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
   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, Unicode, UCS-4, 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 (defvar char-db-file-coding-system 'utf-8-mcs-er)
  73
  74 (defvar char-db-feature-domains
  75   '(ucs daikanwa cns gt jis jis/alt jis/a jis/b
  76         jis-x0212 jis-x0213 cdp shinjigen misc unknown))
  77
  78 (defvar char-db-ignored-attributes '(ideographic-products))
  79
  80 (defun char-attribute-name< (ka kb)
  81   (cond
  82    ((eq '->denotational kb)
  83     t)
  84    ((eq '->subsumptive kb)
  85     (not (eq '->denotational ka)))
  86    ((eq '->denotational ka)
  87     nil)
  88    ((eq '->subsumptive ka)
  89     nil)
  90    ((find-charset ka)
  91     (if (find-charset kb)
  92         (if (<= (charset-id ka) 1)
  93             (if (<= (charset-id kb) 1)
  94                 (cond
  95                  ((= (charset-dimension ka)
  96                      (charset-dimension kb))
  97                   (> (charset-id ka)(charset-id kb)))
  98                  (t
  99                   (> (charset-dimension ka)
 100                      (charset-dimension kb))
 101                   ))
 102               t)
 103           (if (<= (charset-id kb) 1)
 104               nil
 105             (< (charset-id ka)(charset-id kb))))
 106       nil))
 107    ((find-charset kb)
 108     t)
 109    ((symbolp ka)
 110     (cond ((symbolp kb)
 111            (string< (symbol-name ka)
 112                     (symbol-name kb)))
 113           (t)))
 114    ((symbolp kb)
 115     nil)))
 116
 117 (defvar char-db-coded-charset-priority-list
 118   '(ascii
 119     control-1
 120     latin-iso8859-1
 121     latin-iso8859-2
 122     latin-iso8859-3
 123     latin-iso8859-4
 124     latin-iso8859-9
 125     latin-jisx0201
 126     cyrillic-iso8859-5
 127     greek-iso8859-7
 128     thai-tis620
 129     =jis-x0208
 130     =jis-x0208@1978
 131     =jis-x0208@1983
 132     japanese-jisx0212
 133     chinese-gb2312
 134     chinese-cns11643-1
 135     chinese-cns11643-2
 136     chinese-cns11643-3
 137     chinese-cns11643-4
 138     chinese-cns11643-5
 139     chinese-cns11643-6
 140     chinese-cns11643-7
 141     =jis-x0208@1990
 142     =jis-x0213-1-2000
 143     =jis-x0213-2-2000
 144     korean-ksc5601
 145     chinese-isoir165
 146     katakana-jisx0201
 147     hebrew-iso8859-8
 148     chinese-gb12345
 149     latin-viscii
 150     ethiopic-ucs
 151     =big5-cdp
 152     =gt
 153     ideograph-daikanwa-2
 154     ideograph-daikanwa
 155     =cbeta
 156     ideograph-hanziku-1
 157     ideograph-hanziku-2
 158     ideograph-hanziku-3
 159     ideograph-hanziku-4
 160     ideograph-hanziku-5
 161     ideograph-hanziku-6
 162     ideograph-hanziku-7
 163     ideograph-hanziku-8
 164     ideograph-hanziku-9
 165     ideograph-hanziku-10
 166     ideograph-hanziku-11
 167     ideograph-hanziku-12
 168     =gt-k
 169     =ucs@iso
 170     =ucs@unicode
 171     =big5
 172     =big5-eten
 173     =jis-x0208@1997
 174     =jef-china3))
 175
 176 (defun char-db-make-char-spec (char)
 177   (let (ret char-spec)
 178     (cond ((characterp char)
 179            (cond ((and (setq ret (encode-char char '=ucs 'defined-only))
 180                        (not (and (<= #xE000 ret)(<= ret #xF8FF))))
 181                   (setq char-spec (list (cons '=ucs ret)))
 182                   (cond ((setq ret (get-char-attribute char 'name))
 183                          (setq char-spec (cons (cons 'name ret) char-spec))
 184                          )
 185                         ((setq ret (get-char-attribute char 'name*))
 186                          (setq char-spec (cons (cons 'name* ret) char-spec))
 187                          ))
 188                   )
 189                  ((setq ret
 190                         (catch 'tag
 191                           (let ((rest char-db-coded-charset-priority-list)
 192                                 ccs)
 193                             (while rest
 194                               (setq ccs (charset-name
 195                                          (find-charset (car rest))))
 196                               (if (setq ret
 197                                         (encode-char char ccs
 198                                                      'defined-only))
 199                                   (throw 'tag (cons ccs ret)))
 200                               (setq rest (cdr rest))))))
 201                   (setq char-spec (list ret))
 202                   (dolist (ccs (delq (car ret) (charset-list)))
 203                     (if (and (or (charset-iso-final-char ccs)
 204                                  (memq ccs
 205                                        '(=daikanwa
 206                                          =daikanwa@rev2
 207                                          ;; =gt-k
 208                                          )))
 209                              (setq ccs (charset-name ccs))
 210                              (null (assq ccs char-spec))
 211                              (setq ret (encode-char char ccs 'defined-only)))
 212                         (setq char-spec (cons (cons ccs ret) char-spec))))
 213                   (if (null char-spec)
 214                       (setq char-spec (split-char char)))
 215                   (cond ((setq ret (get-char-attribute char 'name))
 216                          (setq char-spec (cons (cons 'name ret) char-spec))
 217                          )
 218                         ((setq ret (get-char-attribute char 'name*))
 219                          (setq char-spec (cons (cons 'name* ret) char-spec))
 220                          ))
 221                   ))
 222            char-spec)
 223           ((consp char)
 224            char))))
 225
 226 (defun char-db-insert-char-spec (char &optional readable column
 227                                       required-features)
 228   (unless column
 229     (setq column (current-column)))
 230   (let (char-spec temp-char)
 231     (setq char-spec (char-db-make-char-spec char))
 232     (unless (or (characterp char) ; char
 233                 (condition-case nil
 234                     (setq char (find-char char-spec))
 235                   (error nil)))
 236       ;; define temporary character
 237       ;;   Current implementation is dirty.
