The antiword program can be found here:

Emacs Lisp package to run AntiWord in Emacs

Here is how to run the antiword program on every .doc file you open, and replace the doc file crud with the antiword output.

    (add-to-list 'auto-mode-alist '("\\.doc\\'" . no-word))
    (defun no-word ()
      "Run antiword on the entire buffer."
      (shell-command-on-region (point-min) (point-max) "antiword - " t t))

See also MimeTypesWithGnus to see how to use antiword to view attached doc files.


This section lists alternatives to the code snipplet above; once a good alternative is found, replace the code above and delete all alternatives.

Here is an alternative which asks you for a coding system, and starts antiword with the -m option to supply the name of a mapping file.

    (defvar no-word-coding-systems '(("greek-iso-8bit" . "8859-7.txt")
                                     ("iso-8859-7" . "8859-7.txt")
                                     ("iso-8859-1" . "8859-1.txt"))
      "Alist mapping coding system to antiword map file.")
    (defun no-word ()
      "Run antiword on the entire buffer."
      (let ((map-file (cdr (assoc (completing-read
                                   "Select coding: (default iso-8859-1) "
                                   nil t nil nil "iso-8859-1")
            (modified (buffer-modified-p)))
         (point-min) (point-max) 
         (format "antiword -m %s -" map-file)
        (set-buffer-modified-p modified)))

The upper version doesn’t change the coding of the buffer. Therefore it doesn’t allow to see the document in another coding than your current coding unless you first do C-x RET c RET but then you are asked twice for the coding. I put another version which uses the no-word-coding-systems variable and displays the document according the choosen coding :

    (defun no-word ()
      "Run antiword on the entire buffer."
      (let* ((no-word-coding (completing-read
                              "Select coding: (default iso-8859-1) "
                              nil t nil nil "iso-8859-1"))
             (map-file (cdr (assoc no-word-coding no-word-coding-systems)))
             (coding-system-for-read (intern no-word-coding))
             (new_name (concat "*" (buffer-name) "*")))
          (shell-command-on-region (point-min) (point-max) (format "antiword -m %s -"  map-file) new_name)  
          (kill-buffer (current-buffer)))
        (switch-to-buffer new_name)))

Note that I didn’t manage to do the same thing within the same XXX.doc buffer. Well, i could have rename the buffer it creates but it doesn’t bother me too much no to have a buffer named XXX.doc since it prevents to accidentally overwrite the word document (which somes might think it’s not a great lost).

Note that you might also want to test wether the file actually contains Microsoft stuff. If so, consider the output of the file program:

    ~% file sounds\!041102.doc 
    sounds!041102.doc: Microsoft Office document data

Therefore in your code:

    (when (string-match "Microsoft "
           (shell-command-to-string (concat "file " buffer-file-name)))

CategoryWThirtyTwo CategoryExternalUtilities