       (setq value (pop plist))
 355       (cond ((eq name :char)
 356              (insert separator)
 357              (insert ":char\t")
 358              (cond ((numberp value)
 359                     (setq value (decode-char '=ucs value)))
 360                    ;; ((consp value)
 361                    ;;  (setq value (or (find-char value)
 362                    ;;                  value)))
 363                    )
 364              (char-db-insert-char-spec value readable)
 365              (insert line-breaking)
 366              (setq separator ""))
 367             ((eq name :radical)
 368              (insert (format "%s%s\t%d ; %c%s"
 369                              separator
 370                              name value
 371                              (ideographic-radical value)
 372                              line-breaking))
 373              (setq separator ""))
 374             (t
 375              (insert (format "%s%s\t%S" separator name value))
 376              (setq separator line-breaking)))
 377       ))
 378   (insert ")"))
 379
 380 (defun char-db-decode-isolated-char (ccs code-point)
 381   (let (ret)
 382     (setq ret
 383           (cond ((eq ccs 'arabic-iso8859-6)
 384                  (decode-char ccs code-point))
 385                 ((and (memq ccs '(=gt-pj-1
 386                                   =gt-pj-2
 387                                   =gt-pj-3
 388                                   =gt-pj-4
 389                                   =gt-pj-5
 390                                   =gt-pj-6
 391                                   =gt-pj-7
 392                                   =gt-pj-8
 393                                   =gt-pj-9
 394                                   =gt-pj-10
 395                                   =gt-pj-11))
 396                       (setq ret (decode-char ccs code-point))
 397                       (setq ret (encode-char ret '=gt 'defined-only)))
 398                  (decode-builtin-char '=gt ret))
 399                 (t
 400                  (decode-builtin-char ccs code-point))))
 401     (cond ((and (<= 0 (char-int ret))
 402                 (<= (char-int ret) #x1F))
 403            (decode-char '=ucs (+ #x2400 (char-int ret))))
 404           ((= (char-int ret) #x7F)
 405            ?\u2421)
 406           (t ret))))
 407
 408 (defvar char-db-convert-obsolete-format t)
 409
 410 (defun char-db-insert-ccs-feature (name value line-breaking)
 411   (insert
 412    (format
 413     (cond ((memq name '(=daikanwa
 414                         =daikanwa@rev1 =daikanwa@rev2
 415                         =gt =gt-k =cbeta =zinbun-oracle))
 416            "(%-18s . %05d)\t; %c")
 417           ((eq name 'mojikyo)
 418            "(%-18s . %06d)\t; %c")
 419           ((>= (charset-dimension name) 2)
 420            "(%-18s . #x%04X)\t; %c")
 421           (t
 422            "(%-18s . #x%02X)\t; %c"))
 423     name
 424     (if (= (charset-iso-graphic-plane name) 1)
 425         (logior value
 426                 (cond ((= (charset-dimension name) 1)
 427                        #x80)
 428                       ((= (charset-dimension name) 2)
 429                        #x8080)
 430                       ((= (charset-dimension name) 3)
 431                        #x808080)
 432                       (t 0)))
 433       value)
 434     (char-db-decode-isolated-char name value)))
 435   (if (and (= (charset-chars name) 94)
 436            (= (charset-dimension name) 2))
 437       (insert (format " [%02d-%02d]"
 438                       (- (lsh value -8) 32)
 439                       (- (logand value 255) 32))))
 440   (insert line-breaking))
 441
 442 (defun char-db-insert-relation-feature (char name value line-breaking
 443                                              ccss readable)
 444   (insert (format "(%-18s%s " name line-breaking))
 445   (let ((lbs (concat "\n" (make-string (current-column) ?\ )))
 446         separator cell sources required-features
 447         ret)
 448     (while (consp value)
 449       (setq cell (car value))
 450       (if (integerp cell)
 451           (setq cell (decode-char '=ucs cell)))
 452       (cond
 453        ((eq name '->subsumptive)
 454         (when (or (not (some (lambda (atr)
 455                                (get-char-attribute cell atr))
 456                              char-db-ignored-attributes))
 457                   (some (lambda (ccs)
 458                           (encode-char cell ccs 'defined-only))
 459                         ccss))
 460           (if separator
 461               (insert lbs))
 462           (let ((char-db-ignored-attributes
 463                  (cons '<-subsumptive
 464                        char-db-ignored-attributes)))
 465             (insert-char-attributes cell readable))
 466           (setq separator lbs))
 467         )
 468        ((characterp cell)
 469         (setq sources
 470               (get-char-attribute
 471                char (intern (format "%s*sources" name))))
 472         (setq required-features nil)
 473         (dolist (source sources)
 474           (cond
 475            ((memq source '(JP JP/Jouyou shinjigen-1))
 476             (setq required-features
 477                   (union required-features
 478                          '(=jis-x0208
 479                            =jis-x0208@1990