 238       (setq temp-char (define-char (cons '(ideograph-daikanwa . 0)
 239                                          char-spec)))
 240       (remove-char-attribute temp-char 'ideograph-daikanwa)
 241       (setq char temp-char))
 242     (insert-char-attributes char
 243                             readable
 244                             (union (mapcar #'car char-spec)
 245                                    required-features))
 246     (when temp-char
 247       ;; undefine temporary character
 248       ;;   Current implementation is dirty.
 249       (setq char-spec (char-attribute-alist temp-char))
 250       (while char-spec
 251         (remove-char-attribute temp-char (car (car char-spec)))
 252         (setq char-spec (cdr char-spec))))))
 253
 254 (defun char-db-insert-alist (alist &optional readable column)
 255   (unless column
 256     (setq column (current-column)))
 257   (let ((line-breaking
 258          (concat "\n" (make-string (1+ column) ?\ )))
 259         name value
 260         ret al ; cal
 261         key
 262         lbs cell rest separator)
 263     (insert "(")
 264     (while alist
 265       (setq name (car (car alist))
 266             value (cdr (car alist)))
 267       (cond ((eq name 'char)
 268              (insert "(char . ")
 269              (if (setq ret (condition-case nil
 270                                (find-char value)
 271                              (error nil)))
 272                  (progn
 273                    (setq al nil
 274                          ;; cal nil
 275                          )
 276                    (while value
 277                      (setq key (car (car value)))
 278                      ;; (if (find-charset key)
 279                      ;;     (setq cal (cons key cal))
 280                      (setq al (cons key al))
 281                      ;; )
 282                      (setq value (cdr value)))
 283                    (insert-char-attributes ret
 284                                            readable
 285                                            (or al 'none) ; cal
 286                                            ))
 287                (insert (prin1-to-string value)))
 288              (insert ")")
 289              (insert line-breaking))
 290             ((consp value)
 291              (insert (format "(%-18s " name))
 292              (setq lbs (concat "\n" (make-string (current-column) ?\ )))
 293              (while (consp value)
 294                (setq cell (car value))
 295                (if (and (consp cell)
 296                         (consp (car cell))
 297                         (setq ret (condition-case nil
 298                                       (find-char cell)
 299                                     (error nil)))
 300                         )
 301                    (progn
 302                      (setq rest cell
 303                            al nil
 304                            ;; cal nil
 305                            )
 306                      (while rest
 307                        (setq key (car (car rest)))
 308                        ;; (if (find-charset key)
 309                        ;;     (setq cal (cons key cal))
 310                        (setq al (cons key al))
 311                        ;; )
 312                        (setq rest (cdr rest)))
 313                      (if separator
 314                          (insert lbs))
 315                      (insert-char-attributes ret
 316                                              readable
 317                                              al ; cal
 318                                              )
 319                      (setq separator lbs))
 320                  (if separator
 321                      (insert separator))
 322                  (insert (prin1-to-string cell))
 323                  (setq separator " "))
 324                (setq value (cdr value)))
 325              (insert ")")
 326              (insert line-breaking))
 327             (t
 328              (insert (format "(%-18s . %S)%s"
 329                              name value
 330                              line-breaking))))
 331       (setq alist (cdr alist))))
 332   (insert ")"))
 333
 334 (defun char-db-insert-char-reference (plist &optional readable column)
 335   (unless column
 336     (setq column (current-column)))
 337   (let ((line-breaking
 338          (concat "\n" (make-string (1+ column) ?\ )))
 339         (separator "")
 340         name value)
 341     (insert "(")
 342     (while plist
 343       (setq name (pop plist))
 344       (setq value (pop plist))
 345       (cond ((eq name :char)
 346              (insert separator)
 347              (insert ":char\t")
 348              (cond ((numberp value)
 349                     (setq value (decode-char '=ucs value)))
 350                    ;; ((consp value)
 351                    ;;  (setq value (or (find-char value)
 352                    ;;                  value)))
 353                    )
 354              (char-db-insert-char-spec value readable)
 355              (insert line-breaking)
 356              (setq separator ""))
 357             ((eq name :radical)
 358              (insert (format "%s%s\t%d ; %c%s"
 359                              separator
 360                              name value
 361                              (aref ideographic-radicals value)
 362                              line-breaking))
 363              (setq separator ""))
 364             (t
 365              (insert (format "%s%s\t%S" separator name value))
 366              (setq separator line-breaking)))
 367       ))
 368   (insert ")"))
 369
 370 (defun char-db-decode-isolated-char (ccs code-point)
 371   (let (ret)
 372     (setq ret
 373           (cond ((eq ccs 'arabic-iso8859-6)
 374                  (decode-char ccs code-point))
 375                 ((and (memq ccs '(=gt-pj-1
 376                                   =gt-pj-2
 377                                   =gt-pj-3
 378                                   =gt-pj-4
 379                                   =gt-pj-5
 380                                   =gt-pj-6
 381                                   =gt-pj-7
 382                                   =gt-pj-8
 383                                   =gt-pj-9
 384                                   =gt-pj-10
 385                                   =gt-pj-11))
 386                       (setq ret (decode-char ccs code-point))
 387                       (setq ret (encode-char ret '=gt 'defined-only)))
 388                  (decode-builtin-char '=gt ret))
 389                 (t
 390                  (decode-builtin-char ccs code-point))))
 391     (cond ((and (<= 0 (char-int ret))
 392                 (<= (char-int ret) #x1F))
 393            (decode-char '=ucs (+ #x2400 (char-int ret))))
 394           ((= (char-int ret) #x7F)
 395            ?\u2421)
 396           (t ret))))
 397
 398 (defvar char-db-convert-obsolete-format t)
 399
 400 (defun char-db-insert-ccs-feature (name value line-breaking)
