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