[Orgmode] Context-sensitive word count in org mode (elisp)

[Paul Sexton]

I am trying to reduce the word count in a document I am writing. The
existing word count functionality for emacs is surprisingly lacking.
I wanted a word count function for org mode which excluded tables and 
comments, and ended up writing one myself. 

If this function is called with a region highlighted, it counts the words in
the region. Otherwise it counts words in the whole buffer.

It ignores commented lines and tables. LaTeX-style macros such as 
\foo{bar,baz} are counted as 1 word, as a compromise (more often than not 
they should count as 0, but they do sometimes expand to 1 or more words
in the final document). 

Limitations:
- Does not ignore BEGIN_SRC/END_SRC or inline src_* blocks (babel).
  Should be easy enough to add however.
- There is probably a better way of identifying latex macros
  than my 'latex-macro-regexp' below.
- Ignores all org links. I couldn't figure out how to extract "description"
  text from links, but I didn't look very hard.

Improvements welcome.

Paul

------------------------------------------------------------------------

(defun in-comment-p ()
  "Return non-nil if point is in a comment."
  (if (or (null comment-start-skip)
          (eq (preceding-char) ?\r))
      nil
    (save-excursion
      (let ((pos (point)))
        (re-search-backward "^\\|\r" nil t)
        (or (looking-at comment-start-skip)
            (re-search-forward comment-start-skip pos t))))))

(defun in-org-table-p ()
  "Return non-nil if point is in an org-mode table."
  (if (or (not (boundp 'org-table-any-line-regexp))
          (null org-table-any-line-regexp)
          (eq (preceding-char) ?\r))
      nil
    (save-excursion
      (let ((pos (point)))
        (re-search-backward "^\\|\r" nil t)
        (looking-at org-table-any-line-regexp)))))


(defvar latex-macro-regexp "\\\\[A-Za-z]+\\(\\[[^]]*\\]\\|\\){\\([^}]*\\)}")


(defun org-word-count (beg end)
  (interactive "r")
  (unless mark-active
    (setf beg (point-min)
          end (point-max)))
  (let ((wc 0))
    (save-excursion
      (goto-char beg)
      (while (< (point) end)
        (re-search-forward "\\w+\\W*")
        (cond
         ((or (in-comment-p) (in-org-table-p))
          nil)
         ((looking-at org-any-link-re)
          (goto-char (match-end 0)))
         ((save-excursion
            (backward-char)
            (looking-at latex-macro-regexp))
          (goto-char (match-end 0))
          (setf wc (+ 2 wc)))
         (t
          (incf wc)))))
    (message (format "%d words in %s." wc
                     (if mark-active "region" "buffer")))))