Org ad hoc code, quick hacks and workarounds

Table of Contents

{Back to Worg's index}

This page is for ad hoc bits of code. Feel free to add quick hacks and workaround. Go crazy.

Hacking Org: Working within Org-mode.

Org Agenda

Picking up a random task in the global TODO list

Tony day shared this gist to pick up a random task.

Building and Managing Org

Generating autoloads and Compiling Org without make

Compilation is optional, but you must update the autoloads file each time you update org, even when you run org uncompiled!

Starting with Org 7.9 you'll find functions for creating the autoload files and do byte-compilation in mk/org-fixup.el. When you execute the commands below, your current directory must be where org has been unpacked into, in other words the file README should be found in your current directory and the directories lisp and etc should be subdirectories of it. The command emacs should be found in your PATH and start the Emacs version you are using. To make just the autoloads file do:

emacs -batch -Q -L lisp -l ../mk/org-fixup -f org-make-autoloads

To make the autoloads file and byte-compile org:

emacs -batch -Q -L lisp -l ../mk/org-fixup -f org-make-autoloads-compile

To make the autoloads file and byte-compile all of org again:

emacs -batch -Q -L lisp -l ../mk/org-fixup -f org-make-autoloads-compile-force

If you are not using Git, you'll have to make fake version strings first if org-version.el is not already available (if it is, you could also edit the version strings there).

emacs -batch -Q -L lisp -l ../mk/org-fixup \
--eval '(let ((org-fake-release "7.9.1")(org-fake-git-version "7.9.1-fake"))\

The above assumes a POSIX shell for its quoting. Windows CMD.exe has quite different quoting rules and this won't work, so your other option is to start Emacs like this

emacs -Q -L lisp -l ../mk/org-fixup

then paste the following into the *scratch* buffer

(let ((org-fake-release     "7.9.1")
      (org-fake-git-version "7.9.1-fake"))

position the cursor after the closing paren and press C-j or C-x C-e to evaluate the form. Of course you can replace org-make-autoloads with org-make-autoloads-compile or even org-make-autoloads-compile-force if you wish with both variants.

For older org versions only (that do not yet have mk/org-fixup.el), you can use the definitions below. To use this function, adjust the variables my/org-lisp-directory and my/org-compile-sources to suit your needs. If you have byte-compiled org, but want to run org uncompiled again, just remove all *.elc files in the lisp/ directory, set my/org-compile-sources to nil.

(defvar my/org-lisp-directory "~/.emacs.d/org/lisp/"
  "Directory where your org-mode files live.")

(defvar my/org-compile-sources t
  "If `nil', never compile org-sources. `my/compile-org' will only create
the autoloads file `org-loaddefs.el' then. If `t', compile the sources, too.")

;; Customize: (must end with a slash!)
(setq my/org-lisp-directory "~/.emacs.d/org/lisp/")

;; Customize:
(setq  my/org-compile-sources t)

(defun my/compile-org(&optional directory)
  "Generate autoloads file org-loaddefs.el.  Optionally compile
   all *.el files that come with org-mode."
  (defun my/compile-org()
    "Generate autoloads file org-loaddefs.el.  Optionally compile
     all *.el files that come with org-mode."
    (let ((dirlisp (file-name-directory my/org-lisp-directory)))
      (add-to-list 'load-path dirlisp)
      (require 'autoload)
      (let ((generated-autoload-file (concat dirlisp "org-loaddefs.el")))
        ;; create the org-loaddefs file
        (update-directory-autoloads dirlisp)
        (when my/org-compile-sources
          ;; optionally byte-compile
          (byte-recompile-directory dirlisp 0 'force)))))

Reload Org

As of Org version 6.23b (released Sunday Feb 22, 2009) there is a new function to reload org files.

Normally you want to use the compiled files since they are faster. If you update your org files you can easily reload them with

M-x org-reload

If you run into a bug and want to generate a useful backtrace you can reload the source files instead of the compiled files with

C-u M-x org-reload

and turn on the "Enter Debugger On Error" option. Redo the action that generates the error and cut and paste the resulting backtrace. To switch back to the compiled version just reload again with

M-x org-reload

Structure Movement and Editing

Show next/prev heading tidily

  • Dan Davison These close the current heading and open the next/previous heading.
(defun ded/org-show-next-heading-tidily ()
  "Show next entry, keeping other entries closed."
  (if (save-excursion (end-of-line) (outline-invisible-p))
      (progn (org-show-entry) (show-children))
    (unless (and (bolp) (org-on-heading-p))
      (error "Boundary reached"))
    (org-reveal t)

(defun ded/org-show-previous-heading-tidily ()
  "Show previous entry, keeping other entries closed."
  (let ((pos (point)))
    (unless (and (< (point) pos) (bolp) (org-on-heading-p))
      (goto-char pos)
      (error "Boundary reached"))
    (org-reveal t)

(setq org-use-speed-commands t)
(add-to-list 'org-speed-commands-user
             '("n" ded/org-show-next-heading-tidily))
(add-to-list 'org-speed-commands-user
             '("p" ded/org-show-previous-heading-tidily))

Promote all items in subtree

  • Matt Lundin

This function will promote all items in a subtree. Since I use subtrees primarily to organize projects, the function is somewhat unimaginatively called my-org-un-project:

(defun my-org-un-project ()
  (org-map-entries 'org-do-promote "LEVEL>1" 'tree)
  (org-cycle t))

Using M-up and M-down to transpose paragraphs

From Paul Sexton: By default, if used within ordinary paragraphs in org mode, M-up and M-down transpose lines (not sentences). The following code makes these keys transpose paragraphs, keeping the point at the start of the moved paragraph. Behavior in tables and headings is unaffected. It would be easy to modify this to transpose sentences.

(defun org-transpose-paragraphs (arg)
 (when (and (not (or (org-at-table-p) (org-on-heading-p) (org-at-item-p)))
            (thing-at-point 'sentence))
   (transpose-paragraphs arg)
   (re-search-forward "[[:graph:]]")
   (goto-char (match-beginning 0))

(add-to-list 'org-metaup-hook 
 (lambda () (interactive) (org-transpose-paragraphs -1)))
(add-to-list 'org-metadown-hook 
 (lambda () (interactive) (org-transpose-paragraphs 1)))

Changelog support for org headers

– James TD Smith

Put the following in your .emacs, and C-x 4 a and other functions which use add-log-current-defun like magit-add-log will pick up the nearest org headline as the "current function" if you add a changelog entry from an org buffer.

(defun org-log-current-defun ()
    (if (looking-at org-complex-heading-regexp)
        (match-string 4))))

(add-hook 'org-mode-hook
          (lambda ()
            (make-variable-buffer-local 'add-log-current-defun-function)
            (setq add-log-current-defun-function 'org-log-current-defun)))

Different org-cycle-level behavior

– Ryan Thompson

In recent org versions, when your point (cursor) is at the end of an empty header line (like after you first created the header), the TAB key (org-cycle) has a special behavior: it cycles the headline through all possible levels. However, I did not like the way it determined "all possible levels," so I rewrote the whole function, along with a couple of supporting functions.

The original function's definition of "all possible levels" was "every level from 1 to one more than the initial level of the current headline before you started cycling." My new definition is "every level from 1 to one more than the previous headline's level." So, if you have a headline at level 4 and you use ALT+RET to make a new headline below it, it will cycle between levels 1 and 5, inclusive.

The main advantage of my custom org-cycle-level function is that it is stateless: the next level in the cycle is determined entirely by the contents of the buffer, and not what command you executed last. This makes it more predictable, I hope.

(require 'cl)

(defun org-point-at-end-of-empty-headline ()
  "If point is at the end of an empty headline, return t, else nil."
  (and (looking-at "[ \t]*$")
         (beginning-of-line 1)
         (looking-at (concat "^\\(\\*+\\)[ \t]+\\(" org-todo-regexp "\\)?[ \t]*")))))

(defun org-level-increment ()
  "Return the number of stars that will be added or removed at a
time to headlines when structure editing, based on the value of
  (if org-odd-levels-only 2 1))

(defvar org-previous-line-level-cached nil)

(defun org-recalculate-previous-line-level ()
  "Same as `org-get-previous-line-level', but does not use cached
value. It does *set* the cached value, though."
  (set 'org-previous-line-level-cached
       (let ((current-level (org-current-level))
             (prev-level (when (> (line-number-at-pos) 1)
         (cond ((null current-level) nil) ; Before first headline
               ((null prev-level) 0)      ; At first headline

(defun org-get-previous-line-level ()
  "Return the outline depth of the last headline before the
current line. Returns 0 for the first headline in the buffer, and
nil if before the first headline."
  ;; This calculation is quite expensive, with all the regex searching
  ;; and stuff. Since org-cycle-level won't change lines, we can reuse
  ;; the last value of this command.
  (or (and (eq last-command 'org-cycle-level)

