(J90-6D40): Unify JIS X 0208:1978 and 1983 code points.
[chise/xemacs-chise.git-] / lisp / utf-2000 / char-db-util.el
index 0bc9803..b0d07b7 100644 (file)
    ((symbolp kb)
     nil)))
 
+(defun char-db-insert-char-spec (char &optional readable column)
+  (unless column
+    (setq column (current-column)))
+  (let (char-spec ret al cal key temp-char)
+    (cond ((characterp char)
+          (cond ((and (setq ret (get-char-attribute char 'ucs))
+                      (not (and (<= #xE000 ret)(<= ret #xF8FF))))
+                 (setq char-spec (list (cons 'ucs ret)))
+                 (if (setq ret (get-char-attribute char 'name))
+                     (setq char-spec (cons (cons 'name ret) char-spec)))
+                 )
+                ((setq ret (split-char char))
+                 (setq char-spec (list ret))
+                 (dolist (ccs (delq (car ret) (charset-list)))
+                   (if (or (and (>= (charset-iso-final-char ccs) ?0)
+                                (setq ret (get-char-attribute char ccs)))
+                           (eq ccs 'ideograph-daikanwa))
+                       (setq char-spec (cons (cons ccs ret) char-spec))))
+                 (if (setq ret (get-char-attribute char 'name))
+                     (setq char-spec (cons (cons 'name ret) char-spec)))
+                 )))
+         ((consp char)
+          (setq char-spec char)
+          (setq char nil)))
+    (unless (or char
+               (condition-case nil
+                   (setq char (find-char char-spec))
+                 (error nil)))
+      ;; define temporary character
+      ;;   Current implementation is dirty.
+      (setq temp-char (define-char (cons '(ideograph-daikanwa . 0)
+                                        char-spec)))
+      (remove-char-attribute temp-char 'ideograph-daikanwa)
+      (setq char temp-char))
+    (setq al nil
+         cal nil)
+    (while char-spec
+      (setq key (car (car char-spec)))
+      (if (find-charset key)
+         (setq cal (cons key cal))
+       (setq al (cons key al)))
+      (setq char-spec (cdr char-spec)))
+    (insert-char-attributes char
+                           readable
+                           (or al 'none) cal)
+    (when temp-char
+      ;; undefine temporary character
+      ;;   Current implementation is dirty.
+      (setq char-spec (char-attribute-alist temp-char))
+      (while char-spec
+       (remove-char-attribute temp-char (car (car char-spec)))
+       (setq char-spec (cdr char-spec))))))
+
 (defun char-db-insert-alist (alist &optional readable column)
   (unless column
     (setq column (current-column)))
       (cond ((eq name 'char)
             (insert "(char . ")
             (if (setq ret (condition-case nil
-                              (define-char value)
+                              (find-char value)
                             (error nil)))
                 (progn
                   (setq al nil
               (if (and (consp cell)
                        (consp (car cell))
                        (setq ret (condition-case nil
-                                     (define-char cell)
+                                     (find-char cell)
                                    (error nil)))
                        )
                   (progn
       (setq alist (cdr alist))))
   (insert ")"))
 
+(defun char-db-insert-char-reference (plist &optional readable column)
+  (unless column
+    (setq column (current-column)))
+  (let ((line-breaking
+        (concat "\n" (make-string (1+ column) ?\ )))
+       name value)
+    (insert "(")
+    (while plist
+      (setq name (pop plist))
+      (setq value (pop plist))
+      (cond ((eq name :char)
+            (insert ":char\t")
+            (cond ((numberp value)
+                   (setq value (decode-char 'ucs value)))
+                   ;; ((consp value)
+                   ;;  (setq value (or (find-char value)
+                   ;;                  value)))
+                  )
+            (char-db-insert-char-spec value readable)
+             (insert line-breaking))
+            (t
+            (insert (format "%s\t%S%s"
+                            name value
+                            line-breaking))))
+      ))
+  (insert ")"))
+
+(defun char-db-decode-isolated-char (ccs code-point)
+  (let (ret)
+    (setq ret
+         (cond ((eq ccs 'arabic-iso8859-6)
+                (decode-char ccs code-point))
+               ((and (memq ccs '(ideograph-gt-pj-1
+                                 ideograph-gt-pj-2
+                                 ideograph-gt-pj-3
+                                 ideograph-gt-pj-4
+                                 ideograph-gt-pj-5
+                                 ideograph-gt-pj-6
+                                 ideograph-gt-pj-7
+                                 ideograph-gt-pj-8
+                                 ideograph-gt-pj-9
+                                 ideograph-gt-pj-10
+                                 ideograph-gt-pj-11))
+                     (setq ret (decode-char ccs code-point))
+                     (setq ret (get-char-attribute ret 'ideograph-gt)))
+                (decode-builtin-char 'ideograph-gt ret))
+               (t
+                (decode-builtin-char ccs code-point))))
+    (cond ((and (<= 0 (char-int ret))
+               (<= (char-int ret) #x1F))
+          (decode-char 'ucs (+ #x2400 (char-int ret))))
+         ((= (char-int ret) #x7F)
+          ?\u2421)