 401   (insert
 402    (format
 403     (cond ((memq name '(=daikanwa
 404                         =daikanwa@rev1 =daikanwa@rev2
 405                         =gt =gt-k =cbeta))
 406            "(%-18s . %05d)\t; %c")
 407           ((eq name 'mojikyo)
 408            "(%-18s . %06d)\t; %c")
 409           ((>= (charset-dimension name) 2)
 410            "(%-18s . #x%04X)\t; %c")
 411           (t
 412            "(%-18s . #x%02X)\t; %c"))
 413     name
 414     (if (= (charset-iso-graphic-plane name) 1)
 415         (logior value
 416                 (cond ((= (charset-dimension name) 1)
 417                        #x80)
 418                       ((= (charset-dimension name) 2)
 419                        #x8080)
 420                       ((= (charset-dimension name) 3)
 421                        #x808080)
 422                       (t 0)))
 423       value)
 424     (char-db-decode-isolated-char name value)))
 425   (if (and (= (charset-chars name) 94)
 426            (= (charset-dimension name) 2))
 427       (insert (format " [%02d-%02d]"
 428                       (- (lsh value -8) 32)
 429                       (- (logand value 255) 32))))
 430   (insert line-breaking))
 431
 432 (defun insert-char-attributes (char &optional readable attributes column)
 433   (unless column
 434     (setq column (current-column)))
 435   (let (name value ; has-long-ccs-name
 436         rest
 437         radical strokes
 438         (line-breaking
 439          (concat "\n" (make-string (1+ column) ?\ )))
 440         lbs cell separator ret
 441         key al cal
 442         dest-ccss
 443         sources required-features
 444         ccss)
 445     (let (atr-d)
 446       (setq attributes
 447             (sort (if attributes
 448                       (if (consp attributes)
 449                           (progn
 450                             (dolist (name attributes)
 451                               (unless (memq name char-db-ignored-attributes)
 452                                 (if (find-charset name)
 453                                     (push name ccss))
 454                                 (push name atr-d)))
 455                             atr-d))
 456                     (dolist (name (char-attribute-list))
 457                       (unless (memq name char-db-ignored-attributes)
 458                         (if (find-charset name)
 459                             (push name ccss))
 460                         (push name atr-d)))
 461                     atr-d)
 462                   #'char-attribute-name<)))
 463     (insert "(")
 464     (when (and (memq 'name attributes)
 465                (setq value (get-char-attribute char 'name)))
 466       (insert (format
 467                (if (> (+ (current-column) (length value)) 48)
 468                    "(name . %S)%s"
 469                  "(name               . %S)%s")
 470                value line-breaking))
 471       (setq attributes (delq 'name attributes))
 472       )
 473     (when (and (memq 'name* attributes)
 474                (setq value (get-char-attribute char 'name*)))
 475       (insert (format
 476                (if (> (+ (current-column) (length value)) 48)
 477                    "(name* . %S)%s"
 478                  "(name*              . %S)%s")
 479                value line-breaking))
 480       (setq attributes (delq 'name* attributes))
 481       )
 482     (when (and (memq 'script attributes)
 483                (setq value (get-char-attribute char 'script)))
 484       (insert (format "(script\t\t%s)%s"
 485                       (mapconcat (function prin1-to-string)
 486                                  value " ")
 487                       line-breaking))
 488       (setq attributes (delq 'script attributes))
 489       )
 490     ;; (when (and (memq '<-denotational attributes)
 491     ;;            (setq value (get-char-attribute char '<-denotational))
 492     ;;            (null (cdr value))
 493     ;;            (setq value (encode-char (car value) 'ucs 'defined-only)))
 494     ;;   (insert (format "(%-18s . #x%04X)\t; %c%s"
 495     ;;                   '=>ucs value (decode-char 'ucs value)
 496     ;;                   line-breaking))
 497     ;;   (setq attributes (delq '<-denotational attributes)))
 498     (dolist (name '(=>ucs =>ucs*))
 499       (when (and (memq name attributes)
 500                  (setq value (get-char-attribute char name)))
 501         (insert (format "(%-18s . #x%04X)\t; %c%s"
 502                         name value (decode-char '=ucs value)
 503                         line-breaking))
 504         (setq attributes (delq name attributes))))
 505     (dolist (name '(=>ucs@gb =>ucs@cns =>ucs@jis =>ucs@ks =>ucs@big5))
 506       (when (and (memq name attributes)
 507                  (setq value (get-char-attribute char name)))
 508         (insert (format "(%-18s . #x%04X)\t; %c%s"
 509                         name value
 510                         (decode-char (intern
 511                                       (concat "="
 512                                               (substring
 513                                                (symbol-name name) 2)))
 514                                      value)
 515                         line-breaking))
 516         (setq attributes (delq name attributes))
 517         ))
 518     (dolist (name '(=>daikanwa))
 519       (when (and (memq name attributes)
 520                  (setq value (get-char-attribute char name)))
 521         (insert
 522          (if (integerp value)
 523              (format "(%-18s . %05d)\t; %c%s"
 524                      name value (decode-char '=daikanwa value)
 525                      line-breaking)
 526            (format "(%-18s %s)\t; %c%s"
 527                    name
 528                    (mapconcat (function prin1-to-string)
 529                               value " ")
 530                    (char-representative-of-daikanwa char)
 531                    line-breaking)))
 532         (setq attributes (delq name attributes))))
 533     (when (and (memq 'general-category attributes)
 534                (setq value (get-char-attribute char 'general-category)))
 535       (insert (format
 536                "(general-category\t%s) ; %s%s"
 537                (mapconcat (lambda (cell)
 538                             (format "%S" cell))
 539                           value " ")
 540                (cond ((rassoc value unidata-normative-category-alist)
 541                       "Normative Category")
 542                      ((rassoc value unidata-informative-category-alist)
 543                       "Informative Category")
 544                      (t
 545                       "Unknown Category"))
 546                line-breaking))
 547       (setq attributes (delq 'general-category attributes))
 548       )
 549     (when (and (memq 'bidi-category attributes)
 550                (setq value (get-char-attribute char 'bidi-category)))
 551       (insert (format "(bidi-category\t. %S)%s"