(defun org-cycle-level ()
  (let ((org-adapt-indentation nil))
    (when (org-point-at-end-of-empty-headline)
      (setq this-command 'org-cycle-level) ;Only needed for caching
      (let ((cur-level (org-current-level))
            (prev-level (org-get-previous-line-level)))
         ;; If first headline in file, promote to top-level.
         ((= prev-level 0)
          (loop repeat (/ (- cur-level 1) (org-level-increment))
                do (org-do-promote)))
         ;; If same level as prev, demote one.
         ((= prev-level cur-level)
         ;; If parent is top-level, promote to top level if not already.
         ((= prev-level 1)
          (loop repeat (/ (- cur-level 1) (org-level-increment))
                do (org-do-promote)))
         ;; If top-level, return to prev-level.
         ((= cur-level 1)
          (loop repeat (/ (- prev-level 1) (org-level-increment))
                do (org-do-demote)))
         ;; If less than prev-level, promote one.
         ((< cur-level prev-level)
         ;; If deeper than prev-level, promote until higher than
         ;; prev-level.
         ((> cur-level prev-level)
          (loop repeat (+ 1 (/ (- cur-level prev-level) (org-level-increment)))
                do (org-do-promote))))

Count words in an Org buffer

#FIXME: Does not fit too well under Structure. Any idea where to put it? Paul Sexton posted this function to count words in an Org buffer:

(defun org-word-count (beg end
                           &optional count-latex-macro-args?
  "Report the number of words in the Org mode buffer or selected region.
- comments
- tables
- source code blocks (#+BEGIN_SRC ... #+END_SRC, and inline blocks)
- hyperlinks (but does count words in hyperlink descriptions)
- tags, priorities, and TODO keywords in headers
- sections tagged as 'not for export'.

The text of footnote definitions is ignored, unless the optional argument
COUNT-FOOTNOTES? is non-nil.

If the optional argument COUNT-LATEX-MACRO-ARGS? is non-nil, the word count
includes LaTeX macro arguments (the material between {curly braces}).
Otherwise, and by default, every LaTeX macro counts as 1 word regardless
of its arguments."
  (interactive "r")
  (unless mark-active
    (setf beg (point-min)
      end (point-max)))
  (let ((wc 0)
    (latex-macro-regexp "\\\\[A-Za-z]+\\(\\[[^]]*\\]\\|\\){\\([^}]*\\)}"))
      (goto-char beg)
      (while (< (point) end)
         ;; Ignore comments.
         ((or (org-in-commented-line) (org-at-table-p))
         ;; Ignore hyperlinks. But if link has a description, count
         ;; the words within the description.
         ((looking-at org-bracket-link-analytic-regexp)
          (when (match-string-no-properties 5)
            (let ((desc (match-string-no-properties 5)))
                (incf wc (length (remove "" (org-split-string
                                             desc "\\W")))))))
          (goto-char (match-end 0)))
         ((looking-at org-any-link-re)
          (goto-char (match-end 0)))
         ;; Ignore source code blocks.
         ((org-in-regexps-block-p "^#\\+BEGIN_SRC\\W" "^#\\+END_SRC\\W")
         ;; Ignore inline source blocks, counting them as 1 word.
            (looking-at org-babel-inline-src-block-regexp))
          (goto-char (match-end 0))
          (setf wc (+ 2 wc)))
         ;; Count latex macros as 1 word, ignoring their arguments.
            (looking-at latex-macro-regexp))
          (goto-char (if count-latex-macro-args?
                         (match-beginning 2)
                       (match-end 0)))
          (setf wc (+ 2 wc)))
         ;; Ignore footnotes.
         ((and (not count-footnotes?)
               (or (org-footnote-at-definition-p)
          (let ((contexts (org-context)))
             ;; Ignore tags and TODO keywords, etc.
             ((or (assoc :todo-keyword contexts)
                  (assoc :priority contexts)
                  (assoc :keyword contexts)
                  (assoc :checkbox contexts))
             ;; Ignore sections marked with tags that are
             ;; excluded from export.
             ((assoc :tags contexts)
              (if (intersection (org-get-tags-at) org-export-exclude-tags
                                :test 'equal)
                  (org-forward-same-level 1)
              (incf wc))))))
        (re-search-forward "\\w+\\W*")))
    (message (format "%d words in %s." wc
                     (if mark-active "region" "buffer")))))

Check for misplaced SCHEDULED and DEADLINE cookies

The SCHEDULED and DEADLINE cookies should be used on the line right below the headline – like this:

* A headline
,  SCHEDULED: <2012-04-09 lun.>

This is what org-scheduled and org-deadline (and other similar commands) do. And the manual explicitely tell people to stick to this format (see the section "8.3.1 Inserting deadlines or schedules").

If you think you might have subtrees with misplaced SCHEDULED and DEADLINE cookies, this command lets you check the current buffer:

(defun org-check-misformatted-subtree ()
  "Check misformatted entries in the current buffer."
   (lambda ()
     (when (and (move-beginning-of-line 2)
        (not (looking-at org-heading-regexp)))
       (if (or (and (org-get-scheduled-time (point))
            (not (looking-at (concat "^.*" org-scheduled-regexp))))
           (and (org-get-deadline-time (point))
            (not (looking-at (concat "^.*" org-deadline-regexp)))))
       (when (y-or-n-p "Fix this subtree? ")
         (message "Call the function again when you're done fixing this subtree.")
     (message "All subtrees checked."))))))

Sorting list by checkbox type

You can use a custom function to sort list by checkbox type. Here is a function suggested by Carsten:

(defun org-sort-list-by-checkbox-type ()
  "Sort list items according to Checkbox state."
   nil ?f
   (lambda ()
     (if (looking-at org-list-full-item-re)
     (cdr (assoc (match-string 3)
             '(("[X]" . 1) ("[-]" . 2) ("[ ]" . 3) (nil . 4))))

Use the function above directly on the list. If you want to use an equivalent function after C-c ^ f, use this one instead:

(defun org-sort-list-by-checkbox-type-1 ()
  (lambda ()
    (if (looking-at org-list-full-item-re)
        (cdr (assoc (match-string 3)
                    '(("[X]" . 1) ("[-]" . 2) ("[ ]" . 3) (nil . 4))))

Org Table

Align all tables in a file

Andrew Young provided this function in this thread:

(defun my-align-all-tables ()
  (org-table-map-tables 'org-table-align 'quietly))

Transpose table

Since Org 7.8, you can use org-table-transpose-table-at-point (which see.) There are also other solutions:

  • with org-babel and Emacs Lisp: provided by Thomas S. Dye in the mailing list, see gmane or gnu
  • with org-babel and R: provided by Dan Davison in the mailing list (old #+TBLR: syntax), see gmane or gnu
  • with field coordinates in formulas (@# and $#): see Worg.

Manipulate hours/minutes/seconds in table formulas

Both Bastien and Martin Halder have posted code (Bastien's code and Martin's code) for interpreting dd:dd or dd:dd:dd strings (where "d" is any digit) as time values in Org-mode table formula. These functions have now been wrapped up into a with-time macro which can be used in table formula to translate table cell values to and from numerical values for algebraic manipulation.

Here is the code implementing this macro.

(defun org-time-string-to-seconds (s)
  "Convert a string HH:MM:SS to a number of seconds."
   ((and (stringp s)
         (string-match "\\([0-9]+\\):\\([0-9]+\\):\\([0-9]+\\)" s))
    (let ((hour (string-to-number (match-string 1 s)))
          (min (string-to-number (match-string 2 s)))
          (sec (string-to-number (match-string 3 s))))
      (+ (* hour 3600) (* min 60) sec)))
   ((and (stringp s)
         (string-match "\\([0-9]+\\):\\([0-9]+\\)" s))
    (let ((min (string-to-number (match-string 1 s)))
          (sec (string-to-number (match-string 2 s))))
      (+ (* min 60) sec)))
   ((stringp s) (string-to-number s))
   (t s)))

(defun org-time-seconds-to-string (secs)
  "Convert a number of seconds to a time string."
  (cond ((>= secs 3600) (format-seconds "%h:%.2m:%.2s" secs))
        ((>= secs 60) (format-seconds "%m:%.2s" secs))
        (t (format-seconds "%s" secs))))

(defmacro with-time (time-output-p &rest exprs)
  "Evaluate an org-table formula, converting all fields that look
like time data to integer seconds.  If TIME-OUTPUT-P then return
the result as a time value."
   (if time-output-p 'org-time-seconds-to-string 'identity)
   (cons 'progn
          (lambda (expr)
            `,(cons (car expr)
                     (lambda (el)
                       (if (listp el)
                           (list 'with-time nil el)
                         (org-time-string-to-seconds el)))
                     (cdr expr))))

Which allows the following forms of table manipulation such as adding and subtracting time values.

| Date             | Start | Lunch |  Back |   End |  Sum |
| [2011-03-01 Tue] |  8:00 | 12:00 | 12:30 | 18:15 | 9:45 |
#+TBLFM: $6='(with-time t (+ (- $5 $4) (- $3 $2)))

and dividing time values by integers

|  time | miles | minutes/mile |
| 34:43 |   2.9 |        11:58 |
| 32:15 |  2.77 |        11:38 |
| 33:56 |   3.0 |        11:18 |
| 52:22 |  4.62 |        11:20 |
#+TBLFM: $3='(with-time t (/ $1 $2))

Update: As of Org version 7.6, you can use the T flag (both in Calc and Elisp formulas) to compute time durations. For example:

| Task 1 | Task 2 |   Total |
|  35:00 |  35:00 | 1:10:00 |
#+TBLFM: @2$3=$1+$2;T

Dates computation

Xin Shi asked for a way to calculate the duration of dates stored in an org table.