 480                            =jis-x0213-1-2000
 481                            =jis-x0213-2-2000
 482                            =jis-x0212
 483                            =jis-x0208@1983
 484                            =jis-x0208@1978))))
 485            ((eq source 'CN)
 486             (setq required-features
 487                   (union required-features
 488                          '(=gb2312
 489                            =gb12345
 490                            =iso-ir165)))))
 491           (cond
 492            ((find-charset (setq ret (intern (format "=%s" source))))
 493             (setq required-features
 494                   (cons ret required-features)))
 495            (t (setq required-features
 496                     (cons source required-features)))))
 497         (cond ((string-match "@JP" (symbol-name name))
 498                (setq required-features
 499                      (union required-features
 500                             '(=jis-x0208
 501                               =jis-x0208@1990
 502                               =jis-x0213-1-2000
 503                               =jis-x0213-2-2000
 504                               =jis-x0212
 505                               =jis-x0208@1983
 506                               =jis-x0208@1978))))
 507               ((string-match "@CN" (symbol-name name))
 508                (setq required-features
 509                      (union required-features
 510                             '(=gb2312
 511                               =gb12345
 512                               =iso-ir165)))))
 513         (if separator
 514             (insert lbs))
 515         (if readable
 516             (insert (format "%S" cell))
 517           (char-db-insert-char-spec cell readable
 518                                     nil
 519                                     required-features))
 520         (setq separator lbs))
 521        ((consp cell)
 522         (if separator
 523             (insert lbs))
 524         (if (consp (car cell))
 525             (char-db-insert-char-spec cell readable)
 526           (char-db-insert-char-reference cell readable))
 527         (setq separator lbs))
 528        (t
 529         (if separator
 530             (insert separator))
 531         (insert (prin1-to-string cell))
 532         (setq separator " ")))
 533       (setq value (cdr value)))
 534     (insert ")")
 535     (insert line-breaking)))
 536
 537 (defun insert-char-attributes (char &optional readable attributes column)
 538   (unless column
 539     (setq column (current-column)))
 540   (let (name value ; has-long-ccs-name
 541         rest
 542         radical strokes
 543         (line-breaking
 544          (concat "\n" (make-string (1+ column) ?\ )))
 545         lbs cell separator ret
 546         key al cal
 547         dest-ccss ; sources required-features
 548         ccss)
 549     (let (atr-d)
 550       (setq attributes
 551             (sort (if attributes
 552                       (if (consp attributes)
 553                           (progn
 554                             (dolist (name attributes)
 555                               (unless (memq name char-db-ignored-attributes)
 556                                 (if (find-charset name)
 557                                     (push name ccss))
 558                                 (push name atr-d)))
 559                             atr-d))
 560                     (dolist (name (char-attribute-list))
 561                       (unless (memq name char-db-ignored-attributes)
 562                         (if (find-charset name)
 563                             (push name ccss))
 564                         (push name atr-d)))
 565                     atr-d)
 566                   #'char-attribute-name<)))
 567     (insert "(")
 568     (when (memq '<-subsumptive attributes)
 569       (unless readable
 570         (when (setq value (get-char-attribute char '<-subsumptive))
 571           (char-db-insert-relation-feature char '<-subsumptive value
 572                                            line-breaking
 573                                            ccss readable)))
 574       (setq attributes (delq '<-subsumptive attributes)))
 575     (when (and (memq '<-denotational attributes)
 576                (setq value (get-char-attribute char '<-denotational)))
 577       (char-db-insert-relation-feature char '<-denotational value
 578                                        line-breaking
 579                                        ccss readable)
 580       (setq attributes (delq '<-denotational attributes)))
 581     (when (and (memq 'name attributes)
 582                (setq value (get-char-attribute char 'name)))
 583       (insert (format
 584                (if (> (+ (current-column) (length value)) 48)
 585                    "(name . %S)%s"
 586                  "(name               . %S)%s")
 587                value line-breaking))
 588       (setq attributes (delq 'name attributes))
 589       )
 590     (when (and (memq 'name* attributes)
 591                (setq value (get-char-attribute char 'name*)))
 592       (insert (format
 593                (if (> (+ (current-column) (length value)) 48)
 594                    "(name* . %S)%s"
 595                  "(name*              . %S)%s")