 552                       value
 553                       line-breaking))
 554       (setq attributes (delq 'bidi-category attributes))
 555       )
 556     (unless (or (not (memq 'mirrored attributes))
 557                 (eq (setq value (get-char-attribute char 'mirrored 'empty))
 558                     'empty))
 559       (insert (format "(mirrored\t\t. %S)%s"
 560                       value
 561                       line-breaking))
 562       (setq attributes (delq 'mirrored attributes))
 563       )
 564     (cond
 565      ((and (memq 'decimal-digit-value attributes)
 566            (setq value (get-char-attribute char 'decimal-digit-value)))
 567       (insert (format "(decimal-digit-value . %S)%s"
 568                       value
 569                       line-breaking))
 570       (setq attributes (delq 'decimal-digit-value attributes))
 571       (when (and (memq 'digit-value attributes)
 572                  (setq value (get-char-attribute char 'digit-value)))
 573         (insert (format "(digit-value\t . %S)%s"
 574                         value
 575                         line-breaking))
 576         (setq attributes (delq 'digit-value attributes))
 577         )
 578       (when (and (memq 'numeric-value attributes)
 579                  (setq value (get-char-attribute char 'numeric-value)))
 580         (insert (format "(numeric-value\t . %S)%s"
 581                         value
 582                         line-breaking))
 583         (setq attributes (delq 'numeric-value attributes))
 584         )
 585       )
 586      (t
 587       (when (and (memq 'digit-value attributes)
 588                  (setq value (get-char-attribute char 'digit-value)))
 589         (insert (format "(digit-value\t. %S)%s"
 590                         value
 591                         line-breaking))
 592         (setq attributes (delq 'digit-value attributes))
 593         )
 594       (when (and (memq 'numeric-value attributes)
 595                  (setq value (get-char-attribute char 'numeric-value)))
 596         (insert (format "(numeric-value\t. %S)%s"
 597                         value
 598                         line-breaking))
 599         (setq attributes (delq 'numeric-value attributes))
 600         )))
 601     (when (and (memq 'iso-10646-comment attributes)
 602                (setq value (get-char-attribute char 'iso-10646-comment)))
 603       (insert (format "(iso-10646-comment\t. %S)%s"
 604                       value
 605                       line-breaking))
 606       (setq attributes (delq 'iso-10646-comment attributes))
 607       )
 608     (when (and (memq 'morohashi-daikanwa attributes)
 609                (setq value (get-char-attribute char 'morohashi-daikanwa)))
 610       (insert (format "(morohashi-daikanwa\t%s)%s"
 611                       (mapconcat (function prin1-to-string) value " ")
 612                       line-breaking))
 613       (setq attributes (delq 'morohashi-daikanwa attributes))
 614       )
 615     (setq radical nil
 616           strokes nil)
 617     (when (and (memq 'ideographic-radical attributes)
 618                (setq value (get-char-attribute char 'ideographic-radical)))
 619       (setq radical value)
 620       (insert (format "(ideographic-radical . %S)\t; %c%s"
 621                       radical
 622                       (aref ideographic-radicals radical)
 623                       line-breaking))
 624       (setq attributes (delq 'ideographic-radical attributes))
 625       )
 626     (let (key)
 627       (dolist (domain
 628                (append
 629                 char-db-feature-domains
 630                 (let (dest domain)
 631                   (dolist (feature (char-attribute-list))
 632                     (setq feature (symbol-name feature))
 633                     (when (string-match
 634                            "\\(radical\\|strokes\\)@\\([^@*]+\\)\\(\\*\\|$\\)"
 635                            feature)
 636                       (setq domain (intern (match-string 2 feature)))
 637                      (unless (memq domain dest)
 638                        (setq dest (cons domain dest)))))
 639                   (sort dest #'string<))))
 640         (setq key (intern (format "%s@%s" 'ideographic-radical domain)))
 641         (when (and (memq key attributes)
 642                    (setq value (get-char-attribute char key)))
 643           (setq radical value)
 644           (insert (format "(%s . %S)\t; %c%s"
 645                           key
 646                           radical
 647                           (aref ideographic-radicals radical)
 648                           line-breaking))
 649           (setq attributes (delq key attributes))
 650           )
 651         (setq key (intern (format "%s@%s" 'ideographic-strokes domain)))
 652         (when (and (memq key attributes)
 653                    (setq value (get-char-attribute char key)))
 654           (setq strokes value)
 655           (insert (format "(%s . %S)%s"
 656                           key
 657                           strokes
 658                           line-breaking))
 659           (setq attributes (delq key attributes))
 660           )
 661         (setq key (intern (format "%s@%s" 'total-strokes domain)))
 662         (when (and (memq key attributes)
 663                    (setq value (get-char-attribute char key)))
 664           (insert (format "(%s       . %S)%s"
 665                           key
 666                           value
 667                           line-breaking))
 668           (setq attributes (delq key attributes))
 669           )
 670         (dolist (feature '(ideographic-radical
 671                            ideographic-strokes
 672                            total-strokes))
 673           (setq key (intern (format "%s@%s*sources" feature domain)))
 674           (when (and (memq key attributes)
 675                      (setq value (get-char-attribute char key)))
 676             (insert (format "(%s%s" key line-breaking))
 677             (dolist (cell value)
 678               (insert (format " %s" cell)))
 679             (insert ")")
 680             (insert line-breaking)
 681             (setq attributes (delq key attributes))
 682             ))
 683         ))
 684     (when (and (memq 'ideographic-strokes attributes)
 685                (setq value (get-char-attribute char 'ideographic-strokes)))
 686       (setq strokes value)
 687       (insert (format "(ideographic-strokes . %S)%s"
 688                       strokes
 689                       line-breaking))
 690       (setq attributes (delq 'ideographic-strokes attributes))
 691       )
 692     (when (and (memq 'kangxi-radical attributes)
 693                (setq value (get-char-attribute char 'kangxi-radical)))
 694       (unless (eq value radical)
 695         (insert (format "(kangxi-radical\t . %S)\t; %c%s"
 696                         value
 697                         (aref ideographic-radicals value)