Nick Dokos suggested:

Try the following:

| Start Date |   End Date | Duration |
| 2004.08.07 | 2005.07.08 |      335 |
#+TBLFM: $3=(date(<$2>)-date(<$1>))

See this thread as well as this post (which is really a followup on the above). The problem that this last article pointed out was solved in this post and Chris Randle's original musings are here.

Hex computation

As with Times computation, the following code allows Computation with Hex values in Org-mode tables using the with-hex macro.

Here is the code implementing this macro.

(defun org-hex-strip-lead (str)
  (if (and (> (length str) 2) (string= (substring str 0 2) "0x"))
      (substring str 2) str))

(defun org-hex-to-hex (int)
  (format "0x%x" int))

(defun org-hex-to-dec (str)
   ((and (stringp str)
         (string-match "\\([0-9a-f]+\\)" (setf str (org-hex-strip-lead str))))
    (let ((out 0))
       (lambda (ch)
         (setf out (+ (* out 16)
                      (if (and (>= ch 48) (<= ch 57)) (- ch 48) (- ch 87)))))
       (coerce (match-string 1 str) 'list))
   ((stringp str) (string-to-number str))
   (t str)))

(defmacro with-hex (hex-output-p &rest exprs)
  "Evaluate an org-table formula, converting all fields that look
    like hexadecimal to decimal integers.  If HEX-OUTPUT-P then
    return the result as a hex value."
   (if hex-output-p 'org-hex-to-hex 'identity)
   (cons 'progn
          (lambda (expr)
            `,(cons (car expr)
                    (mapcar (lambda (el)
                              (if (listp el)
                                  (list 'with-hex nil el)
                                (org-hex-to-dec el)))
                            (cdr expr))))

Which allows the following forms of table manipulation such as adding and subtracting hex values.

0x10 0x0 0x10 16
0x20 0x1 0x21 33
0x30 0x2 0x32 50
0xf0 0xf 0xff 255

Field coordinates in formulas (@# and $#)

– Michael Brand

Following are some use cases that can be implemented with the “field coordinates in formulas” described in the corresponding chapter in the Org manual.

  • Copy a column from a remote table into a column

    current column $3 = remote column $2:

    #+TBLFM: $3 = remote(FOO, @@#$2)
  • Copy a row from a remote table transposed into a column

    current column $1 = transposed remote row @1:

    #+TBLFM: $1 = remote(FOO, @$#$@#)
  • Transpose table

    – Michael Brand

    This is more like a demonstration of using “field coordinates in formulas” and is bound to be slow for large tables. See the discussion in the mailing list on gmane or gnu. For more efficient solutions see Worg.

    To transpose this 4x7 table

    | year | 2004 | 2005 | 2006 | 2007 | 2008 | 2009 |
    | min  |  401 |  501 |  601 |  701 |  801 |  901 |
    | avg  |  402 |  502 |  602 |  702 |  802 |  902 |
    | max  |  403 |  503 |  603 |  703 |  803 |  903 |

    start with a 7x4 table without any horizontal line (to have filled also the column header) and yet empty:

    |   |   |   |   |
    |   |   |   |   |
    |   |   |   |   |
    |   |   |   |   |
    |   |   |   |   |
    |   |   |   |   |
    |   |   |   |   |

    Then add the TBLFM line below. After recalculation this will end up with the transposed copy:

    | year | min | avg | max |
    | 2004 | 401 | 402 | 403 |
    | 2005 | 501 | 502 | 503 |
    | 2006 | 601 | 602 | 603 |
    | 2007 | 701 | 702 | 703 |
    | 2008 | 801 | 802 | 803 |
    | 2009 | 901 | 902 | 903 |
    #+TBLFM: @<$<..@>$> = remote(FOO, @$#$@#)

    The formula simply exchanges row and column numbers by taking

    • the absolute remote row number @$# from the current column number $#
    • the absolute remote column number $@# from the current row number @#

    Formulas to be taken over from the remote table will have to be transformed manually.

  • Dynamic variation of ranges

    – Michael Brand

    In this example all columns next to quote are calculated from the column quote and show the average change of the time series quote[year] during the period of the preceding 1, 2, 3 or 4 years:

    | year | quote |   1 a |   2 a |   3 a |   4 a |
    | 2005 |    10 |       |       |       |       |
    | 2006 |    12 | 0.200 |       |       |       |
    | 2007 |    14 | 0.167 | 0.183 |       |       |
    | 2008 |    16 | 0.143 | 0.155 | 0.170 |       |
    | 2009 |    18 | 0.125 | 0.134 | 0.145 | 0.158 |
    #+TBLFM: @I$3..@>$>=if(@# >= $#, ($2 / subscr(@-I$2..@+I$2, @# + 1 - $#)) ^ (1 / ($# - 2)) - 1, string("")) +.0; f-3

    The important part of the formula without the field blanking is:

    ($2 / subscr(@-I$2..@+I$2, @# + 1 - $#)) ^ (1 / ($# - 2)) - 1

    which is the Emacs Calc implementation of the equation

    AvgChange(i, a) = (quote[i] / quote[i - a]) ^ (1 / a) - 1

    where i is the current time and a is the length of the preceding period.

Change the column sequence in one row only

– Michael Brand

The functions below can be used to change the column sequence in one row only, without affecting the other rows above and below like with M-<left> or M-<right> (org-table-move-column). Please see the docstring of the functions for more explanations. Below is one example per function, with this original table as the starting point for each example:

| a | b | c  | d  |
| e | 9 | 10 | 11 |
| f | g | h  | i  |
  • Move in row left
    1. place point at "10" in original table
    2. result of M-x my-org-table-move-column-in-row-left:
      | a | b  | c | d  |
      | e | 10 | 9 | 11 |
      | f | g  | h | i  |
  • Move in row right
    1. place point at "9" in original table
    2. result of M-x my-org-table-move-column-in-row-right:
      | a | b  | c | d  |
      | e | 10 | 9 | 11 |
      | f | g  | h | i  |
  • Rotate in row left
    1. place point at "9" in original table
    2. result of M-x my-org-table-rotate-column-in-row-left:
      | a | b  | c  | d |
      | e | 10 | 11 | 9 |
      | f | g  | h  | i |
  • Rotate in row right
    1. place point at "9" in original table
    2. result of M-x my-org-table-rotate-column-in-row-right:
      | a | b  | c | d  |
      | e | 11 | 9 | 10 |
      | f | g  | h | i  |
  • The functions
    (defun my-org-table-move-column-in-row-right ()
      "Move column to the right, limited to the current row."
      (my-org-table-move-column-in-row nil))
    (defun my-org-table-move-column-in-row-left ()
      "Move column to the left, limited to the current row."
      (my-org-table-move-column-in-row 'left))
    (defun my-org-table-move-column-in-row (&optional left)
      "Move the current column to the right, limited to the current row.
    With arg LEFT, move to the left.  For repeated invocation the point follows
    the value and changes to the target colum.  Does not fix formulas."
      ;; derived from `org-table-move-column'
      (interactive "P")
      (if (not (org-at-table-p))
          (error "Not at a table"))
      (let* ((col (org-table-current-column))
             (col1 (if left (1- col) col))
             ;; Current cursor position
             (colpos (if left (1- col) (1+ col))))
        (if (and left (= col 1))
            (error "Cannot move column further left"))
        (if (and (not left) (looking-at "[^|\n]*|[^|\n]*$"))
            (error "Cannot move column further right"))
        (org-table-goto-column col1 t)
        (and (looking-at "|\\([^|\n]+\\)|\\([^|\n]+\\)|")
             (replace-match "|\\2|\\1|"))
        (org-table-goto-column colpos)
    (defun my-org-table-rotate-column-in-row-right ()
      "Rotate column to the right, limited to the current row."
      (my-org-table-rotate-column-in-row nil))
    (defun my-org-table-rotate-column-in-row-left ()
      "Rotate column to the left, limited to the current row."
      (my-org-table-rotate-column-in-row 'left))
    (defun my-org-table-rotate-column-in-row (&optional left)
      "Rotate the current column to the right, limited to the current row.
    With arg LEFT, rotate to the left.  The boundaries of the rotation range are
    the current and the most right column for both directions.  For repeated
    invocation the point stays on the current column.  Does not fix formulas."
      ;; derived from `org-table-move-column'
      (interactive "P")
      (if (not (org-at-table-p))
          (error "Not at a table"))
      (let ((col (org-table-current-column)))
        (org-table-goto-column col t)
        (and (looking-at (if left
             (replace-match "|\\2|\\1|"))
        (org-table-goto-column col)
  • Key bindings

    As hack I have this in an Org buffer to change temporarily to the desired behavior with C-c C-c on one of the three snippets:

