bug#1770: 23.0.60; (message-check 'illegible-text ...) fails on eight-bit chars

bug#1770: 23.0.60; (message-check 'illegible-text ...) fails on eight-bit chars

[Reiner Steib]

Package: emacs,gnus
Version: 23.0.60

When replying to an article with a bogus charset declaration
(e.g. charset="ISO 8859-15" produced by Knode; "ISO-8859-15" would be
correct. An example in gmane.test is
<873ag15s04.not-fsf@marauder.physik.uni-ulm.de>) with Gnus, the buffer
contains eight-bit-control characters.

When sending the reply, Gnus asks: "Use ASCII as charset? " (see
`mml-parse-1').

Option 1: Answer `y'.

Result: The reply is sent with charset=us-ascii, but it contains 8bit
        characters.


Option 2: Answer `n'.  The Gnus asks a second time.  Answer `n' again.

Result: The same as above.
        See <87wsddtn9r.fsf@marauder.physik.uni-ulm.de> in gmane.test


Expected behavior:

This following code from `message-fix-before-sending' should kick in:
(This is what happens in Emacs 22 with current Gnus CVS trunk,
i.e. the same Gnus code base as Emacs 23.)

  (message-check 'illegible-text
    (let (char found choice)
      (message-goto-body)
      (while (progn
	       (skip-chars-forward mm-7bit-chars)
	       (when (get-text-property (point) 'no-illegible-text)
		 ;; There is a signed or encrypted raw message part
		 ;; that is considered to be safe.
		 (goto-char (or (next-single-property-change
				 (point) 'no-illegible-text)
				(point-max))))
	       (setq char (char-after)))
	(when (or (< (mm-char-int char) 128)
		  (and (mm-multibyte-p)
		       (memq (char-charset char)
			     '(eight-bit-control eight-bit-graphic
						 control-1))
		       (not (get-text-property
			     (point) 'untranslated-utf-8))))
	  (message-overlay-put (message-make-overlay (point) (1+ (point)))
			       'face 'highlight)
	  (setq found t))
	(forward-char))
      (when found
	(setq choice
	      (gnus-multiple-choice
	       "Non-printable characters found.  Continue sending?"
	       `((?d "Remove non-printable characters and send")
		 (?r ,(format
		       "Replace non-printable characters with \"%s\" and send"
		       message-replacement-char))
		 (?i "Ignore non-printable characters and send")
		 (?e "Continue editing"))))
	(if (eq choice ?e)
	  (error "Non-printable characters"))
	(message-goto-body)
	(skip-chars-forward mm-7bit-chars)
	(while (not (eobp))
	  (when (let ((char (char-after)))
		  (or (< (mm-char-int char) 128)
		      (and (mm-multibyte-p)
			   ;; FIXME: Wrong for Emacs 23 (unicode) and for
			   ;; things like undecable utf-8.  Should at least
			   ;; use find-coding-systems-region.
			   (memq (char-charset char)
				 '(eight-bit-control eight-bit-graphic
						     control-1))
			   (not (get-text-property
				 (point) 'untranslated-utf-8)))))
	    (if (eq choice ?i)
		(message-kill-all-overlays)
	      (delete-char 1)
	      (when (eq choice ?r)
		(insert message-replacement-char))))
	  (forward-char)
	  (skip-chars-forward mm-7bit-chars)))))

In Emacs 23, (char-charset char) returns `eight-bit'.  Is adding
eight-bit next to eight-bit-graphic sufficient?  The comment (by Dave
Love, CC-ed if I got X-Debbugs-CC right) seems to suggest that there's
more to be done.

Bye, Reiner.



In GNU Emacs 23.0.60.1 (i686-pc-linux-gnu, GTK+ Version 2.12.9)
 of 2009-01-01 on primula
Windowing system distributor `The X.Org Foundation', version 11.0.10400090
Important settings:
  value of $LC_ALL: nil
  value of $LC_COLLATE: nil
  value of $LC_CTYPE: nil
  value of $LC_MESSAGES: nil
  value of $LC_MONETARY: nil
  value of $LC_NUMERIC: nil
  value of $LC_TIME: nil
  value of $LANG: en_US.UTF-8
  value of $XMODIFIERS: nil
  locale-coding-system: utf-8-unix
  default-enable-multibyte-characters: t
-- 
       ,,,
      (o o)
---ooO-(_)-Ooo---  |  PGP key available  |  http://rsteib.home.pages.de/