 596                value line-breaking))
 597       (setq attributes (delq 'name* attributes))
 598       )
 599     (when (and (memq 'script attributes)
 600                (setq value (get-char-attribute char 'script)))
 601       (insert (format "(script\t\t%s)%s"
 602                       (mapconcat (function prin1-to-string)
 603                                  value " ")
 604                       line-breaking))
 605       (setq attributes (delq 'script attributes))
 606       )
 607     (dolist (name '(=>ucs =>ucs*))
 608       (when (and (memq name attributes)
 609                  (setq value (get-char-attribute char name)))
 610         (insert (format "(%-18s . #x%04X)\t; %c%s"
 611                         name value (decode-char '=ucs value)
 612                         line-breaking))
 613         (setq attributes (delq name attributes))))
 614     (dolist (name '(=>ucs@gb =>ucs@cns =>ucs@jis =>ucs@ks =>ucs@big5))
 615       (when (and (memq name attributes)
 616                  (setq value (get-char-attribute char name)))
 617         (insert (format "(%-18s . #x%04X)\t; %c%s"
 618                         name value
 619                         (decode-char (intern
 620                                       (concat "="
 621                                               (substring
 622                                                (symbol-name name) 2)))
 623                                      value)
 624                         line-breaking))
 625         (setq attributes (delq name attributes))
 626         ))
 627     (dolist (name '(=>daikanwa))
 628       (when (and (memq name attributes)
 629                  (setq value (get-char-attribute char name)))
 630         (insert
 631          (if (integerp value)
 632              (format "(%-18s . %05d)\t; %c%s"
 633                      name value (decode-char '=daikanwa value)
 634                      line-breaking)
 635            (format "(%-18s %s)\t; %c%s"
 636                    name
 637                    (mapconcat (function prin1-to-string)
 638                               value " ")
 639                    (char-representative-of-daikanwa char)
 640                    line-breaking)))
 641         (setq attributes (delq name attributes))))
 642     (when (and (memq 'general-category attributes)
 643                (setq value (get-char-attribute char 'general-category)))
 644       (insert (format
 645                "(general-category\t%s) ; %s%s"
 646                (mapconcat (lambda (cell)
 647                             (format "%S" cell))
 648                           value " ")
 649                (cond ((rassoc value unidata-normative-category-alist)
 650                       "Normative Category")
 651                      ((rassoc value unidata-informative-category-alist)
 652                       "Informative Category")
 653                      (t
 654                       "Unknown Category"))
 655                line-breaking))
 656       (setq attributes (delq 'general-category attributes))
 657       )
 658     (when (and (memq 'bidi-category attributes)
 659                (setq value (get-char-attribute char 'bidi-category)))
 660       (insert (format "(bidi-category\t. %S)%s"
 661                       value
 662                       line-breaking))
 663       (setq attributes (delq 'bidi-category attributes))
 664       )
 665     (unless (or (not (memq 'mirrored attributes))
 666                 (eq (setq value (get-char-attribute char 'mirrored 'empty))
 667                     'empty))
 668       (insert (format "(mirrored\t\t. %S)%s"
 669                       value
 670                       line-breaking))
 671       (setq attributes (delq 'mirrored attributes))
 672       )
 673     (cond
 674      ((and (memq 'decimal-digit-value attributes)
 675            (setq value (get-char-attribute char 'decimal-digit-value)))
 676       (insert (format "(decimal-digit-value . %S)%s"
 677                       value
 678                       line-breaking))
 679       (setq attributes (delq 'decimal-digit-value attributes))
 680       (when (and (memq 'digit-value attributes)
 681                  (setq value (get-char-attribute char 'digit-value)))
 682         (insert (format "(digit-value\t . %S)%s"
 683                         value
 684                         line-breaking))
 685         (setq attributes (delq 'digit-value attributes))
 686         )
 687       (when (and (memq 'numeric-value attributes)
 688                  (setq value (get-char-attribute char 'numeric-value)))
 689         (insert (format "(numeric-value\t . %S)%s"
 690                         value
 691                         line-breaking))
 692         (setq attributes (delq 'numeric-value attributes))
 693         )
 694       )
 695      (t
 696       (when (and (memq 'digit-value attributes)
 697                  (setq value (get-char-attribute char 'digit-value)))
 698         (insert (format "(digit-value\t. %S)%s"
 699                         value
 700                         line-breaking))
 701         (setq attributes (delq 'digit-value attributes))
 702         )
 703       (when (and (memq 'numeric-value attributes)
 704                  (setq value (get-char-attribute char 'numeric-value)))
 705         (insert (format "(numeric-value\t. %S)%s"