+         (t ret))))
+
+(defvar char-db-convert-obsolete-format t)
+
 (defun insert-char-attributes (char &optional readable
                                    attributes ccs-attributes
                                    column)
     (when (and (memq 'name attributes)
               (setq value (get-char-attribute char 'name)))
       (insert (format
-              (if (> (length value) 47)
+              (if (> (+ (current-column) (length value)) 48)
                   "(name . %S)%s"
-                "(name\t\t. %S)%s")
+                "(name               . %S)%s")
               value line-breaking))
       (setq attributes (delq 'name attributes))
       )
                      line-breaking))
       (setq attributes (delq 'script attributes))
       )
+    (when (and (memq '=>ucs attributes)
+              (setq value (get-char-attribute char '=>ucs)))
+      (insert (format "(=>ucs\t\t. #x%04X)\t; %c%s"
+                     value (decode-char 'ucs value)
+                     line-breaking))
+      (setq attributes (delq '=>ucs attributes))
+      )
     (when (and (memq '->ucs attributes)
               (setq value (get-char-attribute char '->ucs)))
-      (insert (format "(->ucs\t\t. #x%04X)\t; %c%s"
+      (insert (format (if char-db-convert-obsolete-format
+                         "(=>ucs\t\t. #x%04X)\t; %c%s"
+                       "(->ucs\t\t. #x%04X)\t; %c%s")
                      value (decode-char 'ucs value)
                      line-breaking))
       (setq attributes (delq '->ucs attributes))
                      line-breaking))
       (setq attributes (delq '->decomposition attributes))
       )
-    (when (and (memq '->uppercase attributes)
-              (setq value (get-char-attribute char '->uppercase)))
-      (insert (format "(->uppercase\t%s)%s"
-                     (mapconcat (lambda (code)
-                                  (cond ((symbolp code)
-                                         (symbol-name code))
-                                        ((integerp code)
-                                         (format "#x%04X" code))
-                                        (t
-                                         (format "%s %S"
-                                                 line-breaking code))))
-                                value " ")
-                     line-breaking))
-      (setq attributes (delq '->uppercase attributes))
-      )
-    (when (and (memq '->lowercase attributes)
-              (setq value (get-char-attribute char '->lowercase)))
-      (insert (format "(->lowercase\t%s)%s"
-                     (mapconcat (lambda (code)
-                                  (cond ((symbolp code)
-                                         (symbol-name code))
-                                        ((integerp code)
-                                         (format "#x%04X" code))
-                                        (t
-                                         (format "%s %S"
-                                                 line-breaking code))))
-                                value " ")
-                     line-breaking))
-      (setq attributes (delq '->lowercase attributes))
-      )
-    (when (and (memq '->titlecase attributes)
-              (setq value (get-char-attribute char '->titlecase)))
-      (insert (format "(->titlecase\t%s)%s"
-                     (mapconcat (lambda (code)
-                                  (cond ((symbolp code)
-                                         (symbol-name code))
-                                        ((integerp code)
-                                         (format "#x%04X" code))
-                                        (t
-                                         (format "%s %S"
-                                                 line-breaking code))))
-                                value " ")
-                     line-breaking))
-      (setq attributes (delq '->titlecase attributes))
-      )
+    (if (equal (get-char-attribute char '->titlecase)
+              (get-char-attribute char '->uppercase))
+       (setq attributes (delq '->titlecase attributes)))
     (when (and (memq '->mojikyo attributes)
               (setq value (get-char-attribute char '->mojikyo)))
       (insert (format "(->mojikyo\t\t. %06d)\t; %c%s"
                 (insert (format "(%-18s . #x%04X)%s"
                                 name value
                                 line-breaking)))
-               ((string-match "^->" (symbol-name name))
-                (insert
-                 (format "(%-18s %s)%s"
-                         name
-                         (mapconcat (lambda (code)
-                                      (cond ((symbolp code)
-                                             (symbol-name code))
-                                            ((integerp code)
-                                             (format "#x%04X" code))
-                                            (t
-                                             (format "%s%S"
-                                                     line-breaking code))))
-                                    value " ")
-                         line-breaking)))
+               ((memq name '(->lowercase
+                             ->uppercase ->titlecase
+                             ->fullwidth <-fullwidth
+                             ->vulgar-ideograph <-vulgar-ideograph
+                             ->ancient-ideograph <-ancient-ideograph
