(decode-builtin-char): Use `charset-name'.
[chise/xemacs-chise.git-] / lisp / utf-2000 / char-db-util.el
index dc740d2..acbc639 100644 (file)
     ("So" symbol       other)
     ))
 
+(defconst ideographic-radicals
+  (let ((v (make-vector 215 nil))
+       (i 1))
+    (while (< i 215)
+      (aset v i (int-char (+ #x2EFF i)))
+      (setq i (1+ i)))
+    (if (< (charset-iso-final-char (car (split-char (aref v 34)))) ?0)
+       (aset v 34 (make-char 'chinese-gb2312 #x62 #x3A)))
+    v))
+
+(defun char-attribute-name< (ka kb)
+  (cond
+   ((find-charset ka)
+    (cond
+     ((find-charset kb)
+      (cond
+       ((= (charset-dimension ka)
+          (charset-dimension kb))
+       (cond ((= (charset-chars ka)(charset-chars kb))
+              (cond
+               ((>= (charset-iso-final-char ka) ?@)
+                (if (>= (charset-iso-final-char kb) ?@)
+                    (< (charset-iso-final-char ka)
+                       (charset-iso-final-char kb))
+                  t))
+               ((>= (charset-iso-final-char ka) ?0)
+                (cond
+                 ((>= (charset-iso-final-char kb) ?@)
+                  nil)
+                 ((>= (charset-iso-final-char kb) ?0)
+                  (< (charset-iso-final-char ka)
+                     (charset-iso-final-char kb)))
+                 (t)))))
+             ((<= (charset-chars ka)(charset-chars kb)))))
+       (t
+       (< (charset-dimension ka)
+          (charset-dimension kb))
+       )))
+     ((symbolp kb)
+      nil)
+     (t
+      t)))
+   ((find-charset kb)
+    t)
+   ((symbolp ka)
+    (cond ((symbolp kb)
+          (string< (symbol-name ka)
+                   (symbol-name kb)))
+         (t)))
+   ((symbolp kb)
+    nil)))
+
 (defun insert-char-data (char)
   (let ((data (char-attribute-alist char))
-       cell ret name has-long-ccs-name rest)
+       cell ret has-long-ccs-name rest
+       radical strokes)
     (when data
       (save-restriction
        (narrow-to-region (point)(point))
          )
        (when (setq cell (assq '->ucs data))
          (setq cell (cdr cell))
-         (insert (format "(->ucs\t\t. #x%04X)
+         (insert (format "(->ucs\t\t. #x%04X)\t; %c
     "
-                         cell))
+                         cell (decode-char 'ucs cell)))
          (setq data (del-alist '->ucs data))
          )
        (when (setq cell (assq 'general-category data))
                          cell))
          (setq data (del-alist 'mirrored data))
          )
-       (when (setq cell (assq 'decimal-digit-value data))
+       (cond
+        ((setq cell (assq 'decimal-digit-value data))
          (setq cell (cdr cell))
          (insert (format "(decimal-digit-value . %S)
     "
            (setq data (del-alist 'numeric-value data))
            )
          )
-       (setq data (sort data
-                        (lambda (a b)
-                          (let ((ka (car a))
-                                (kb (car b)))
-                            (cond ((find-charset ka)
-                                   (cond ((find-charset kb)
-                                          (cond ((= (charset-dimension ka)
-                                                    (charset-dimension kb))
-                                                 (< (charset-final ka)
-                                                    (charset-final kb)))
-                                                (t
-                                                 (< (charset-dimension ka)
-                                                    (charset-dimension kb))
-                                                 )))
-                                         (t)))
-                                  ((find-charset kb)
-                                   t)
-                                  ((symbolp ka)
-                                   (cond ((symbolp kb)
-                                          (string< (symbol-name ka)
-                                                   (symbol-name kb)))
-                                         (t)))
-                                  ((symbolp kb)
-                                   nil))))))
+        (t
+         (when (setq cell (assq 'digit-value data))
+           (setq cell (cdr cell))
+           (insert (format "(digit-value\t. %S)
+    "
+                           cell))
+           (setq data (del-alist 'digit-value data))
+           )
+         (when (setq cell (assq 'numeric-value data))