 706                         value
 707                         line-breaking))
 708         (setq attributes (delq 'numeric-value attributes))
 709         )))
 710     (when (and (memq 'iso-10646-comment attributes)
 711                (setq value (get-char-attribute char 'iso-10646-comment)))
 712       (insert (format "(iso-10646-comment\t. %S)%s"
 713                       value
 714                       line-breaking))
 715       (setq attributes (delq 'iso-10646-comment attributes))
 716       )
 717     (when (and (memq 'morohashi-daikanwa attributes)
 718                (setq value (get-char-attribute char 'morohashi-daikanwa)))
 719       (insert (format "(morohashi-daikanwa\t%s)%s"
 720                       (mapconcat (function prin1-to-string) value " ")
 721                       line-breaking))
 722       (setq attributes (delq 'morohashi-daikanwa attributes))
 723       )
 724     (setq radical nil
 725           strokes nil)
 726     (when (and (memq 'ideographic-radical attributes)
 727                (setq value (get-char-attribute char 'ideographic-radical)))
 728       (setq radical value)
 729       (insert (format "(ideographic-radical . %S)\t; %c%s"
 730                       radical
 731                       (ideographic-radical radical)
 732                       line-breaking))
 733       (setq attributes (delq 'ideographic-radical attributes))
 734       )
 735     (when (and (memq 'shuowen-radical attributes)
 736                (setq value (get-char-attribute char 'shuowen-radical)))
 737       (insert (format "(shuowen-radical\t. %S)\t; %c%s"
 738                       value
 739                       (shuowen-radical value)
 740                       line-breaking))
 741       (setq attributes (delq 'shuowen-radical attributes))
 742       )
 743     (let (key)
 744       (dolist (domain
 745                (append
 746                 char-db-feature-domains
 747                 (let (dest domain)
 748                   (dolist (feature (char-attribute-list))
 749                     (setq feature (symbol-name feature))
 750                     (when (string-match
 751                            "\\(radical\\|strokes\\)@\\([^@*]+\\)\\(\\*\\|$\\)"
 752                            feature)
 753                       (setq domain (intern (match-string 2 feature)))
 754                      (unless (memq domain dest)
 755                        (setq dest (cons domain dest)))))
 756                   (sort dest #'string<))))
 757         (setq key (intern (format "%s@%s" 'ideographic-radical domain)))
 758         (when (and (memq key attributes)
 759                    (setq value (get-char-attribute char key)))
 760           (setq radical value)
 761           (insert (format "(%s . %S)\t; %c%s"
 762                           key
 763                           radical
 764                           (ideographic-radical radical)
 765                           line-breaking))
 766           (setq attributes (delq key attributes))
 767           )
 768         (setq key (intern (format "%s@%s" 'ideographic-strokes domain)))
 769         (when (and (memq key attributes)
 770                    (setq value (get-char-attribute char key)))
 771           (setq strokes value)
 772           (insert (format "(%s . %S)%s"
 773                           key
 774                           strokes
 775                           line-breaking))
 776           (setq attributes (delq key attributes))
 777           )
 778         (setq key (intern (format "%s@%s" 'total-strokes domain)))
 779         (when (and (memq key attributes)
 780                    (setq value (get-char-attribute char key)))
 781           (insert (format "(%s       . %S)%s"
 782                           key
 783                           value
 784                           line-breaking))
 785           (setq attributes (delq key attributes))
 786           )
 787         (dolist (feature '(ideographic-radical
 788                            ideographic-strokes
 789                            total-strokes))
 790           (setq key (intern (format "%s@%s*sources" feature domain)))
 791           (when (and (memq key attributes)
 792                      (setq value (get-char-attribute char key)))
 793             (insert (format "(%s%s" key line-breaking))
 794             (dolist (cell value)
 795               (insert (format " %s" cell)))
 796             (insert ")")
 797             (insert line-breaking)
 798             (setq attributes (delq key attributes))
 799             ))
 800         ))
 801     (when (and (memq 'ideographic-strokes attributes)
 802                (setq value (get-char-attribute char 'ideographic-strokes)))
 803       (setq strokes value)
 804       (insert (format "(ideographic-strokes . %S)%s"
 805                       strokes
 806                       line-breaking))
 807       (setq attributes (delq 'ideographic-strokes attributes))
 808       )
 809     (when (and (memq 'kangxi-radical attributes)
 810                (setq value (get-char-attribute char 'kangxi-radical)))
 811       (unless (eq value radical)
 812         (insert (format "(kangxi-radical\t . %S)\t; %c%s"
 813                         value
 814                         (ideographic-radical value)
 815                         line-breaking))