    - move in row:
      #+begin_src emacs-lisp :results silent
        (org-defkey org-mode-map [(meta left)]
        (org-defkey org-mode-map [(meta right)]
        (org-defkey org-mode-map [(left)]  'org-table-previous-field)
        (org-defkey org-mode-map [(right)] 'org-table-next-field)
    - rotate in row:
      #+begin_src emacs-lisp :results silent
        (org-defkey org-mode-map [(meta left)]
        (org-defkey org-mode-map [(meta right)]
        (org-defkey org-mode-map [(left)]  'org-table-previous-field)
        (org-defkey org-mode-map [(right)] 'org-table-next-field)
    - back to original:
      #+begin_src emacs-lisp :results silent
        (org-defkey org-mode-map [(meta left)]  'org-metaleft)
        (org-defkey org-mode-map [(meta right)] 'org-metaright)
        (org-defkey org-mode-map [(left)]  'backward-char)
        (org-defkey org-mode-map [(right)] 'forward-char)
  • reasons why this is not put into the Org core

    I consider this as only a hack for several reasons:

    • Generalization: The existing org-table-move-column function could be enhanced with additional optional parameters to incorporate these functionalities and could be used as the only function for better maintainability. Now it's only a copy/paste hack of several similar functions with simple modifications.
    • Bindings: Should be convenient for repetition like M-<right>. What should be bound where, what has to be left unbound?
    • Does not fix formulas. Could be resolved for field formulas but most probably not for column or range formulas and this can lead to confusion. AFAIK all "official" table manipulations fix formulas.
    • Completeness: Not all variations and combinations are covered yet
      • left-right, up-down
      • move, rotate with range to end, rotate with range to begin
      • whole column/row, only in-row/in-column

Capture and Remember

Customize the size of the frame for remember

(Note: this hack is likely out of date due to the development of org-capture.)

#FIXME: gmane link? On emacs-orgmode, Ryan C. Thompson suggested this:

I am using org-remember set to open a new frame when used, and the default frame size is much too large. To fix this, I have designed some advice and a custom variable to implement custom parameters for the remember frame:

(defcustom remember-frame-alist nil
  "Additional frame parameters for dedicated remember frame."
  :type 'alist
  :group 'remember)

(defadvice remember (around remember-frame-parameters activate)
  "Set some frame parameters for the remember frame."
  (let ((default-frame-alist (append remember-frame-alist

Setting remember-frame-alist to ((width . 80) (height . 15))) give a reasonable size for the frame.

Handling Links

Quickaccess to the link part of hyperlinks

Christian Moe asked, if there is a simpler way to copy the link part of an org hyperling other than to use `C-c C-l C-a C-k C-g', which is indeed kind of cumbersome.

The thread offered two ways:

Using a keyboard macro:

(fset 'getlink
      (lambda (&optional arg) 
        "Keyboard macro." 
        (interactive "p") 
        (kmacro-exec-ring-item (quote ("\C-c\C-l\C-a\C-k\C-g" 0 "%d")) arg)))

or a function:

(defun my-org-extract-link ()
  "Extract the link location at point and put it on the killring."
  (when (org-in-regexp org-bracket-link-regexp 1)
    (kill-new (org-link-unescape (org-match-string-no-properties 1)))))

They put the link destination on the killring and can be easily bound to a key.

Insert link with HTML title as default description

When using `org-insert-link' (`C-c C-l') it might be useful to extract contents from HTML <title> tag and use it as a default link description. Here is a way to accomplish this:

(require 'mm-url) ; to include mm-url-decode-entities-string

(defun my-org-insert-link ()
  "Insert org link where default description is set to html title."
  (let* ((url (read-string "URL: "))
         (title (get-html-title-from-url url)))
    (org-insert-link nil url title)))

(defun get-html-title-from-url (url)
  "Return content in <title> tag."
  (let (x1 x2 (download-buffer (url-retrieve-synchronously url)))
      (set-buffer download-buffer)
      (setq x1 (search-forward "<title>"))
      (search-forward "</title>")
      (setq x2 (search-backward "<"))
      (mm-url-decode-entities-string (buffer-substring-no-properties x1 x2)))))

Then just use `M-x my-org-insert-link' instead of `org-insert-link'.

Archiving Content in Org-Mode

Preserve top level headings when archiving to a file

  • Matt Lundin

To preserve (somewhat) the integrity of your archive structure while archiving lower level items to a file, you can use the following defadvice:

(defadvice org-archive-subtree (around my-org-archive-subtree activate)
  (let ((org-archive-location
         (if (save-excursion (org-back-to-heading)
                             (> (org-outline-level) 1))
             (concat (car (split-string org-archive-location "::"))
                     "::* "
                     (car (org-get-outline-path)))

Thus, if you have an outline structure such as…

* Heading
** Subheading
*** Subsubheading

…archiving "Subsubheading" to a new file will set the location in the new file to the top level heading:

* Heading
** Subsubheading

While this hack obviously destroys the outline hierarchy somewhat, it at least preserves the logic of level one groupings.

A slightly more complex version of this hack will not only keep the archive organized by top-level headings, but will also preserve the tags found on those headings:

(defun my-org-inherited-no-file-tags ()
  (let ((tags (org-entry-get nil "ALLTAGS" 'selective))
        (ltags (org-entry-get nil "TAGS")))
    (mapc (lambda (tag)
            (setq tags
                  (replace-regexp-in-string (concat tag ":") "" tags)))
          (append org-file-tags (when ltags (split-string ltags ":" t))))
    (if (string= ":" tags) nil tags)))

(defadvice org-archive-subtree (around my-org-archive-subtree-low-level activate)
  (let ((tags (my-org-inherited-no-file-tags))
         (if (save-excursion (org-back-to-heading)
                             (> (org-outline-level) 1))
             (concat (car (split-string org-archive-location "::"))
                     "::* "
                     (car (org-get-outline-path)))
    (with-current-buffer (find-file-noselect (org-extract-archive-file))
        (while (org-up-heading-safe))
        (org-set-tags-to tags)))))

Archive in a date tree

Posted to Org-mode mailing list by Osamu Okano [2010-04-21 Wed].

(Make sure org-datetree.el is loaded for this to work.)

;; (setq org-archive-location "%s_archive::date-tree")
(defadvice org-archive-subtree
  (around org-archive-subtree-to-data-tree activate)
  "org-archive-subtree to date-tree"
      (string= "date-tree"
      (let* ((dct (decode-time (org-current-time)))
             (y (nth 5 dct))
             (m (nth 4 dct))
             (d (nth 3 dct))
             (this-buffer (current-buffer))
             (location (org-get-local-archive-location))
             (afile (org-extract-archive-file location))
              (format "%s::*** %04d-%02d-%02d %s" afile y m d
                      (format-time-string "%A" (encode-time 0 0 0 d m y)))))
        (message "afile=%s" afile)
        (unless afile
          (error "Invalid `org-archive-location'"))
          (switch-to-buffer (find-file-noselect afile))
          (org-datetree-find-year-create y)
          (org-datetree-find-month-create y m)
          (org-datetree-find-day-create y m d)
          (switch-to-buffer this-buffer))

Add inherited tags to archived entries

To make org-archive-subtree keep inherited tags, Osamu OKANO suggests to advise the function like this:

(defadvice org-archive-subtree
  (before add-inherited-tags-before-org-archive-subtree activate)
    "add inherited tags before org-archive-subtree"
    (org-set-tags-to (org-get-tags-at)))

Using and Managing Org-Metadata

Remove redundant tags of headlines

– David Maus

A small function that processes all headlines in current buffer and removes tags that are local to a headline and inherited by a parent headline or the #+FILETAGS: statement.

(defun dmj/org-remove-redundant-tags ()
  "Remove redundant tags of headlines in current buffer.

A tag is considered redundant if it is local to a headline and
inherited by a parent headline."
  (when (eq major-mode 'org-mode)
       '(lambda ()
          (let ((alltags (split-string (or (org-entry-get (point) "ALLTAGS") "") ":"))
                local inherited tag)
            (dolist (tag alltags)
              (if (get-text-property 0 'inherited tag)
                  (push tag inherited) (push tag local)))
            (dolist (tag local)
              (if (member tag inherited) (org-toggle-tag tag 'off)))))
       t nil))))

Remove empty property drawers

David Maus proposed this:

(defun dmj:org:remove-empty-propert-drawers ()
  "*Remove all empty property drawers in current file."
  (unless (eq major-mode 'org-mode)
    (error "You need to turn on Org mode for this function."))
    (goto-char (point-min))
    (while (re-search-forward ":PROPERTIES:" nil t)
    (org-remove-empty-drawer-at "PROPERTIES" (match-beginning 0))))))

Group task list by a property

This advice allows you to group a task list in Org-Mode. To use it, set the variable org-agenda-group-by-property to the name of a property in the option list for a TODO or TAGS search. The resulting agenda view will group tasks by that property prior to searching.

