Reformatted.
[chise/xemacs-chise.git.1] / lisp / utf-2000 / char-db-util.el
index 5aaa6e2..84e2313 100644 (file)
@@ -69,6 +69,8 @@
       (setq i (1+ i)))
     v))
 
+(defvar char-db-file-coding-system 'utf-8-mcs-er)
+
 (defvar char-db-feature-domains
   '(ucs daikanwa cns gt jis jis/alt jis/a jis/b
        jis-x0213 misc unknown))
 
 (defun char-attribute-name< (ka kb)
   (cond
-   ((find-charset ka)
-    (cond
-     ((find-charset kb)
-      (if (<= (charset-id ka) 0)
-         (if (<= (charset-id kb) 0)
-             (cond
-              ((= (charset-dimension ka)
-                  (charset-dimension kb))
-               (> (charset-id ka)(charset-id kb)))
-              (t
-               (> (charset-dimension ka)
-                  (charset-dimension kb))
-               ))
-           t)
-       (if (<= (charset-id kb) 0)
-           nil
-         (< (charset-id ka)(charset-id kb)))))
-     ((symbolp kb)
-      nil)
-     (t
-      t)))
    ((eq '->denotational kb)
     t)
    ((eq '->subsumptive kb)
     nil)
    ((eq '->subsumptive ka)
     nil)
+   ((find-charset ka)
+    (if (find-charset kb)
+       (if (<= (charset-id ka) 1)
+           (if (<= (charset-id kb) 1)
+               (cond
+                ((= (charset-dimension ka)
+                    (charset-dimension kb))
+                 (> (charset-id ka)(charset-id kb)))
+                (t
+                 (> (charset-dimension ka)
+                    (charset-dimension kb))
+                 ))
+             t)
+         (if (<= (charset-id kb) 1)
+             nil
+           (< (charset-id ka)(charset-id kb))))
+      nil))
    ((find-charset kb)
     t)
    ((symbolp ka)
                                      required-features)
   (unless column
     (setq column (current-column)))
-  (let (char-spec al cal key temp-char)
+  (let (char-spec temp-char)
     (setq char-spec (char-db-make-char-spec char))
     (unless (or (characterp char) ; char
                (condition-case nil
                                         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)))
-    ;;   (unless (memq key char-db-ignored-attributes)
-    ;;     (if (find-charset key)
-    ;;         (if (encode-char char key 'defined-only)
-    ;;             (setq cal (cons key cal)))
-    ;;       (setq al (cons key al))))
-    ;;   (setq char-spec (cdr char-spec)))
-    ;; (unless cal
-    ;;   (setq char-spec (char-db-make-char-spec char))
-    ;;   (while char-spec
-    ;;     (setq key (car (car char-spec)))
-    ;;     (unless (memq key char-db-ignored-attributes)
-    ;;       (if (find-charset key)
-    ;;           (setq cal (cons key cal))
-    ;;         (setq al (cons key al))))
-    ;;     (setq char-spec (cdr char-spec)))
-    ;;   )
-    ;; (unless (or cal
-    ;;             (memq 'ideographic-structure al))
-    ;;   (push 'ideographic-structure al))
-    ;; (dolist (feature required-features)
-    ;;   (if (find-charset feature)
-    ;;       (if (encode-char char feature 'defined-only)
-    ;;           (setq cal (adjoin feature cal)))
-    ;;     (setq al (adjoin feature al))))
     (insert-char-attributes char
                            readable
-                            ;; (or al 'none) cal
-                           (union (mapcar #'car char-spec)
-                                  required-features)
-                           )
+                            (union (mapcar #'car char-spec)
+                                  required-features))
     (when temp-char
       ;; undefine temporary character
       ;;   Current implementation is dirty.
                             (error nil)))
                 (progn
                   (setq al nil
-                        cal nil)
+                        ;; cal nil
+                        )
                   (while value
                     (setq key (car (car value)))
                      ;; (if (find-charset key)
                   (progn
                     (setq rest cell
                           al nil
-                          cal nil)
+                          ;; cal nil
+                          )
                     (while rest
                       (setq key (car (car rest)))
                        ;; (if (find-charset key)
 (defvar char-db-convert-obsolete-format t)
 