 698                         line-breaking))
 699         (or radical
 700             (setq radical value)))
 701       (setq attributes (delq 'kangxi-radical attributes))
 702       )
 703     (when (and (memq 'kangxi-strokes attributes)
 704                (setq value (get-char-attribute char 'kangxi-strokes)))
 705       (unless (eq value strokes)
 706         (insert (format "(kangxi-strokes\t . %S)%s"
 707                         value
 708                         line-breaking))
 709         (or strokes
 710             (setq strokes value)))
 711       (setq attributes (delq 'kangxi-strokes attributes))
 712       )
 713     (when (and (memq 'japanese-radical attributes)
 714                (setq value (get-char-attribute char 'japanese-radical)))
 715       (unless (eq value radical)
 716         (insert (format "(japanese-radical\t . %S)\t; %c%s"
 717                         value
 718                         (aref ideographic-radicals value)
 719                         line-breaking))
 720         (or radical
 721             (setq radical value)))
 722       (setq attributes (delq 'japanese-radical attributes))
 723       )
 724     (when (and (memq 'japanese-strokes attributes)
 725                (setq value (get-char-attribute char 'japanese-strokes)))
 726       (unless (eq value strokes)
 727         (insert (format "(japanese-strokes\t . %S)%s"
 728                         value
 729                         line-breaking))
 730         (or strokes
 731             (setq strokes value)))
 732       (setq attributes (delq 'japanese-strokes attributes))
 733       )
 734     (when (and (memq 'cns-radical attributes)
 735                (setq value (get-char-attribute char 'cns-radical)))
 736       (insert (format "(cns-radical\t . %S)\t; %c%s"
 737                       value
 738                       (aref ideographic-radicals value)
 739                       line-breaking))
 740       (setq attributes (delq 'cns-radical attributes))
 741       )
 742     (when (and (memq 'cns-strokes attributes)
 743                (setq value (get-char-attribute char 'cns-strokes)))
 744       (unless (eq value strokes)
 745         (insert (format "(cns-strokes\t . %S)%s"
 746                         value
 747                         line-breaking))
 748         (or strokes
 749             (setq strokes value)))
 750       (setq attributes (delq 'cns-strokes attributes))
 751       )
 752     (when (and (memq 'shinjigen-1-radical attributes)
 753                (setq value (get-char-attribute char 'shinjigen-1-radical)))
 754       (unless (eq value radical)
 755         (insert (format "(shinjigen-1-radical . %S)\t; %c%s"
 756                         value
 757                         (aref ideographic-radicals value)
 758                         line-breaking))
 759         (or radical
 760             (setq radical value)))
 761       (setq attributes (delq 'shinjigen-1-radical attributes))
 762       )
 763     (when (and (memq 'ideographic- attributes)
 764                (setq value (get-char-attribute char 'ideographic-)))
 765       (insert "(ideographic-       ")
 766       (setq lbs (concat "\n" (make-string (current-column) ?\ ))
 767             separator nil)
 768       (while (consp value)
 769         (setq cell (car value))
 770         (if (integerp cell)
 771             (setq cell (decode-char '=ucs cell)))
 772         (cond ((characterp cell)
 773                (if separator
 774                    (insert lbs))
 775                (if readable
 776                    (insert (format "%S" cell))
 777                  (char-db-insert-char-spec cell readable))
 778                (setq separator lbs))
 779               ((consp cell)
 780                (if separator
 781                    (insert lbs))
 782                (if (consp (car cell))
 783                    (char-db-insert-char-spec cell readable)
 784                  (char-db-insert-char-reference cell readable))
 785                (setq separator lbs))
 786               (t
 787                (if separator
 788                    (insert separator))
 789                (insert (prin1-to-string cell))
 790                (setq separator " ")))
 791         (setq value (cdr value)))
 792       (insert ")")
 793       (insert line-breaking)
 794       (setq attributes (delq 'ideographic- attributes)))
 795     (when (and (memq 'total-strokes attributes)
 796                (setq value (get-char-attribute char 'total-strokes)))
 797       (insert (format "(total-strokes       . %S)%s"
 798                       value
 799                       line-breaking))
 800       (setq attributes (delq 'total-strokes attributes))
 801       )
 802     (when (and (memq '->ideograph attributes)
 803                (setq value (get-char-attribute char '->ideograph)))
 804       (insert (format "(->ideograph\t%s)%s"
 805                       (mapconcat (lambda (code)
 806                                    (cond ((symbolp code)
 807                                           (symbol-name code))
 808                                          ((integerp code)
 809                                           (format "#x%04X" code))
 810                                          (t
 811                                           (format "%s %S"
 812                                                   line-breaking code))))
 813                                  value " ")
 814                       line-breaking))
 815       (setq attributes (delq '->ideograph attributes))
 816       )
 817     (when (and (memq '->decomposition attributes)
 818                (setq value (get-char-attribute char '->decomposition)))
 819       (insert (format "(->decomposition\t%s)%s"
 820                       (mapconcat (lambda (code)
 821                                    (cond ((symbolp code)
 822                                           (symbol-name code))
 823                                          ((characterp code)
 824                                           (if readable
 825                                               (format "%S" code)
 826                                             (format "#x%04X"
 827                                                     (char-int code))
 828                                             ))
 829                                          ((integerp code)
 830                                           (format "#x%04X" code))
 831                                          (t
 832                                           (format "%s%S" line-breaking code))))
 833                                  value " ")
 834                       line-breaking))
 835       (setq attributes (delq '->decomposition attributes))
 836       )
 837     (if (equal (get-char-attribute char '->titlecase)
 838                (get-char-attribute char '->uppercase))
 839         (setq attributes (delq '->titlecase attributes)))
 840     (when (and (memq '->mojikyo attributes)
 841                (setq value (get-char-attribute char '->mojikyo)))