(defvar org-agenda-group-by-property nil
  "Set this in org-mode agenda views to group tasks by property")

(defun org-group-bucket-items (prop items)
  (let ((buckets ()))
    (dolist (item items)
      (let* ((marker (get-text-property 0 'org-marker item))
             (pvalue (org-entry-get marker prop t))
             (cell (assoc pvalue buckets)))
        (if cell
            (setcdr cell (cons item (cdr cell)))
          (setq buckets (cons (cons pvalue (list item))
    (setq buckets (mapcar (lambda (bucket)
                            (cons (car bucket)
                                  (reverse (cdr bucket))))
    (sort buckets (lambda (i1 i2)
                    (string< (car i1) (car i2))))))

(defadvice org-finalize-agenda-entries (around org-group-agenda-finalize
                                               (list &optional nosort))
  "Prepare bucketed agenda entry lists"
  (if org-agenda-group-by-property
      ;; bucketed, handle appropriately
      (let ((text ""))
        (dolist (bucket (org-group-bucket-items
          (let ((header (concat "Property "
                                " is "
                                (or (car bucket) "<nil>") ":\n")))
            (add-text-properties 0 (1- (length header))
                                 (list 'face 'org-agenda-structure)
            (setq text
                  (concat text header
                          ;; recursively process
                          (let ((org-agenda-group-by-property nil))
                             (cdr bucket) nosort))
        (setq ad-return-value text))
(ad-activate 'org-finalize-agenda-entries)

A way to tag a task so that when clocking-out user is prompted to take a note.

Thanks to Richard Riley (see this post on the mailing list).

A small hook run when clocking out of a task that prompts for a note when the tag "clockout_note" is found in a headline. It uses the tag ("clockout_note") so inheritance can also be used…

(defun rgr/check-for-clock-out-note()
        (let ((tags (org-get-tags)))
          (and tags (message "tags: %s " tags)
               (when (member "clocknote" tags)

(add-hook 'org-clock-out-hook 'rgr/check-for-clock-out-note)

Dynamically adjust tag position

Here is a bit of code that allows you to have the tags always right-adjusted in the buffer.

This is useful when you have bigger window than default window-size and you dislike the aesthetics of having the tag in the middle of the line.

This hack solves the problem of adjusting it whenever you change the window size. Before saving it will revert the file to having the tag position be left-adjusted so that if you track your files with version control, you won't run into artificial diffs just because the window-size changed.

IMPORTANT: This is probably slow on very big files.

(setq ba/org-adjust-tags-column t)

(defun ba/org-adjust-tags-column-reset-tags ()
  "In org-mode buffers it will reset tag position according to
  (when (and
         (not (string= (buffer-name) "*Remember*"))
         (eql major-mode 'org-mode))
    (let ((b-m-p (buffer-modified-p)))
      (condition-case nil
            (goto-char (point-min))
            (command-execute 'outline-next-visible-heading)
            ;; disable (message) that org-set-tags generates
            (flet ((message (&rest ignored) nil))
              (org-set-tags 1 t))
            (set-buffer-modified-p b-m-p))
        (error nil)))))

(defun ba/org-adjust-tags-column-now ()
  "Right-adjust `org-tags-column' value, then reset tag position."
  (set (make-local-variable 'org-tags-column)
       (- (- (window-width) (length org-ellipsis))))

(defun ba/org-adjust-tags-column-maybe ()
  "If `ba/org-adjust-tags-column' is set to non-nil, adjust tags."
  (when ba/org-adjust-tags-column

(defun ba/org-adjust-tags-column-before-save ()
  "Tags need to be left-adjusted when saving."
  (when ba/org-adjust-tags-column
     (setq org-tags-column 1)

(defun ba/org-adjust-tags-column-after-save ()
  "Revert left-adjusted tag position done by before-save hook."
  (set-buffer-modified-p nil))

; automatically align tags on right-hand side
(add-hook 'window-configuration-change-hook
(add-hook 'before-save-hook 'ba/org-adjust-tags-column-before-save)
(add-hook 'after-save-hook 'ba/org-adjust-tags-column-after-save)
(add-hook 'org-agenda-mode-hook '(lambda ()
                                  (setq org-agenda-tags-column (- (window-width)))))

; between invoking org-refile and displaying the prompt (which
; triggers window-configuration-change-hook) tags might adjust,
; which invalidates the org-refile cache
(defadvice org-refile (around org-refile-disable-adjust-tags)
  "Disable dynamically adjusting tags"
  (let ((ba/org-adjust-tags-column nil))
(ad-activate 'org-refile)

Use an "attach" link type to open files without worrying about their location

– Darlan Cavalcante Moreira

In the setup part in my org-files I put:

#+beginsrc org ,#+LINK: attach elisp:(org-open-file (org-attach-expand "%s")) #+endsrc org

Now I can use the "attach" link type, but org will ask me if I want to allow executing the elisp code. To avoid this you can even set org-confirm-elisp-link-function to nil (I don't like this because it allows any elisp code in links) or you can set org-confirm-elisp-link-not-regexp appropriately.

In my case I use

(setq org-confirm-elisp-link-not-regexp "org-open-file")

This works very well.

Org Agenda and Task Management

Make it easier to set org-agenda-files from multiple directories

  • Matt Lundin
(defun my-org-list-files (dirs ext)
  "Function to create list of org files in multiple subdirectories.
This can be called to generate a list of files for
org-agenda-files or org-refile-targets.

DIRS is a list of directories.

EXT is a list of the extensions of files to be included."
  (let ((dirs (if (listp dirs)
                (list dirs)))
        (ext (if (listp ext)
               (list ext)))
     (lambda (x)
        (lambda (y)
          (setq files
                (append files
                         (concat (file-name-as-directory x) "*" y)))))
     (lambda (x)
       (when (or (string-match "/.#" x)
                 (string-match "#$" x))
         (setq files (delete x files))))

(defvar my-org-agenda-directories '("~/org/")
  "List of directories containing org files.")
(defvar my-org-agenda-extensions '(".org")
  "List of extensions of agenda files")

(setq my-org-agenda-directories '("~/org/" "~/work/"))
(setq my-org-agenda-extensions '(".org" ".ref"))

(defun my-org-set-agenda-files ()
  (setq org-agenda-files (my-org-list-files


The code above will set your "default" agenda files to all files ending in ".org" and ".ref" in the directories "~/org/" and "~/work/". You can change these values by setting the variables my-org-agenda-extensions and my-org-agenda-directories. The function my-org-agenda-files-by-filetag uses these two variables to determine which files to search for filetags (i.e., the larger set from which the subset will be drawn).

You can also easily use my-org-list-files to "mix and match" directories and extensions to generate different lists of agenda files.

Restrict org-agenda-files by filetag

  • Matt Lundin

It is often helpful to limit yourself to a subset of your agenda files. For instance, at work, you might want to see only files related to work (e.g., bugs, clientA, projectxyz, etc.). The FAQ has helpful information on filtering tasks using filetags and custom agenda commands. These solutions, however, require reapplying a filter each time you call the agenda or writing several new custom agenda commands for each context. Another solution is to use directories for different types of tasks and to change your agenda files with a function that sets org-agenda-files to the appropriate directory. But this relies on hard and static boundaries between files.

The following functions allow for a more dynamic approach to selecting a subset of files based on filetags:

(defun my-org-agenda-restrict-files-by-filetag (&optional tag)
  "Restrict org agenda files only to those containing filetag."
  (let* ((tagslist (my-org-get-all-filetags))
         (ftag (or tag
                   (completing-read "Tag: "
                                    (mapcar 'car tagslist)))))
    (org-agenda-remove-restriction-lock 'noupdate)
    (put 'org-agenda-files 'org-restrict (cdr (assoc ftag tagslist)))
    (setq org-agenda-overriding-restriction 'files)))

(defun my-org-get-all-filetags ()
  "Get list of filetags from all default org-files."
  (let ((files org-agenda-files)
        tagslist x)
      (while (setq x (pop files))
        (set-buffer (find-file-noselect x))
         (lambda (y)
           (let ((tagfiles (assoc y tagslist)))
             (if tagfiles
                 (setcdr tagfiles (cons x (cdr tagfiles)))
               (add-to-list 'tagslist (list y x)))))

(defun my-org-get-filetags ()
  "Get list of filetags for current buffer"
  (let ((ftags org-file-tags)
     (lambda (x)
       (org-substring-no-properties x))

Calling my-org-agenda-restrict-files-by-filetag results in a prompt with all filetags in your "normal" agenda files. When you select a tag, org-agenda-files will be restricted to only those files containing the filetag. To release the restriction, type C-c C-x > (org-agenda-remove-restriction-lock).

Highlight the agenda line under cursor

This is useful to make sure what task you are operating on.

(add-hook 'org-agenda-mode-hook '(lambda () (hl-line-mode 1)))

Under XEmacs:

;; hl-line seems to be only for emacs
(require 'highline)
(add-hook 'org-agenda-mode-hook '(lambda () (highline-mode 1)))

;; highline-mode does not work straightaway in tty mode.
;; I use a black background
  '(highline-face ((((type tty) (class color))
                    (:background "white" :foreground "black")))))

Split frame horizontally for agenda

If you would like to split the frame into two side-by-side windows when displaying the agenda, try this hack from Jan Rehders, which uses the `toggle-window-split' from

;; Patch org-mode to use vertical splitting
(defadvice org-prepare-agenda (after org-fix-split)
(ad-activate 'org-prepare-agenda)

Automatically add an appointment when clocking in a task

;; Make sure you have a sensible value for `appt-message-warning-time'
(defvar bzg-org-clock-in-appt-delay 100
  "Number of minutes for setting an appointment by clocking-in")

This function let's you add an appointment for the current entry. This can be useful when you need a reminder.