+           (setq cell (cdr cell))
+           (insert (format "(numeric-value\t. %S)
+    "
+                           cell))
+           (setq data (del-alist 'numeric-value data))
+           )))
+       (when (setq cell (assq 'iso-10646-comment data))
+         (setq cell (cdr cell))
+         (insert (format "(iso-10646-comment\t. %S)
+    "
+                         cell))
+         (setq data (del-alist 'iso-10646-comment data))
+         )
+       (when (setq cell (assq 'morohashi-daikanwa data))
+         (setq cell (cdr cell))
+         (insert (format "(morohashi-daikanwa\t%s)
+    "
+                         (mapconcat (function prin1-to-string) cell " ")))
+         (setq data (del-alist 'morohashi-daikanwa data))
+         )
+       (setq radical nil)
+       (when (setq cell (assq 'ideographic-radical data))
+         (setq radical (cdr cell))
+         (insert (format "(ideographic-radical . %S)\t; %c
+    "
+                         radical
+                         (aref ideographic-radicals radical)))
+         (setq data (del-alist 'ideographic-radical data))
+         )
+       (when (setq cell (assq 'kangxi-radical data))
+         (setq cell (cdr cell))
+         (unless (eq cell radical)
+           (insert (format "(kangxi-radical\t . %S)\t; %c
+    "
+                           cell
+                           (aref ideographic-radicals cell)))
+           (setq radical cell))
+         (setq data (del-alist 'kangxi-radical data))
+         )
+       (when (setq cell (assq 'japanese-radical data))
+         (setq cell (cdr cell))
+         (unless (eq cell radical)
+           (insert (format "(japanese-radical . %S)\t; %c
+    "
+                           cell
+                           (aref ideographic-radicals cell)))
+           (setq radical cell))
+         (setq data (del-alist 'japanese-radical data))
+         )
+       (when (setq cell (assq 'cns-radical data))
+         (setq cell (cdr cell))
+         (insert (format "(cns-radical\t . %S)\t; %c
+    "
+                         cell
+                         (aref ideographic-radicals cell)))
+         (setq data (del-alist 'cns-radical data))
+         )
+       (setq strokes nil)
+       (cond
+        ((setq cell (assq 'ideographic-strokes data))
+         (setq strokes (cdr cell))
+         (insert (format "(ideographic-strokes . %S)
+    "
+                         strokes))
+         (setq data (del-alist 'ideographic-strokes data))
+         (when (setq cell (assq 'kangxi-strokes data))
+           (setq cell (cdr cell))
+           (unless (eq cell strokes)
+             (insert (format "(kangxi-strokes\t . %S)
+    "
+                             cell))
+             (setq strokes cell))
+           (setq data (del-alist 'kangxi-strokes data))
+           )
+         (when (setq cell (assq 'japanese-strokes data))
+           (setq cell (cdr cell))
+           (unless (eq cell strokes)
+             (insert (format "(japanese-strokes\t . %S)
+    "
+                             cell))
+             (setq strokes cell))
+           (setq data (del-alist 'japanese-strokes data))
+           )
+         (when (setq cell (assq 'total-strokes data))
+           (setq cell (cdr cell))
+           (insert (format "(total-strokes\t . %S)
+    "
+                           cell))
+           (setq data (del-alist 'total-strokes data))
+           )
+         )
+        ((setq cell (assq 'total-strokes data))
+         (setq cell (cdr cell))
+         (insert (format "(total-strokes\t. %S)
+    "
+                         cell))
+         (setq data (del-alist 'total-strokes data))
+         ))
+       (when (setq cell (assq '->decomposition data))
+         (setq cell (cdr cell))
+         (insert (format "(->decomposition\t%s)
+    "
+                         (mapconcat (lambda (code)
+                                      (cond ((symbolp code)
+                                             (symbol-name code))
+                                            ((integerp code)
+                                             (format "#x%04X" code))
+                                            (t
+                                             (format "\n     %S" code))))
+                                    cell " ")))
+         (setq data (del-alist '->decomposition data))
+         )
+       (when (setq cell (assq '->uppercase data))
+         (setq cell (cdr cell))
+         (insert (format "(->uppercase\t%s)