 842       (insert (format "(->mojikyo\t\t. %06d)\t; %c%s"
 843                       value (decode-char 'mojikyo value)
 844                       line-breaking))
 845       (setq attributes (delq '->mojikyo attributes))
 846       )
 847     (when (and (memq 'hanyu-dazidian-vol attributes)
 848                (setq value (get-char-attribute char 'hanyu-dazidian-vol)))
 849       (insert (format "(hanyu-dazidian-vol  . %d)%s"
 850                       value line-breaking))
 851       (setq attributes (delq 'hanyu-dazidian-vol attributes))
 852       )
 853     (when (and (memq 'hanyu-dazidian-page attributes)
 854                (setq value (get-char-attribute char 'hanyu-dazidian-page)))
 855       (insert (format "(hanyu-dazidian-page . %d)%s"
 856                       value line-breaking))
 857       (setq attributes (delq 'hanyu-dazidian-page attributes))
 858       )
 859     (when (and (memq 'hanyu-dazidian-char attributes)
 860                (setq value (get-char-attribute char 'hanyu-dazidian-char)))
 861       (insert (format "(hanyu-dazidian-char . %d)%s"
 862                       value line-breaking))
 863       (setq attributes (delq 'hanyu-dazidian-char attributes))
 864       )
 865     (unless readable
 866       (dolist (ignored '(composition
 867                          ->denotational <-subsumptive ->ucs-unified
 868                          ->ideographic-component-forms))
 869         (setq attributes (delq ignored attributes))))
 870     (while attributes
 871       (setq name (car attributes))
 872       (if (setq value (get-char-attribute char name))
 873           (cond ((setq ret (find-charset name))
 874                  (setq name (charset-name ret))
 875                  (if (and (not (memq name dest-ccss))
 876                           (prog1
 877                               (setq value (get-char-attribute char name))
 878                             (setq dest-ccss (cons name dest-ccss))))
 879                      (char-db-insert-ccs-feature name value line-breaking))
 880                  )
 881                 ((string-match "^=>ucs@" (symbol-name name))
 882                  (insert (format "(%-18s . #x%04X)\t; %c%s"
 883                                  name value (decode-char '=ucs value)
 884                                  line-breaking))
 885                  )
 886                 ((eq name 'jisx0208-1978/4X)
 887                  (insert (format "(%-18s . #x%04X)%s"
 888                                  name value
 889                                  line-breaking))
 890                  )
 891                 ((and (not readable)
 892                       (null (get-char-attribute
 893                              char
 894                              (intern (format "%s*sources" name))))
 895                       (not (string-match "\\*sources$" (symbol-name name)))
 896                       (or (eq name '<-identical)
 897                           (string-match "^->halfwidth" (symbol-name name))
 898                           (and
 899                            (string-match "^->fullwidth" (symbol-name name))
 900                            (not
 901                             (and (consp value)
 902                                  (characterp (car value))
 903                                  (encode-char
 904                                   (car value) '=ucs 'defined-only))))
 905                           (string-match "^->simplified" (symbol-name name))
 906                           (string-match "^->vulgar" (symbol-name name))
 907                           (string-match "^->wrong" (symbol-name name))
 908                           (string-match "^->same" (symbol-name name))
 909                           (string-match "^->original" (symbol-name name))
 910                           (string-match "^->ancient" (symbol-name name))
 911                           ))
 912                  )
 913                 ((or (eq name 'ideographic-structure)
 914                      (eq name 'ideographic-)
 915                      (string-match "^\\(->\\|<-\\)" (symbol-name name)))
 916                  (insert (format "(%-18s%s " name line-breaking))
 917                  (setq lbs (concat "\n" (make-string (current-column) ?\ ))
 918                        separator nil)
 919                  (while (consp value)
 920                    (setq cell (car value))
 921                    (if (integerp cell)
 922                        (setq cell (decode-char '=ucs cell)))
 923                    (cond ((eq name '->subsumptive)
 924                           (when (or (not
 925                                      (some (lambda (atr)
 926                                              (get-char-attribute cell atr))
 927                                            char-db-ignored-attributes))
 928                                     (some (lambda (ccs)
 929                                             (encode-char cell ccs
 930                                                          'defined-only))
 931                                           ccss))
 932                             (if separator
 933                                 (insert lbs))
 934                             (let ((char-db-ignored-attributes
 935                                    (cons '<-subsumptive
 936                                          char-db-ignored-attributes)))
 937                               (insert-char-attributes cell readable))
 938                             (setq separator lbs))
 939                           )
 940                          ((characterp cell)
 941                           (setq sources
 942                                 (get-char-attribute
 943                                  char
 944                                  (intern (format "%s*sources" name))))
 945                           (setq required-features nil)
 946                           (dolist (source sources)
 947                             (cond
 948                              ((memq source '(JP JP/Jouyou
 949                                                 shinjigen-1))
 950                               (setq required-features
 951                                     (union required-features
 952                                            '(=jis-x0208
 953                                              =jis-x0208@1990
 954                                              =jis-x0213-1-2000
 955                                              =jis-x0213-2-2000
 956                                              =jis-x0212
 957                                              =jis-x0208@1983
 958                                              =jis-x0208@1978))))
 959                              ((eq source 'CN)
 960                               (setq required-features
 961                                     (union required-features
 962                                            '(=gb2312
 963                                              =gb12345