+                             ->original-ideograph <-original-ideograph
+                             ->simplified-ideograph <-simplified-ideograph
+                             ->wrong-ideograph <-wrong-ideograph
+                             ->same-ideograph
+                             ->ideographic-variants
+                             ->synonyms
+                             ->radical <-radical
+                             ->bopomofo <-bopomofo
+                             ->ideographic <-ideographic
+                             ideographic-structure))
+                (insert (format "(%-18s%s " name line-breaking))
+                (setq lbs (concat "\n" (make-string (current-column) ?\ ))
+                      separator nil)
+                (while (consp value)
+                  (setq cell (car value))
+                   (if (integerp cell)
+                      (setq cell (decode-char 'ucs cell)))
+                  (cond ((characterp cell)
+                         (if separator
+                             (insert lbs))
+                         (char-db-insert-char-spec cell readable)
+                         (setq separator lbs))
+                        ((consp cell)
+                         (if separator
+                             (insert lbs))
+                         (if (consp (car cell))
+                             (char-db-insert-char-spec cell readable)
+                           (char-db-insert-char-reference cell readable))
+                         (setq separator lbs))
+                        (t
+                         (if separator
+                             (insert separator))
+                         (insert (prin1-to-string cell))
+                         (setq separator " ")))
+                  (setq value (cdr value)))
+                (insert ")")
+                (insert line-breaking))
                ((memq name '(ideograph=
                              original-ideograph-of
                              ancient-ideograph-of
                   (setq value (cdr value)))
                 (insert ")")
                 (insert line-breaking))
+               ((string-match "^->" (symbol-name name))
+                (insert
+                 (format "(%-18s %s)%s"
+                         name
+                         (mapconcat (lambda (code)
+                                      (cond ((symbolp code)
+                                             (symbol-name code))
+                                            ((integerp code)
+                                             (format "#x%04X" code))
+                                            (t
+                                             (format "%s%S"
+                                                     line-breaking code))))
+                                    value " ")
+                         line-breaking)))
                ((consp value)
                 (insert (format "(%-18s " name))
                 (setq lbs (concat "\n" (make-string (current-column) ?\ ))
                   (if (and (consp cell)
                            (consp (car cell))
                            (setq ret (condition-case nil
-                                         (define-char cell)
+                                         (find-char cell)
                                        (error nil))))
                       (progn
                         (setq rest cell
                   (if has-long-ccs-name
                       "(%-26s . %06d)\t; %c%s"
                     "(%-18s . %06d)\t; %c%s"))
+                 ((eq name 'ucs)
+                  (if has-long-ccs-name
+                      "(%-26s . #x%04X)\t; %c%s"
+                    "(%-18s . #x%04X)\t; %c%s"))
                  (t
                   (if has-long-ccs-name
-                      "(%-26s . #x%X)\t; %c%s"
-                    "(%-18s . #x%X)\t; %c%s")))
+                      "(%-26s . #x%02X)\t; %c%s"
+                    "(%-18s . #x%02X)\t; %c%s")))
            name
            (if (= (charset-iso-graphic-plane name) 1)
                (logior value
                               #x808080)
                              (t 0)))
              value)
-           (if (and (memq name '(ideograph-gt-pj-1
-                                 ideograph-gt-pj-2
-                                 ideograph-gt-pj-3
-                                 ideograph-gt-pj-4
-                                 ideograph-gt-pj-5
-                                 ideograph-gt-pj-6
-                                 ideograph-gt-pj-7
-                                 ideograph-gt-pj-8
-                                 ideograph-gt-pj-9
-                                 ideograph-gt-pj-10
-                                 ideograph-gt-pj-11))
-                    (setq ret (decode-char name value))
-                    (setq ret (get-char-attribute ret 'ideograph-gt)))
-               (decode-builtin-char 'ideograph-gt ret)
-             (decode-builtin-char name value))
+            (char-db-decode-isolated-char name value)
            line-breaking)))
       (setq ccs-attributes (cdr ccs-attributes)))
     (insert ")")))
                        (if ucs
                            (delete char (char-variants (int-char ucs)))))))
        variant vs)
+    (setq variants (sort variants #'<))
     (while variants
       (setq variant (car variants))
       (if (and (or (null script)