 816         (or radical
 817             (setq radical value)))
 818       (setq attributes (delq 'kangxi-radical attributes))
 819       )
 820     (when (and (memq 'kangxi-strokes attributes)
 821                (setq value (get-char-attribute char 'kangxi-strokes)))
 822       (unless (eq value strokes)
 823         (insert (format "(kangxi-strokes\t . %S)%s"
 824                         value
 825                         line-breaking))
 826         (or strokes
 827             (setq strokes value)))
 828       (setq attributes (delq 'kangxi-strokes attributes))
 829       )
 830     (when (and (memq 'japanese-radical attributes)
 831                (setq value (get-char-attribute char 'japanese-radical)))
 832       (unless (eq value radical)
 833         (insert (format "(japanese-radical\t . %S)\t; %c%s"
 834                         value
 835                         (ideographic-radical value)
 836                         line-breaking))
 837         (or radical
 838             (setq radical value)))
 839       (setq attributes (delq 'japanese-radical attributes))
 840       )
 841     (when (and (memq 'japanese-strokes attributes)
 842                (setq value (get-char-attribute char 'japanese-strokes)))
 843       (unless (eq value strokes)
 844         (insert (format "(japanese-strokes\t . %S)%s"
 845                         value
 846                         line-breaking))
 847         (or strokes
 848             (setq strokes value)))
 849       (setq attributes (delq 'japanese-strokes attributes))
 850       )
 851     (when (and (memq 'cns-radical attributes)
 852                (setq value (get-char-attribute char 'cns-radical)))
 853       (insert (format "(cns-radical\t . %S)\t; %c%s"
 854                       value
 855                       (ideographic-radical value)
 856                       line-breaking))
 857       (setq attributes (delq 'cns-radical attributes))
 858       )
 859     (when (and (memq 'cns-strokes attributes)
 860                (setq value (get-char-attribute char 'cns-strokes)))
 861       (unless (eq value strokes)
 862         (insert (format "(cns-strokes\t . %S)%s"
 863                         value
 864                         line-breaking))
 865         (or strokes
 866             (setq strokes value)))
 867       (setq attributes (delq 'cns-strokes attributes))
 868       )
 869     (when (and (memq 'shinjigen-1-radical attributes)
 870                (setq value (get-char-attribute char 'shinjigen-1-radical)))
 871       (unless (eq value radical)
 872         (insert (format "(shinjigen-1-radical . %S)\t; %c%s"
 873                         value
 874                         (ideographic-radical value)
 875                         line-breaking))
 876         (or radical
 877             (setq radical value)))
 878       (setq attributes (delq 'shinjigen-1-radical attributes))
 879       )
 880     (when (and (memq 'ideographic- attributes)
 881                (setq value (get-char-attribute char 'ideographic-)))
 882       (insert "(ideographic-       ")
 883       (setq lbs (concat "\n" (make-string (current-column) ?\ ))
 884             separator nil)
 885       (while (consp value)
 886         (setq cell (car value))
 887         (if (integerp cell)
 888             (setq cell (decode-char '=ucs cell)))
 889         (cond ((characterp cell)
 890                (if separator
 891                    (insert lbs))
 892                (if readable
 893                    (insert (format "%S" cell))
 894                  (char-db-insert-char-spec cell readable))
 895                (setq separator lbs))
 896               ((consp cell)
 897                (if separator
 898                    (insert lbs))
 899                (if (consp (car cell))
 900                    (char-db-insert-char-spec cell readable)
 901                  (char-db-insert-char-reference cell readable))
 902                (setq separator lbs))
 903               (t
 904                (if separator
 905                    (insert separator))
 906                (insert (prin1-to-string cell))
 907                (setq separator " ")))
 908         (setq value (cdr value)))
 909       (insert ")")
 910       (insert line-breaking)
 911       (setq attributes (delq 'ideographic- attributes)))
 912     (when (and (memq 'total-strokes attributes)
 913                (setq value (get-char-attribute char 'total-strokes)))
 914       (insert (format "(total-strokes       . %S)%s"
 915                       value
 916                       line-breaking))
 917       (setq attributes (delq 'total-strokes attributes))
 918       )
 919     (when (and (memq '->ideograph attributes)
 920                (setq value (get-char-attribute char '->ideograph)))
 921       (insert (format "(->ideograph\t%s)%s"
 922                       (mapconcat (lambda (code)
 923                                    (cond ((symbolp code)
 924                                           (symbol-name code))
 925                                          ((integerp code)
 926                                           (format "#x%04X" code))
 927                                          (t
 928                                           (format "%s %S"
 929                                                   line-breaking code))))