 964                                              =iso-ir165)))))
 965                             (cond
 966                              ((find-charset
 967                                (setq ret (intern (format "=%s" source))))
 968                               (setq required-features
 969                                     (cons ret required-features)))
 970                              (t (setq required-features
 971                                       (cons source required-features)))))
 972                           (cond ((string-match "@JP" (symbol-name name))
 973                                  (setq required-features
 974                                        (union required-features
 975                                               '(=jis-x0208
 976                                                 =jis-x0208@1990
 977                                                 =jis-x0213-1-2000
 978                                                 =jis-x0213-2-2000
 979                                                 =jis-x0212
 980                                                 =jis-x0208@1983
 981                                                 =jis-x0208@1978))))
 982                                 ((string-match "@CN" (symbol-name name))
 983                                  (setq required-features
 984                                        (union required-features
 985                                               '(=gb2312
 986                                                 =gb12345
 987                                                 =iso-ir165)))))
 988                           (if separator
 989                               (insert lbs))
 990                           (if readable
 991                               (insert (format "%S" cell))
 992                             (char-db-insert-char-spec cell readable
 993                                                       nil
 994                                                       required-features))
 995                           (setq separator lbs))
 996                          ((consp cell)
 997                           (if separator
 998                               (insert lbs))
 999                           (if (consp (car cell))
1000                               (char-db-insert-char-spec cell readable)
1001                             (char-db-insert-char-reference cell readable))
1002                           (setq separator lbs))
1003                          (t
1004                           (if separator
1005                               (insert separator))
1006                           (insert (prin1-to-string cell))
1007                           (setq separator " ")))
1008                    (setq value (cdr value)))
1009                  (insert ")")
1010                  (insert line-breaking))
1011                 ((memq name '(ideograph=
1012                               original-ideograph-of
1013                               ancient-ideograph-of
1014                               vulgar-ideograph-of
1015                               wrong-ideograph-of
1016                               ;; simplified-ideograph-of
1017                               ideographic-variants
1018                               ;; ideographic-different-form-of
1019                               ))
1020                  (insert (format "(%-18s%s " name line-breaking))
1021                  (setq lbs (concat "\n" (make-string (current-column) ?\ ))
1022                        separator nil)
1023                  (while (consp value)
1024                    (setq cell (car value))
1025                    (if (and (consp cell)
1026                             (consp (car cell)))
1027                        (progn
1028                          (if separator
1029                              (insert lbs))
1030                          (char-db-insert-alist cell readable)
1031                          (setq separator lbs))
1032                      (if separator
1033                          (insert separator))
1034                      (insert (prin1-to-string cell))
1035                      (setq separator " "))
1036                    (setq value (cdr value)))
1037                  (insert ")")
1038                  (insert line-breaking))
1039                 ((consp value)
1040                  (insert (format "(%-18s " name))
1041                  (setq lbs (concat "\n" (make-string (current-column) ?\ ))
1042                        separator nil)
1043                  (while (consp value)
1044                    (setq cell (car value))
1045                    (if (and (consp cell)
1046                             (consp (car cell))
1047                             (setq ret (condition-case nil
1048                                           (find-char cell)
1049                                         (error nil))))
1050                        (progn
1051                          (setq rest cell
1052                                al nil
1053                                cal nil)
1054                          (while rest
1055                            (setq key (car (car rest)))
1056                            (if (find-charset key)
1057                                (setq cal (cons key cal))
1058                              (setq al (cons key al)))
1059                            (setq rest (cdr rest)))
1060                          (if separator
1061                              (insert lbs))
1062                          (insert-char-attributes ret
1063                                                  readable
1064                                                  al cal)
1065                          (setq separator lbs))
1066                      (setq ret (prin1-to-string cell))
1067                      (if separator
1068                          (if (< (+ (current-column)
1069                                    (length ret)
1070                                    (length separator))
1071                                 76)
1072                              (insert separator)
1073                            (insert lbs)))
1074                      (insert ret)
1075                      (setq separator " "))
1076                    (setq value (cdr value)))
1077                  (insert ")")
1078                  (insert line-breaking))
1079                 (t
1080                  (insert (format "(%-18s . %S)%s"
1081                                  name value
1082                                  line-breaking)))
1083                 ))
1084       (setq attributes (cdr attributes)))
1085     (insert ")")))
1086
1087 (defun insert-char-data (char &optional readable
1088                               attributes)
1089   (save-restriction
1090     (narrow-to-region (point)(point))
1091     (insert "(define-char
1092   '")
1093     (insert-char-attributes char readable attributes)
1094     (insert ")\n")
1095     (goto-char (point-min))
1096     (while (re-search-forward "[ \t]+$" nil t)
1097       (replace-match ""))
1098     ;; from tabify.