+    "
+                         (mapconcat (lambda (code)
+                                      (cond ((symbolp code)
+                                             (symbol-name code))
+                                            ((integerp code)
+                                             (format "#x%04X" code))
+                                            (t
+                                             (format "\n     %S" code))))
+                                    cell " ")))
+         (setq data (del-alist '->uppercase data))
+         )
+       (when (setq cell (assq '->lowercase data))
+         (setq cell (cdr cell))
+         (insert (format "(->lowercase\t%s)
+    "
+                         (mapconcat (lambda (code)
+                                      (cond ((symbolp code)
+                                             (symbol-name code))
+                                            ((integerp code)
+                                             (format "#x%04X" code))
+                                            (t
+                                             (format "\n     %S" code))))
+                                    cell " ")))
+         (setq data (del-alist '->lowercase data))
+         )
+       (when (setq cell (assq '->titlecase data))
+         (setq cell (cdr cell))
+         (insert (format "(->titlecase\t%s)
+    "
+                         (mapconcat (lambda (code)
+                                      (cond ((symbolp code)
+                                             (symbol-name code))
+                                            ((integerp code)
+                                             (format "#x%04X" code))
+                                            (t
+                                             (format "\n     %S" code))))
+                                    cell " ")))
+         (setq data (del-alist '->titlecase data))
+         )
+       (setq data
+             (sort data
+                   (lambda (a b)
+                     (char-attribute-name< (car a)(car b)))))
        (setq rest data)
        (while (and rest
                    (progn
        (while data
          (setq cell (car data))
          (cond ((setq ret (find-charset (car cell)))
-                (insert (format (if has-long-ccs-name
-                                    "(%-26s %s)
+                (insert
+                 (format
+                  (if has-long-ccs-name
+                      (if (eq ret (find-charset 'ideograph-daikanwa))
+                          "(%-26s . %05d)\t; %c
     "
-                                  "(%-18s %s)
+                        "(%-26s . #x%X)\t; %c
     "
-                                  )
-                                (charset-name ret)
-                                (mapconcat (lambda (b)
-                                             (format "#x%02X" b)
-                                             )
-                                           (cdr cell) " "))))
+                        )
+                    (if (eq ret (find-charset 'ideograph-daikanwa))
+                        "(%-18s . %05d)\t; %c
+    "
+                      "(%-18s . #x%X)\t; %c
+    "
+                      ))
+                  (charset-name ret)
+                  (if (= (charset-iso-graphic-plane ret) 1)
+                      (logior (cdr cell)
+                              (cond ((= (charset-dimension ret) 1)
+                                     #x80)
+                                    ((= (charset-dimension ret) 2)
+                                     #x8080)
+                                    ((= (charset-dimension ret) 3)
+                                     #x808080)
+                                    (t 0)))
+                    (cdr cell))
+                  (decode-builtin-char ret (cdr cell)))))
                ((string-match "^->" (symbol-name (car cell)))
                 (insert
                  (format "(%-18s %s)
                                              (format "\n     %S" code))))
                                     (cdr cell) " "))))
                ((consp (cdr cell))
-                (insert (format "%S
+                (insert (format "(%-18s %s)
     "
-                                cell)))
+                                (car cell)
+                                (mapconcat (function prin1-to-string)
+                                           (cdr cell) " "))))
                (t