 930                                  value " ")
 931                       line-breaking))
 932       (setq attributes (delq '->ideograph attributes))
 933       )
 934     ;; (when (and (memq '->decomposition attributes)
 935     ;;            (setq value (get-char-attribute char '->decomposition)))
 936     ;;   (insert (format "(->decomposition\t%s)%s"
 937     ;;                   (mapconcat (lambda (code)
 938     ;;                                (cond ((symbolp code)
 939     ;;                                       (symbol-name code))
 940     ;;                                      ((characterp code)
 941     ;;                                       (if readable
 942     ;;                                           (format "%S" code)
 943     ;;                                         (format "#x%04X"
 944     ;;                                                 (char-int code))
 945     ;;                                         ))
 946     ;;                                      ((integerp code)
 947     ;;                                       (format "#x%04X" code))
 948     ;;                                      (t
 949     ;;                                       (format "%s%S" line-breaking code))))
 950     ;;                              value " ")
 951     ;;                   line-breaking))
 952     ;;   (setq attributes (delq '->decomposition attributes))
 953     ;;   )
 954     (if (equal (get-char-attribute char '->titlecase)
 955                (get-char-attribute char '->uppercase))
 956         (setq attributes (delq '->titlecase attributes)))
 957     (when (and (memq '->mojikyo attributes)
 958                (setq value (get-char-attribute char '->mojikyo)))
 959       (insert (format "(->mojikyo\t\t. %06d)\t; %c%s"
 960                       value (decode-char 'mojikyo value)
 961                       line-breaking))
 962       (setq attributes (delq '->mojikyo attributes))
 963       )
 964     (when (and (memq 'hanyu-dazidian-vol attributes)
 965                (setq value (get-char-attribute char 'hanyu-dazidian-vol)))
 966       (insert (format "(hanyu-dazidian-vol  . %d)%s"
 967                       value line-breaking))
 968       (setq attributes (delq 'hanyu-dazidian-vol attributes))
 969       )
 970     (when (and (memq 'hanyu-dazidian-page attributes)
 971                (setq value (get-char-attribute char 'hanyu-dazidian-page)))
 972       (insert (format "(hanyu-dazidian-page . %d)%s"
 973                       value line-breaking))
 974       (setq attributes (delq 'hanyu-dazidian-page attributes))
 975       )
 976     (when (and (memq 'hanyu-dazidian-char attributes)
 977                (setq value (get-char-attribute char 'hanyu-dazidian-char)))
 978       (insert (format "(hanyu-dazidian-char . %d)%s"
 979                       value line-breaking))
 980       (setq attributes (delq 'hanyu-dazidian-char attributes))
 981       )
 982     (unless readable
 983       (dolist (ignored '(composition
 984                          ->denotational <-subsumptive ->ucs-unified
 985                          ->ideographic-component-forms))
 986         (setq attributes (delq ignored attributes))))
 987     (while attributes
 988       (setq name (car attributes))
 989       (if (setq value (get-char-attribute char name))
 990           (cond ((setq ret (find-charset name))
 991                  (setq name (charset-name ret))
 992                  (if (and (not (memq name dest-ccss))
 993                           (prog1
 994                               (setq value (get-char-attribute char name))
 995                             (setq dest-ccss (cons name dest-ccss))))
 996                      (char-db-insert-ccs-feature name value line-breaking))
 997                  )
 998                 ((string-match "^=>ucs@" (symbol-name name))
 999                  (insert (format "(%-18s . #x%04X)\t; %c%s"
1000                                  name value (decode-char '=ucs value)
1001                                  line-breaking))
1002                  )
1003                 ((eq name 'jisx0208-1978/4X)
1004                  (insert (format "(%-18s . #x%04X)%s"
1005                                  name value
1006                                  line-breaking))
1007                  )
1008                 ((and (not readable)
1009                       (null (get-char-attribute
1010                              char
1011                              (intern (format "%s*sources" name))))
1012                       (not (string-match "\\*sources$" (symbol-name name)))
1013                       (or (eq name '<-identical)
1014                           (string-match "^->compat" (symbol-name name))
1015                           (string-match "^->halfwidth" (symbol-name name))
1016                           (and
1017                            (string-match "^->fullwidth" (symbol-name name))
1018                            (not
1019                             (and (consp value)
1020                                  (characterp (car value))
1021                                  (encode-char
1022                                   (car value) '=ucs 'defined-only))))
1023                           (string-match "^->simplified" (symbol-name name))
1024                           (string-match "^->vulgar" (symbol-name name))