1099     (goto-char (point-min))
1100     (while (re-search-forward "[ \t][ \t][ \t]*" nil t)
1101       (let ((column (current-column))
1102             (indent-tabs-mode t))
1103         (delete-region (match-beginning 0) (point))
1104         (indent-to column)))
1105     (goto-char (point-max))
1106     ;; (tabify (point-min)(point-max))
1107     ))
1108
1109 (defun insert-char-data-with-variant (char &optional printable
1110                                            no-ucs-unified
1111                                            script excluded-script)
1112   (insert-char-data char printable)
1113   (let ((variants (char-variants char))
1114         rest
1115         variant vs ret)
1116     (setq variants (sort variants #'<))
1117     (setq rest variants)
1118     (setq variants (cons char variants))
1119     (while rest
1120       (setq variant (car rest))
1121       (unless (get-char-attribute variant '<-subsumptive)
1122         (if (and (or (null script)
1123                      (null (setq vs (get-char-attribute variant 'script)))
1124                      (memq script vs))
1125                  (or (null excluded-script)
1126                      (null (setq vs (get-char-attribute variant 'script)))
1127                      (not (memq excluded-script vs))))
1128             (unless (and no-ucs-unified (get-char-attribute variant '=ucs))
1129               (insert-char-data variant printable)
1130               (if (setq ret (char-variants variant))
1131                   (while ret
1132                     (or (memq (car ret) variants)
1133                         ;; (get-char-attribute (car ret) '<-subsumptive)
1134                         (setq rest (nconc rest (list (car ret)))))
1135                     (setq ret (cdr ret)))))))
1136       (setq rest (cdr rest)))))
1137
1138 (defun insert-char-range-data (min max &optional script excluded-script)
1139   (let ((code min)
1140         char)
1141     (while (<= code max)
1142       (setq char (decode-char '=ucs code))
1143       (if (encode-char char '=ucs 'defined-only)
1144           (insert-char-data-with-variant char nil 'no-ucs-unified
1145                                          script excluded-script))
1146       (setq code (1+ code)))))
1147
1148 (defun write-char-range-data-to-file (min max file
1149                                           &optional script excluded-script)
1150   (let ((coding-system-for-write char-db-file-coding-system))
1151     (with-temp-buffer
1152       (insert (format ";; -*- coding: %s -*-\n"
1153                       char-db-file-coding-system))
1154       (insert-char-range-data min max script excluded-script)
1155       (write-region (point-min)(point-max) file))))
1156
1157 (defvar what-character-original-window-configuration)
1158
1159 ;;;###autoload
1160 (defun what-char-definition (char)
1161   (interactive (list (char-after)))
1162   (let ((buf (get-buffer-create "*Character Description*"))
1163         (the-buf (current-buffer))
1164         (win-conf (current-window-configuration)))
1165     (pop-to-buffer buf)
1166     (make-local-variable 'what-character-original-window-configuration)
1167     (setq what-character-original-window-configuration win-conf)
1168     (setq buffer-read-only nil)
1169     (erase-buffer)
1170     (condition-case err
1171         (progn
1172           (insert-char-data-with-variant char 'printable)
1173           (unless (char-attribute-alist char)
1174             (insert (format ";; = %c\n"
1175                             (let* ((rest (split-char char))
1176                                    (ccs (pop rest))
1177                                    (code (pop rest)))
1178                               (while rest
1179                                 (setq code (logior (lsh code 8)
1180                                                    (pop rest))))
1181                               (decode-char ccs code)))))
1182           ;; (char-db-update-comment)
1183           (set-buffer-modified-p nil)
1184           (view-mode the-buf (lambda (buf)
1185                                (set-window-configuration
1186                                 what-character-original-window-configuration)
1187                                ))
1188           (goto-char (point-min)))
1189       (error (progn
1190                (set-window-configuration
1191                 what-character-original-window-configuration)
1192                (signal (car err) (cdr err)))))))
1193
1194 (provide 'char-db-util)
1195
1196 ;;; char-db-util.el ends here