(diacritic-composition-pattern): New constant.

(diacritic-compose-region, diacritic-compose-string)
(diacritic-compose-buffer, diacritic-post-read-conversion)
(diacritic-composition-function): New functions.

diff --git a/lisp/ChangeLog b/lisp/ChangeLog
index b950cad7d912839d3727a070e4712e3a3eadebc7..45ee9b4a87a794b44291821cb5048d312d2b1498 100644 (file)
--- a/lisp/ChangeLog
+++ b/lisp/ChangeLog
@@ -1,3 +1,30 @@
+2001-12-07  Dave Love  <fx@gnu.org>
+
+       * language/european.el (diacritic-composition-pattern): New constant.
+       (diacritic-compose-region, diacritic-compose-string) 
+       (diacritic-compose-buffer, diacritic-post-read-conversion) 
+       (diacritic-composition-function): New functions.
+
+       * international/utf-8.el (ucs-mule-to-mule-unicode): New
+       translation table.
+       (ccl-encode-mule-utf-8): Use it.
+       (utf-8-untranslated-to-ucs, utf-8-help-echo, utf-8-compose)
+       (utf-8-post-read-conversion, utf-8-pre-write-conversion): New
+       function.
+       (utf-8-subst-table): New variable.
+       (utf-8-compose-scripts): New option.
+       (mule-utf-8): Update safe-charsets, pre-write and post-read
+       conversion.
+
+       * international/ucs-tables.el, international/utf-8-subst.el: New
+       file.
+
+       * international/characters.el: Don't set word syntax (the default)
+       explicitly.  Add a diacritic category.  Add info for Unicode
+       equivalents of characters in various Mule charsets and for extra
+       Unicode characters.  Don't define specific categories for
+       Indian/Devanagari, since they aren't used.
+
 2001-12-06  Richard M. Stallman  <rms@gnu.org>
 
        * textmodes/fill.el (set-justification): Rename arg VALUE to STYLE.

diff --git a/lisp/language/european.el b/lisp/language/european.el
index 1fd230521ecb1a88e299de03aa3c332e2eeee1b6..4020339376f5ed4d5a78c2e42b535eeccba7cd99 100644 (file)
--- a/lisp/language/european.el
+++ b/lisp/language/european.el
@@ -542,6 +542,66 @@ but select's the Dutch tutorial."))
    (valid-codes (0 . 255))
    (mime-charset . macintosh)))                ; per IANA, rfc1345
 
+(defconst diacritic-composition-pattern "\\C^\\c^+")
+
+;;;###autoload
+(defun diacritic-compose-region (beg end)
+  "Compose diacritic characters in the region.
+When called from a program, expects two arguments,
+positions (integers or markers) specifying the region."
+  (interactive "r")
+  (save-restriction
+    (narrow-to-region beg end)
+    (goto-char (point-min))
+    (while (re-search-forward diacritic-composition-pattern nil t)
+      (compose-region (match-beginning 0) (match-end 0)))))
+
+;;;###autoload
+(defun diacritic-compose-string (string)
+  "Compose diacritic characters in STRING and return the resulting string."
+  (let ((idx 0))
+    (while (setq idx (string-match diacritic-composition-pattern string idx))
+      (compose-string string idx (match-end 0))
+      (setq idx (match-end 0))))
+  string)
+      
+;;;###autoload
+(defun diacritic-compose-buffer ()
+  "Compose diacritic characters in the current buffer."
+  (interactive)
+  (diacritic-compose-region (point-min) (point-max)))
+
+;;;###autoload
+(defun diacritic-post-read-conversion (len)
+  (diacritic-compose-region (point) (+ (point) len))
+  len)
+
+;;;###autoload
+(defun diacritic-composition-function (from to pattern &optional string)
+  "Compose diacritic text in the region FROM and TO.
+The text matches the regular expression PATTERN.
+Optional 4th argument STRING, if non-nil, is a string containing text
+to compose.
+
+The return value is number of composed characters."
+  (if (< (1+ from) to)
+      (prog1 (- to from)
+       (if string
+           (compose-string string from to)
+         (compose-region from to))
+       (- to from))))
+
+;; Register a function to compose Unicode diacrtics and marks.
+(let ((patterns '(("\\C^\\c^+" . diacrtic-composition-function))))
+  (let ((c #x300))
+    (while (<= c #x362)
+      (aset composition-function-table (decode-char 'ucs c) patterns)
+      (setq c (1+ c)))
+    (setq c #x20d0)
+    (while (<= c #x20e3)
+      (aset composition-function-table (decode-char 'ucs c) patterns)
+      (setq c (1+ c)))))
+
 (provide 'european)
 
 ;;; european.el ends here