1025                           (string-match "^->wrong" (symbol-name name))
1026                           (string-match "^->same" (symbol-name name))
1027                           (string-match "^->formed" (symbol-name name))
1028                           (string-match "^->original" (symbol-name name))
1029                           (string-match "^->ancient" (symbol-name name))
1030                           (string-match "^->Oracle-Bones" (symbol-name name))
1031                           ))
1032                  )
1033                 ((or (eq name 'ideographic-structure)
1034                      (eq name 'ideographic-combination)
1035                      (eq name 'ideographic-)
1036                      (eq name '=decomposition)
1037                      (string-match "^=>decomposition" (symbol-name name))
1038                      (string-match "^\\(->\\|<-\\)" (symbol-name name)))
1039                  (char-db-insert-relation-feature char name value
1040                                                   line-breaking
1041                                                   ccss readable))
1042                 ((memq name '(ideograph=
1043                               original-ideograph-of
1044                               ancient-ideograph-of
1045                               vulgar-ideograph-of
1046                               wrong-ideograph-of
1047                               ;; simplified-ideograph-of
1048                               ideographic-variants
1049                               ;; ideographic-different-form-of
1050                               ))
1051                  (insert (format "(%-18s%s " name line-breaking))
1052                  (setq lbs (concat "\n" (make-string (current-column) ?\ ))
1053                        separator nil)
1054                  (while (consp value)
1055                    (setq cell (car value))
1056                    (if (and (consp cell)
1057                             (consp (car cell)))
1058                        (progn
1059                          (if separator
1060                              (insert lbs))
1061                          (char-db-insert-alist cell readable)
1062                          (setq separator lbs))
1063                      (if separator
1064                          (insert separator))
1065                      (insert (prin1-to-string cell))
1066                      (setq separator " "))
1067                    (setq value (cdr value)))
1068                  (insert ")")
1069                  (insert line-breaking))
1070                 ((consp value)
1071                  (insert (format "(%-18s " name))
1072                  (setq lbs (concat "\n" (make-string (current-column) ?\ ))
1073                        separator nil)
1074                  (while (consp value)
1075                    (setq cell (car value))
1076                    (if (and (consp cell)
1077                             (consp (car cell))
1078                             (setq ret (condition-case nil
1079                                           (find-char cell)
1080                                         (error nil))))
1081                        (progn
1082                          (setq rest cell
1083                                al nil
1084                                cal nil)
1085                          (while rest
1086                            (setq key (car (car rest)))
1087                            (if (find-charset key)
1088                                (setq cal (cons key cal))
1089                              (setq al (cons key al)))
1090                            (setq rest (cdr rest)))
1091                          (if separator
1092                              (insert lbs))
1093                          (insert-char-attributes ret
1094                                                  readable
1095                                                  al cal)
1096                          (setq separator lbs))
1097                      (setq ret (prin1-to-string cell))
1098                      (if separator
1099                          (if (< (+ (current-column)
1100                                    (length ret)
1101                                    (length separator))
1102                                 76)
1103                              (insert separator)
1104                            (insert lbs)))
1105                      (insert ret)
1106                      (setq separator " "))
1107                    (setq value (cdr value)))
1108                  (insert ")")
1109                  (insert line-breaking))
1110                 (t
1111                  (insert (format "(%-18s . %S)%s"
1112                                  name value
1113                                  line-breaking)))
1114                 ))
1115       (setq attributes (cdr attributes)))
1116     (insert ")")))
1117
1118 (defun insert-char-data (char &optional readable
1119                               attributes)
1120   (save-restriction
1121     (narrow-to-region (point)(point))
1122     (insert "(define-char
1123   '")
1124     (insert-char-attributes char readable attributes)
1125     (insert ")\n")
1126     (goto-char (point-min))
1127     (while (re-search-forward "[ \t]+$" nil t)
1128       (replace-match ""))
1129     ;; from tabify.