 (defun insert-char-attributes (char &optional readable attributes column)
-  (let (atr-d)
-    (setq attributes
-         (sort (if attributes
-                   (if (consp attributes)
-                       (progn
-                         (dolist (name attributes)
-                           (unless (memq name char-db-ignored-attributes)
-                             (push name atr-d)))
-                         atr-d))
-                 (dolist (name (char-attribute-list))
-                   (unless (memq name char-db-ignored-attributes)
-                     (push name atr-d)))
-                 atr-d)
-               #'char-attribute-name<)))
   (unless column
     (setq column (current-column)))
   (let (name value has-long-ccs-name rest
        lbs cell separator ret
        key al cal
        dest-ccss
-       sources required-features)
+       sources required-features
+       ccss)
+    (let (atr-d)
+      (setq attributes
+           (sort (if attributes
+                     (if (consp attributes)
+                         (progn
+                           (dolist (name attributes)
+                             (unless (memq name char-db-ignored-attributes)
+                               (if (find-charset name)
+                                   (push name ccss))
+                               (push name atr-d)))
+                           atr-d))
+                   (dolist (name (char-attribute-list))
+                     (unless (memq name char-db-ignored-attributes)
+                       (if (find-charset name)
+                           (push name ccss))
+                       (push name atr-d)))
+                   atr-d)
+                 #'char-attribute-name<)))
     (insert "(")
     (when (and (memq 'name attributes)
               (setq value (get-char-attribute char 'name)))
                      line-breaking))
       (setq attributes (delq 'script attributes))
       )
+    ;; (when (and (memq '<-denotational attributes)
+    ;;            (setq value (get-char-attribute char '<-denotational))
+    ;;            (null (cdr value))
+    ;;            (setq value (encode-char (car value) 'ucs 'defined-only)))
+    ;;   (insert (format "(%-18s . #x%04X)\t; %c%s"
+    ;;                   '=>ucs value (decode-char 'ucs value)
+    ;;                   line-breaking))
+    ;;   (setq attributes (delq '<-denotational attributes)))
     (dolist (name '(=>ucs =>ucs*))
       (when (and (memq name attributes)
                 (setq value (get-char-attribute char name)))
                     line-breaking)
           (format "(%-18s %s)\t; %c%s"
                   name
-                  (mapconcat #'prin1-to-string
+                  (mapconcat (function prin1-to-string)
                              value " ")
                   (char-representative-of-daikanwa char)
                   line-breaking)))
       )
     (unless readable
       (dolist (ignored '(composition
-                        ->denotational <-subsumptive ->ucs-unified))
+                        ->denotational <-subsumptive ->ucs-unified
+                        ->ideographic-component-forms
+                        <-same))
        (setq attributes (delq ignored attributes))))
     ;; (setq rest ccs-attributes)
     ;; (while (and rest
                                 line-breaking))
                 )
                ((and (not readable)
-                     (string-match "^->simplified" (symbol-name name)))
+                     (null (get-char-attribute
+                            char
+                            (intern (format "%s*sources" name))))
+                     (not (string-match "\\*sources$" (symbol-name name)))
+                     (or (eq name '<-identical)
+                         (string-match "^->simplified" (symbol-name name))
+                         (string-match "^->vulgar" (symbol-name name))
+                         (string-match "^->wrong" (symbol-name name))
+                         ))
                 )
                ((or (eq name 'ideographic-structure)
                     (eq name 'ideographic-)
                    (if (integerp cell)
                       (setq cell (decode-char '=ucs cell)))
                   (cond ((eq name '->subsumptive)
-                         (if separator
-                             (insert lbs))
-                         (let ((char-db-ignored-attributes
-                                (cons '<-subsumptive
-                                      char-db-ignored-attributes)))
-                           (insert-char-attributes cell readable))
-                         (setq separator lbs))
+                         (when (or (not
+                                    (some (lambda (atr)
+                                            (get-char-attribute cell atr))
+                                          char-db-ignored-attributes))
+                                   (some (lambda (ccs)
+                                           (encode-char cell ccs
+                                                        'defined-only))
+                                         ccss))
+                           (if separator
+                               (insert lbs))
+                           (let ((char-db-ignored-attributes
+                                  (cons '<-subsumptive
+                                        char-db-ignored-attributes)))
+                             (insert-char-attributes cell readable))
+                           (setq separator lbs))
+                         )
                         ((characterp cell)
                          (setq sources
                                (get-char-attribute
                                 (intern (format "%s*sources" name))))
                          (setq required-features nil)
                          (dolist (source sources)
-                           (setq required-features
-                                 (cons
-                                  (if (find-charset
-                                       (setq ret (intern
-                                                  (format "=%s" source))))
-                                      ret
-                                    source)
-                                  required-features)))
-                         (when (string-match "@JP" (symbol-name name))
-                           (setq required-features
-                                 (union required-features
-                                        '(=jis-x0208
-                                          =jis-x0208-1990
-                                          =jis-x0213-1-2000
-                                          =jis-x0213-2-2000
-                                          =jis-x0212
-                                          =jis-x0208-1983
-                                          =jis-x0208-1978))))
+                           (cond
+                            ((memq source '(JP JP/Jouyou
+                                               shinjigen-1))
+                             (setq required-features
+                                   (union required-features
+                                          '(=jis-x0208
+                                            =jis-x0208-1990
+                                            =jis-x0213-1-2000
+                                            =jis-x0213-2-2000
+                                            =jis-x0212
+                                            =jis-x0208-1983
+                                            =jis-x0208-1978))))
+                            ((eq source 'CN)
+                             (setq required-features
+                                   (union required-features
+                                          '(=gb2312
+                                            =gb12345
+                                            =iso-ir165)))))
+                           (cond
+                            ((find-charset
+                              (setq ret (intern (format "=%s" source))))
+                             (setq required-features
+                                   (cons ret required-features)))
+                            (t (setq required-features
+                                     (cons source required-features)))))
+                         (cond ((string-match "@JP" (symbol-name name))
+                                (setq required-features
+                                      (union required-features
+                                             '(=jis-x0208
+                                               =jis-x0208-1990
+                                               =jis-x0213-1-2000
+                                               =jis-x0213-2-2000
+                                               =jis-x0212
+                                               =jis-x0208-1983
+                                               =jis-x0208-1978))))
+                               ((string-match "@CN" (symbol-name name))
+                                (setq required-features
+                                      (union required-features
+                                             '(=gb2312
+                                               =gb12345
+                                               =iso-ir165)))))
                          (if separator
                              (insert lbs))
                          (if readable
 
 (defun write-char-range-data-to-file (min max file
                                          &optional script excluded-script)
-  (let ((coding-system-for-write 'utf-8-mcs))
+  (let ((coding-system-for-write char-db-file-coding-system))
     (with-temp-buffer
-      (insert ";; -*- coding: utf-8-mcs -*-\n")
+      (insert (format ";; -*- coding: %s -*-\n"
+                     char-db-file-coding-system))
       (insert-char-range-data min max script excluded-script)
       (write-region (point-min)(point-max) file))))