(defun bzg-org-clock-in-add-appt (&optional n)
  "Add an appointment for the Org entry at point in N minutes."
    (org-back-to-heading t)
    (looking-at org-complex-heading-regexp)
    (let* ((msg (match-string-no-properties 4))
       (ct-time (decode-time))
       (appt-min (+ (cadr ct-time)
            (or n bzg-org-clock-in-appt-delay)))
       (appt-time ; define the time for the appointment
        (progn (setf (cadr ct-time) appt-min) ct-time)))
      (appt-add (format-time-string
         "%H:%M" (apply 'encode-time appt-time)) msg)
      (if (interactive-p) (message "New appointment for %s" msg)))))

You can advise org-clock-in so that C-c C-x C-i will automatically add an appointment:

(defadvice org-clock-in (after org-clock-in-add-appt activate)
  "Add an appointment when clocking a task in."

You may also want to delete the associated appointment when clocking out. This function does this:

(defun bzg-org-clock-out-delete-appt nil
  "When clocking out, delete any associated appointment."
    (org-back-to-heading t)
    (looking-at org-complex-heading-regexp)
    (let* ((msg (match-string-no-properties 4)))
      (setq appt-time-msg-list
        (delete nil
             (lambda (appt)
               (if (not (string-match (regexp-quote msg)
                          (cadr appt))) appt))

And here is the advice for org-clock-out (C-c C-x C-o)

(defadvice org-clock-out (before org-clock-out-delete-appt activate)
  "Delete an appointment when clocking a task out."

IMPORTANT: You can add appointment by clocking in in both an org-mode and an org-agenda-mode buffer. But clocking out from agenda buffer with the advice above will bring an error.

Using external programs for appointments reminders

Read this rich thread from the org-mode list.

Remove from agenda time grid lines that are in an appointment

The agenda shows lines for the time grid. Some people think that these lines are a distraction when there are appointments at those times. You can get rid of the lines which coincide exactly with the beginning of an appointment. Michael Ekstrand has written a piece of advice that also removes lines that are somewhere inside an appointment:

(defun org-time-to-minutes (time)
  "Convert an HHMM time to minutes"
  (+ (* (/ time 100) 60) (% time 100)))

(defun org-time-from-minutes (minutes)
  "Convert a number of minutes to an HHMM time"
  (+ (* (/ minutes 60) 100) (% minutes 60)))

(defadvice org-agenda-add-time-grid-maybe (around mde-org-agenda-grid-tweakify
                                                  (list ndays todayp))
  (if (member 'remove-match (car org-agenda-time-grid))
      (flet ((extract-window
              (let ((start (get-text-property 1 'time-of-day line))
                    (dur (get-text-property 1 'duration line)))
                 ((and start dur)
                  (cons start
                         (+ dur (org-time-to-minutes start)))))
                 (start start)
                 (t nil)))))
        (let* ((windows (delq nil (mapcar 'extract-window list)))
                (list (car org-agenda-time-grid)
                      (cadr org-agenda-time-grid)
                       (lambda (time)
                         (find-if (lambda (w)
                                    (if (numberp w)
                                        (equal w time)
                                      (and (>= time (car w))
                                           (< time (cdr w)))))
                       (caddr org-agenda-time-grid)))))
(ad-activate 'org-agenda-add-time-grid-maybe)

Disable version control for Org mode agenda files

– David Maus

Even if you use Git to track your agenda files you might not need vc-mode to be enabled for these files.

(add-hook 'find-file-hook 'dmj/disable-vc-for-agenda-files-hook)
(defun dmj/disable-vc-for-agenda-files-hook ()
  "Disable vc-mode for Org agenda files."
  (if (and (fboundp 'org-agenda-file-p)
           (org-agenda-file-p (buffer-file-name)))
      (remove-hook 'find-file-hook 'vc-find-file-hook)
    (add-hook 'find-file-hook 'vc-find-file-hook)))

Easy customization of TODO colors

– Ryan C. Thompson

Here is some code I came up with some code to make it easier to customize the colors of various TODO keywords. As long as you just want a different color and nothing else, you can customize the variable org-todo-keyword-faces and use just a string color (i.e. a string of the color name) as the face, and then org-get-todo-face will convert the color to a face, inheriting everything else from the standard org-todo face.

To demonstrate, I currently have org-todo-keyword-faces set to

(("IN PROGRESS" . "dark orange")
 ("WAITING" . "red4")
 ("CANCELED" . "saddle brown"))

Here's the code, in a form you can put in your .emacs

(eval-after-load 'org-faces
    (defcustom org-todo-keyword-faces nil
      "Faces for specific TODO keywords.
This is a list of cons cells, with TODO keywords in the car and
faces in the cdr.  The face can be a symbol, a color, or a
property list of attributes, like (:foreground \"blue\" :weight
bold :underline t)."
      :group 'org-faces
      :group 'org-todo
      :type '(repeat
               (string :tag "Keyword")
               (choice color (sexp :tag "Face")))))))

(eval-after-load 'org
    (defun org-get-todo-face-from-color (color)
      "Returns a specification for a face that inherits from org-todo
 face and has the given color as foreground. Returns nil if
 color is nil."
      (when color
        `(:inherit org-warning :foreground ,color)))

    (defun org-get-todo-face (kwd)
      "Get the right face for a TODO keyword KWD.
If KWD is a number, get the corresponding match group."
      (if (numberp kwd) (setq kwd (match-string kwd)))
      (or (let ((face (cdr (assoc kwd org-todo-keyword-faces))))
            (if (stringp face)
                (org-get-todo-face-from-color face)
          (and (member kwd org-done-keywords) 'org-done)

Add an effort estimate on the fly when clocking in

You can use org-clock-in-prepare-hook to add an effort estimate. This way you can easily have a "tea-timer" for your tasks when they don't already have an effort estimate.

(add-hook 'org-clock-in-prepare-hook

(defun my-org-mode-ask-effort ()
  "Ask for an effort estimate when clocking in."
  (unless (org-entry-get (point) "Effort")
    (let ((effort
        "Effort: "
        (org-entry-get-multivalued-property (point) "Effort"))))
      (unless (equal effort "")
    (org-set-property "Effort" effort)))))

Or you can use a default effort for such a timer:

(add-hook 'org-clock-in-prepare-hook

(defvar org-clock-default-effort "1:00")

(defun my-org-mode-add-default-effort ()
  "Add a default effort estimation."
  (unless (org-entry-get (point) "Effort")
    (org-set-property "Effort" org-clock-default-effort)))

Use idle timer for automatic agenda views

From John Wiegley's mailing list post (March 18, 2010):

I have the following snippet in my .emacs file, which I find very useful. Basically what it does is that if I don't touch my Emacs for 5 minutes, it displays the current agenda. This keeps my tasks "always in mind" whenever I come back to Emacs after doing something else, whereas before I had a tendency to forget that it was there.

(defun jump-to-org-agenda ()
  (let ((buf (get-buffer "*Org Agenda*"))
    (if buf
        (if (setq wind (get-buffer-window buf))
            (select-window wind)
          (if (called-interactively-p)
                (select-window (display-buffer buf t t))
                ;; (org-agenda-redo)
            (with-selected-window (display-buffer buf)
              ;; (org-agenda-redo)
      (call-interactively 'org-agenda-list)))
  ;;(let ((buf (get-buffer "*Calendar*")))
  ;;  (unless (get-buffer-window buf)
  ;;    (org-agenda-goto-calendar)))

(run-with-idle-timer 300 t 'jump-to-org-agenda)

Refresh the agenda view regularly

Hack sent by Kiwon Um:

(defun kiwon/org-agenda-redo-in-other-window ()
  "Call org-agenda-redo function even in the non-agenda buffer."
  (let ((agenda-window (get-buffer-window org-agenda-buffer-name t)))
    (when agenda-window
      (with-selected-window agenda-window (org-agenda-redo)))))
(run-at-time nil 300 'kiwon/org-agenda-redo-in-other-window)

Reschedule agenda items to today with a single command

This was suggested by Carsten in reply to David Abrahams:

(defun org-agenda-reschedule-to-today ()
  (flet ((org-read-date (&rest rest) (current-time)))
    (call-interactively 'org-agenda-schedule)))

Mark subtree DONE along with all subheadings

Bernt Hansen suggested this command:

(defun bh/mark-subtree-done ()
  (let ((limit (point)))
      (while (> (point) limit)
    (org-todo "DONE")
    (outline-previous-visible-heading 1))
      (org-todo "DONE"))))

Then M-x bh/mark-subtree-done.

Mark heading done when all checkboxes are checked.