1130     (goto-char (point-min))
1131     (while (re-search-forward "[ \t][ \t][ \t]*" nil t)
1132       (let ((column (current-column))
1133             (indent-tabs-mode t))
1134         (delete-region (match-beginning 0) (point))
1135         (indent-to column)))
1136     (goto-char (point-max))
1137     ;; (tabify (point-min)(point-max))
1138     ))
1139
1140 (defun insert-char-data-with-variant (char &optional printable
1141                                            no-ucs-unified
1142                                            script excluded-script)
1143   (insert-char-data char printable)
1144   (let ((variants (char-variants char))
1145         rest
1146         variant vs ret)
1147     (setq variants (sort variants #'<))
1148     (setq rest variants)
1149     (setq variants (cons char variants))
1150     (while rest
1151       (setq variant (car rest))
1152       (unless (get-char-attribute variant '<-subsumptive)
1153         (if (and (or (null script)
1154                      (null (setq vs (get-char-attribute variant 'script)))
1155                      (memq script vs))
1156                  (or (null excluded-script)
1157                      (null (setq vs (get-char-attribute variant 'script)))
1158                      (not (memq excluded-script vs))))
1159             (unless (and no-ucs-unified (get-char-attribute variant '=ucs))
1160               (insert-char-data variant printable)
1161               (if (setq ret (char-variants variant))
1162                   (while ret
1163                     (or (memq (car ret) variants)
1164                         ;; (get-char-attribute (car ret) '<-subsumptive)
1165                         (setq rest (nconc rest (list (car ret)))))
1166                     (setq ret (cdr ret)))))))
1167       (setq rest (cdr rest)))))
1168
1169 (defun insert-char-range-data (min max &optional script excluded-script)
1170   (let ((code min)
1171         char)
1172     (while (<= code max)
1173       (setq char (decode-char '=ucs code))
1174       (if (encode-char char '=ucs 'defined-only)
1175           (insert-char-data-with-variant char nil 'no-ucs-unified
1176                                          script excluded-script))
1177       (setq code (1+ code)))))
1178
1179 (defun write-char-range-data-to-file (min max file
1180                                           &optional script excluded-script)
1181   (let ((coding-system-for-write char-db-file-coding-system))
1182     (with-temp-buffer
1183       (insert (format ";; -*- coding: %s -*-\n"
1184                       char-db-file-coding-system))
1185       (insert-char-range-data min max script excluded-script)
1186       (write-region (point-min)(point-max) file))))
1187
1188 (defvar what-character-original-window-configuration)
1189
1190 ;;;###autoload
1191 (defun what-char-definition (char)
1192   (interactive (list (char-after)))
1193   (let ((buf (get-buffer-create "*Character Description*"))
1194         (the-buf (current-buffer))
1195         (win-conf (current-window-configuration)))
1196     (pop-to-buffer buf)
1197     (make-local-variable 'what-character-original-window-configuration)
1198     (setq what-character-original-window-configuration win-conf)
1199     (setq buffer-read-only nil)
1200     (erase-buffer)
1201     (condition-case err
1202         (progn
1203           (insert-char-data-with-variant char 'printable)
1204           (unless (char-attribute-alist char)
1205             (insert (format ";; = %c\n"
1206                             (let* ((rest (split-char char))
1207                                    (ccs (pop rest))
1208                                    (code (pop rest)))
1209                               (while rest
1210                                 (setq code (logior (lsh code 8)
1211                                                    (pop rest))))
1212                               (decode-char ccs code)))))
1213           ;; (char-db-update-comment)
1214           (set-buffer-modified-p nil)
1215           (view-mode the-buf (lambda (buf)
1216                                (set-window-configuration
1217                                 what-character-original-window-configuration)
1218                                ))
1219           (goto-char (point-min)))
1220       (error (progn
1221                (set-window-configuration
1222                 what-character-original-window-configuration)
1223                (signal (car err) (cdr err)))))))
1224
1225 (provide 'char-db-util)
1226
1227 ;;; char-db-util.el ends here