                 (insert (format "(%-18s . %S)
     "
        (tabify (point-min)(point-max))
        ))))
 
-(defun insert-char-range-data (min max)
-  (let ((code min))
-    (while (<= code max)
-      (insert-char-data (int-char code))
-      (setq code (1+ code))
-      )))
-
-(defun write-char-range-data-to-file (min max file)
-  (with-temp-buffer
-    (insert-char-range-data min max)
-    (write-region (point-min)(point-max) file)))
+(defun decode-builtin-char (charset code-point)
+  (setq charset (get-charset charset))
+  (if (and (not (eq (charset-name charset) 'ideograph-daikanwa))
+          (or (memq (charset-name charset)
+                    '(ascii latin-viscii-upper
+                            latin-viscii-lower
+                            arabic-iso8859-6
+                            japanese-jisx0213-1
+                            japanese-jisx0213-2))
+              (= (char-int (charset-iso-final-char charset)) 0)))
+      (decode-char charset code-point)
+    (let ((table (charset-mapping-table charset)))
+      (if table
+         (prog2
+             (set-charset-mapping-table charset nil)
+             (decode-char charset code-point)
+           (set-charset-mapping-table charset table))
+       (decode-char charset code-point)))))
 
+;;;###autoload
 (defun char-db-update-comment ()
   (interactive)
   (save-excursion
     (goto-char (point-min))
-    (let (cdef table)
+    (let (cdef table char)
       (while (re-search-forward "^[ \t]*\\(([^.()]+)\\)" nil t)
        (goto-char (match-beginning 1))
        (setq cdef (read (current-buffer)))
        (when (find-charset (car cdef))
          (goto-char (match-end 0))
-         (if (setq table (charset-mapping-table (car cdef)))
-             (set-charset-mapping-table (car cdef) nil))
-         (delete-region (point) (point-at-eol))
-         (insert (format "\t; %c" (apply #'make-char cdef)))
-         (if table
-             (set-charset-mapping-table (car cdef) table))
+         (setq char
+               (if (and
+                    (not (eq (car cdef) 'ideograph-daikanwa))
+                    (or (memq (car cdef) '(ascii latin-viscii-upper
+                                                 latin-viscii-lower
+                                                 arabic-iso8859-6
+                                                 japanese-jisx0213-1
+                                                 japanese-jisx0213-2))
+                        (= (char-int (charset-iso-final-char (car cdef)))
+                           0)))
+                   (apply (function make-char) cdef)
+                 (if (setq table (charset-mapping-table (car cdef)))
+                     (set-charset-mapping-table (car cdef) nil))
+                 (prog1
+                     (apply (function make-char) cdef)
+                   (if table
+                       (set-charset-mapping-table (car cdef) table)))))
+         (when (not (or (< (char-int char) 32)
+                        (and (<= 128 (char-int char))
+                             (< (char-int char) 160))))
+           (delete-region (point) (point-at-eol))
+           (insert (format "\t; %c" char)))
          )))))
 
+(defun insert-char-data-with-variant (char)
+  (insert-char-data char)
+  (let ((variants (or (char-variants char)
+                     (let ((ucs (get-char-attribute char '->ucs)))
+                       (if ucs
+                           (delete char (char-variants (int-char ucs))))))))
+    (while variants
+      (insert-char-data (car variants))
+      (setq variants (cdr variants))
+      )))
+
+(defun insert-char-range-data (min max)
+  (let ((code min)
+       char)
+    (while (<= code max)
+      (setq char (int-char code))
+      (insert-char-data-with-variant char)
+      (setq code (1+ code))
+      )))
+
+(defun write-char-range-data-to-file (min max file)
+  (let ((coding-system-for-write 'utf-8))
+    (with-temp-buffer
+      (insert-char-range-data min max)
+      (write-region (point-min)(point-max) file))))
+
+(defvar what-character-original-window-configuration)
+
 ;;;###autoload
 (defun what-char-definition (char)
   (interactive (list (char-after)))
     (erase-buffer)
     (condition-case err
        (progn
-         (insert-char-data char)
+         (insert-char-data-with-variant char)
+          ;; (char-db-update-comment)
          (set-buffer-modified-p nil)
          (view-mode the-buf (lambda (buf)
                               (set-window-configuration