An item consists of a list with checkboxes. When all of the checkboxes are checked, the item should be considered complete and its TODO state should be automatically changed to DONE. The code below does that. This version is slightly enhanced over the one in the mailing list (see to reset the state back to TODO if a checkbox is unchecked.

Note that the code requires that a checkbox statistics cookie (the [/] or [%] thingie in the headline - see the Checkboxes section in the manual) be present in order for it to work. Note also that it is too dumb to figure out whether the item has a TODO state in the first place: if there is a statistics cookie, a TODO/DONE state will be added willy-nilly any time that the statistics cookie is changed.

;; see
(eval-after-load 'org-list
  '(add-hook 'org-checkbox-statistics-hook (function ndk/checkbox-list-complete)))

(defun ndk/checkbox-list-complete ()
    (org-back-to-heading t)
    (let ((beg (point)) end)
      (setq end (point))
      (goto-char beg)
      (if (re-search-forward "\\[\\([0-9]*%\\)\\]\\|\\[\\([0-9]*\\)/\\([0-9]*\\)\\]" end t)
            (if (match-end 1)
                (if (equal (match-string 1) "100%")
                    ;; all done - do the state change
                    (org-todo 'done)
                  (org-todo 'todo))
              (if (and (> (match-end 2) (match-beginning 2))
                       (equal (match-string 2) (match-string 3)))
                  (org-todo 'done)
                (org-todo 'todo)))))))

Exporting org files

Export Org to Org and handle includes.

Nick Dokos came up with this useful function:

(defun org-to-org-handle-includes ()
  "Copy the contents of the current buffer to OUTFILE,
recursively processing #+INCLUDEs."
  (let* ((s (buffer-string))
     (fname (buffer-file-name))
     (ofname (format "" (file-name-sans-extension fname))))
    (setq result
        (insert s)
    (find-file ofname)
    (delete-region (point-min) (point-max))
    (insert result)

Specifying LaTeX commands to floating environments

The keyword placement can be used to specify placement options to floating environments (like \begin{figure} and \begin{table}}) in LaTeX export. Org passes along everything passed in options as long as there are no spaces. One can take advantage of this to pass other LaTeX commands and have their scope limited to the floating environment.

For example one can set the fontsize of a table different from the default normal size by putting something like \footnotesize right after the placement options. During LaTeX export using the #+ATTR_LaTeX: line below:

#+ATTR_LaTeX: placement=[<options>]\footnotesize

exports the associated floating environment as shown in the following block.


It should be noted that this hack does not work for beamer export of tables since the table environment is not used. As an ugly workaround, one can use the following:

#+LATEX: {\footnotesize
#+ATTR_LaTeX: align=rr
,| some | table |
,| ..   | ..    |
#+LATEX: }

Styling code sections with CSS

Code sections (marked with #+begin_src and #+end_src) are exported to HTML using <pre> tags, and assigned CSS classes by their content type. For example, Perl content will have an opening tag like <pre class"src src-perl">=. You can use those classes to add styling to the output, such as here where a small language tag is added at the top of each kind of code box:

(setq org-export-html-style
 "<style type=\"text/css\">
      .src             { background-color: #F5FFF5; position: relative; overflow: visible; }
      .src:before      { position: absolute; top: -15px; background: #ffffff; padding: 1px; border: 1px solid #000000; font-size: small; }
      .src-sh:before   { content: 'sh'; }
      .src-bash:before { content: 'sh'; }
      .src-R:before    { content: 'R'; }
      .src-perl:before { content: 'Perl'; }
      .src-sql:before  { content: 'SQL'; }
      .example         { background-color: #FFF5F5; }

Additionally, we use color to distinguish code output (the .example class) from input (all the .src-* classes).


How do I preview LaTeX fragments when in a LaTeX source block?

When editing LaTeX source blocks, you may want to preview LaTeX fragments just like in an Org-mode buffer. You can do this by using the usual keybinding C-c C-x C-l after loading this snipped:

(define-key org-src-mode-map "\C-c\C-x\C-l" 'org-edit-preview-latex-fragment)

(defun org-edit-preview-latex-fragment ()
  "Write latex fragment from source to parent buffer and preview it."
  (org-src-in-org-buffer (org-preview-latex-fragment)))

Thanks to Sebastian Hofer for sharing this.

Hacking Org: Working with Org-mode and other Emacs Packages.


Anything users may find the snippet below interesting:

(defvar org-remember-anything
  '((name . "Org Remember")
    (candidates . (lambda () (mapcar 'car org-remember-templates)))
    (action . (lambda (name)
                (let* ((orig-template org-remember-templates)
                        (list (assoc name orig-template))))
                  (call-interactively 'org-remember))))))

You can add it to your 'anything-sources' variable and open remember directly from anything. I imagine this would be more interesting for people with many remember templates, so that you are out of keys to assign those to.

Org-mode and saveplace.el

Fix a problem with saveplace.el putting you back in a folded position:

(add-hook 'org-mode-hook
          (lambda ()
        (when (outline-invisible-p)
        (outline-previous-visible-heading 1)

Using ido-mode for org-refile (and archiving via refile)

First set up ido-mode, for example using:

; use ido mode for completion
(setq ido-everywhere t)
(setq ido-enable-flex-matching t)
(setq ido-max-directory-size 100000)
(ido-mode (quote both))

Now to enable it in org-mode, use the following:

(setq org-completion-use-ido t)
(setq org-refile-use-outline-path nil)
(setq org-refile-allow-creating-parent-nodes 'confirm)

The last line enables the creation of nodes on the fly.

If you refile into files that are not in your agenda file list, you can add them as target like this (replace file1\done, etc with your files):

(setq org-refile-targets '((org-agenda-files :maxlevel . 5) (("~/org/file1_done" "~/org/file2_done") :maxlevel . 5) ))

For refiling it is often not useful to include targets that have a DONE state. It's easy to remove them by using the verify-refile-target hook.

; Exclude DONE state tasks from refile targets; taken from
; added check to only include headlines, e.g. line must have at least one child
(defun my/verify-refile-target ()
  "Exclude todo keywords with a DONE state from refile targets"
  (or (not (member (nth 2 (org-heading-components)) org-done-keywords)))
      (save-excursion (org-goto-first-child))
(setq org-refile-target-verify-function 'my/verify-refile-target)

Now when looking for a refile target, you can use the full power of ido to find them. Ctrl-R can be used to switch between different options that ido offers.

Using ido-completing-read to find attachments

– Matt Lundin.

Org-attach is great for quickly linking files to a project. But if you use org-attach extensively you might find yourself wanting to browse all the files you've attached to org headlines. This is not easy to do manually, since the directories containing the files are not human readable (i.e., they are based on automatically generated ids). Here's some code to browse those files using ido (obviously, you need to be using ido):

(load-library "find-lisp")

;; Adapted from

(defun my-ido-find-org-attach ()
  "Find files in org-attachment directory"
  (let* ((enable-recursive-minibuffers t)
         (files (find-lisp-find-files org-attach-directory "."))
          (mapcar (lambda (x)
                    (cons (file-name-nondirectory x)
          (remove-duplicates (mapcar #'car file-assoc-list)
                             :test #'string=))
         (filename (ido-completing-read "Org attachments: " filename-list nil t))
         (longname (cdr (assoc filename file-assoc-list))))
     (if (file-directory-p longname)
       (file-name-directory longname)))
    (setq ido-exit 'refresh
          ido-text-init ido-text
          ido-rotate-temp t)

(add-hook 'ido-setup-hook 'ido-my-keys)

(defun ido-my-keys ()
  "Add my keybindings for ido."
  (define-key ido-completion-map (kbd "C-;") 'my-ido-find-org-attach))

To browse your org attachments using ido fuzzy matching and/or the completion buffer, invoke ido-find-file as usual (C-x C-f) and then press C-;.

Link to Gnus messages by Message-Id

In a recent thread on the Org-Mode mailing list, there was some discussion about linking to Gnus messages without encoding the folder name in the link. The following code hooks in to the store-link function in Gnus to capture links by Message-Id when in nnml folders, and then provides a link type "mid" which can open this link. The mde-org-gnus-open-message-link function uses the mde-mid-resolve-methods variable to determine what Gnus backends to scan. It will go through them, in order, asking each to locate the message and opening it from the first one that reports success.

It has only been tested with a single nnml backend, so there may be bugs lurking here and there.

The logic for finding the message was adapted from an Emacs Wiki article.

