How to convert a MULE encoded file to UTF-8?

Question

I am trying to modify a buffer written by desktop-save to a file. The file is named .emacs.desktop and I am trying to modify it from a Perl script. It turns out that it is saved in emacs-mule encoding.

From M-x list-coding-systems :

U -- utf-8 (alias: mule-utf-8)
  UTF-8 (no signature (BOM))
M -- emacs-mule
  Emacs 21 internal format used in buffer and string.

I would like to read the file in Perl as UTF-8, how can this be done? (I would prefer a solution that does not depend on a running Emacs server)

(An example: emacs-mule saves the letter æ as two bytes 0x81 0xe6, ISO Latin 1 saves it as a single byte 0xe6, whereas UTF-8 saves it as two bytes 0xc3 0xa6.. )

Answer 1

Here is the solution I ended up with:

• Assume an Emacs server with name my_server is up and running.

• Convert emacs-mule to utf-8 by calling emacsclient from the Perl script:

  emacsclient -s my_server -e '(my-convert-mule-utf8 \".emacs.desktop\")' >/dev/null 2>&1`
  

where my-convert-mule-utf8 is

(defun my-convert-mule-utf8 (fn)
  (let*  ((temp-fn (concat fn ".utf8"))
         (coding-system-for-write 'utf-8))
    (with-temp-file temp-fn
      (insert-file-contents fn)
      (when (re-search-forward "coding: \\(emacs-mule\\);" (line-end-position) t)
        (replace-match "utf-8" nil t nil 1)))))

• To convert utf-8 to emacs-mule:

  emacsclient -s my_server -e '(my-convert-utf8-mule \".emacs.desktop\")' >/dev/null 2>&1
  

where my-convert-utf8-mule is:

(defun my-convert-utf8-mule (fn)
  (let* ((temp-fn (concat fn ".utf8"))
         (coding-system-for-write 'emacs-mule))
    (with-temp-file fn
      (insert-file-contents temp-fn)
      (when (re-search-forward "coding: \\(utf-8\\);" (line-end-position) t)
        (replace-match "emacs-mule" nil t nil 1)))))