;; Support for saving Gnus messages by Message-ID
(defun mde-org-gnus-save-by-mid ()
  (when (memq major-mode '(gnus-summary-mode gnus-article-mode))
    (when (eq major-mode 'gnus-article-mode)
    (let* ((group gnus-newsgroup-name)
           (method (gnus-find-method-for-group group)))
      (when (eq 'nnml (car method))
        (let* ((article (gnus-summary-article-number))
               (header (gnus-summary-article-header article))
               (from (mail-header-from header))
                  (let ((mid (mail-header-id header)))
                    (if (string-match "<\\(.*\\)>" mid)
                        (match-string 1 mid)
                      (error "Malformed message ID header %s" mid)))))
               (date (mail-header-date header))
               (subject (gnus-summary-subject-string)))
          (org-store-link-props :type "mid" :from from :subject subject
                                :message-id message-id :group group
                                :link (org-make-link "mid:" message-id))
          (apply 'org-store-link-props
                 :description (org-email-link-description)

(defvar mde-mid-resolve-methods '()
  "List of methods to try when resolving message ID's.  For Gnus,
it is a cons of 'gnus and the select (type and name).")
(setq mde-mid-resolve-methods
      '((gnus nnml "")))

(defvar mde-org-gnus-open-level 1
  "Level at which Gnus is started when opening a link")
(defun mde-org-gnus-open-message-link (msgid)
  "Open a message link with Gnus"
  (require 'gnus)
  (require 'org-table)
  (catch 'method-found
    (message "[MID linker] Resolving %s" msgid)
    (dolist (method mde-mid-resolve-methods)
       ((and (eq (car method) 'gnus)
             (eq (cadr method) 'nnml))
        (funcall (cdr (assq 'gnus org-link-frame-setup))
        (when gnus-other-frame-object
          (select-frame gnus-other-frame-object))
        (let* ((msg-info (nnml-find-group-number
                          (concat "<" msgid ">")
                          (cdr method)))
               (group (and msg-info (car msg-info)))
               (message (and msg-info (cdr msg-info)))
               (qname (and group
                           (if (gnus-methods-equal-p
                                (cdr method)
                             (gnus-group-full-name group (cdr method))))))
          (when msg-info
            (gnus-summary-read-group qname nil t)
            (gnus-summary-goto-article message nil t))
          (throw 'method-found t)))
       (t (error "Unknown link type"))))))

(eval-after-load 'org-gnus
     (add-to-list 'org-store-link-functions 'mde-org-gnus-save-by-mid)
     (org-add-link-type "mid" 'mde-org-gnus-open-message-link)))

Store link to a message when sending in Gnus

Ulf Stegemann came up with this solution (see his original message):

(defun ulf-message-send-and-org-gnus-store-link (&optional arg)
  "Send message with `message-send-and-exit' and store org link to message copy.
If multiple groups appear in the Gcc header, the link refers to
the copy in the last group."
  (interactive "P")
    (let ((gcc (car (last
               (mail-fetch-field "gcc" nil t) " ,")))))
          (buf (current-buffer))
          (message-kill-buffer-on-exit nil)
          id to from subject desc link newsgroup xarchive)
        (message-send-and-exit arg)
         ;; gcc group found ...
         (and gcc
                (progn (set-buffer buf)
                       (setq id (org-remove-angle-brackets
                                 (mail-fetch-field "Message-ID")))
                       (setq to (mail-fetch-field "To"))
                       (setq from (mail-fetch-field "From"))
                       (setq subject (mail-fetch-field "Subject"))))
              (org-store-link-props :type "gnus" :from from :subject subject
                                    :message-id id :group gcc :to to)
              (setq desc (org-email-link-description))
              (setq link (org-gnus-article-link
                          gcc newsgroup id xarchive))
              (setq org-stored-links
                    (cons (list link desc) org-stored-links)))
         ;; no gcc group found ...
         (message "Can not create Org link: No Gcc header found."))))))

(define-key message-mode-map [(control c) (control meta c)]

Send html messages and attachments with Wanderlust

– David Maus

Note: The module Org-mime in Org's contrib directory provides similar functionality for both Wanderlust and Gnus. The hack below is still somewhat different: It allows you to toggle sending of html messages within Wanderlust transparently. I.e. html markup of the message body is created right before sending starts.

Send HTML message

Putting the code below in your .emacs adds following four functions:

  • dmj/wl-send-html-message

    Function that does the job: Convert everything between "–text follows this line–" and first mime entity (read: attachment) or end of buffer into html markup using `org-export-region-as-html' and replaces original body with a multipart MIME entity with the plain text version of body and the html markup version. Thus a recipient that prefers html messages can see the html markup, recipients that prefer or depend on plain text can see the plain text.

    Cannot be called interactively: It is hooked into SEMI's `mime-edit-translate-hook' if message should be HTML message.

  • dmj/wl-send-html-message-draft-init

    Cannot be called interactively: It is hooked into WL's `wl-mail-setup-hook' and provides a buffer local variable to toggle.

  • dmj/wl-send-html-message-draft-maybe

    Cannot be called interactively: It is hooked into WL's `wl-draft-send-hook' and hooks `dmj/wl-send-html-message' into `mime-edit-translate-hook' depending on whether HTML message is toggled on or off

  • dmj/wl-send-html-message-toggle

    Toggles sending of HTML message. If toggled on, the letters "HTML" appear in the mode line.

    Call it interactively! Or bind it to a key in `wl-draft-mode'.

If you have to send HTML messages regularly you can set a global variable `dmj/wl-send-html-message-toggled-p' to the string "HTML" to toggle on sending HTML message by default.

The image here shows an example of how the HTML message looks like in Google's web front end. As you can see you have the whole markup of Org at your service: bold, italics, tables, lists…

So even if you feel uncomfortable with sending HTML messages at least you send HTML that looks quite good.

(defun dmj/wl-send-html-message ()
  "Send message as html message.
Convert body of message to html using
  (require 'org)
    (let (beg end html text)
      (goto-char (point-min))
      (re-search-forward "^--text follows this line--$")
      ;; move to beginning of next line
      (beginning-of-line 2)
      (setq beg (point))
      (if (not (re-search-forward "^--\\[\\[" nil t))
          (setq end (point-max))
        ;; line up
        (end-of-line 0)
        (setq end (point)))
      ;; grab body
      (setq text (buffer-substring-no-properties beg end))
      ;; convert to html
        (insert text)
        ;; handle signature
        (when (re-search-backward "^-- \n" nil t)
          ;; preserve link breaks in signature
          (insert "\n#+BEGIN_VERSE\n")
          (goto-char (point-max))
          (insert "\n#+END_VERSE\n")
          ;; grab html
          (setq html (org-export-region-as-html
                      (point-min) (point-max) t 'string))))
      (delete-region beg end)
        "--" "<<alternative>>-{\n"
        "--" "[[text/plain]]\n" text
        "--" "[[text/html]]\n"  html
        "--" "}-<<alternative>>\n")))))

(defun dmj/wl-send-html-message-toggle ()
  "Toggle sending of html message."
  (setq dmj/wl-send-html-message-toggled-p
        (if dmj/wl-send-html-message-toggled-p
            nil "HTML"))
  (message "Sending html message toggled %s"
           (if dmj/wl-send-html-message-toggled-p
               "on" "off")))

(defun dmj/wl-send-html-message-draft-init ()
  "Create buffer local settings for maybe sending html message."
  (unless (boundp 'dmj/wl-send-html-message-toggled-p)
    (setq dmj/wl-send-html-message-toggled-p nil))
  (make-variable-buffer-local 'dmj/wl-send-html-message-toggled-p)
  (add-to-list 'global-mode-string
               '(:eval (if (eq major-mode 'wl-draft-mode)

(defun dmj/wl-send-html-message-maybe ()
  "Maybe send this message as html message.

If buffer local variable `dmj/wl-send-html-message-toggled-p' is
non-nil, add `dmj/wl-send-html-message' to
  (if dmj/wl-send-html-message-toggled-p
      (add-hook 'mime-edit-translate-hook 'dmj/wl-send-html-message)
    (remove-hook 'mime-edit-translate-hook 'dmj/wl-send-html-message)))

(add-hook 'wl-draft-reedit-hook 'dmj/wl-send-html-message-draft-init)
(add-hook 'wl-mail-setup-hook 'dmj/wl-send-html-message-draft-init)
(add-hook 'wl-draft-send-hook 'dmj/wl-send-html-message-maybe)

Attach HTML of region or subtree

Instead of sending a complete HTML message you might only send parts of an Org file as HTML for the poor souls who are plagued with non-proportional fonts in their mail program that messes up pretty ASCII tables.

This short function does the trick: It exports region or subtree to HTML, prefixes it with a MIME entity delimiter and pushes to killring and clipboard. If a region is active, it uses the region, the complete subtree otherwise.

(defun dmj/org-export-region-as-html-attachment (beg end arg)
  "Export region between BEG and END as html attachment.
If BEG and END are not set, use current subtree.  Region or
subtree is exported to html without header and footer, prefixed
with a mime entity string and pushed to clipboard and killring.
When called with prefix, mime entity is not marked as
  (interactive "r\nP")
    (let* ((beg (if (region-active-p) (region-beginning)
           (end (if (region-active-p) (region-end)
           (html (concat "--[[text/html"
                         (if arg "" "\nContent-Disposition: attachment")
                         (org-export-region-as-html beg end t 'string))))
      (when (fboundp 'x-set-selection)
        (ignore-errors (x-set-selection 'PRIMARY html))
        (ignore-errors (x-set-selection 'CLIPBOARD html)))
      (message "html export done, pushed to kill ring and clipboard"))))

Adopting for Gnus

The whole magic lies in the special strings that mark a HTML attachment. So you might just have to find out what these special strings are in message-mode and modify the functions accordingly.

Add sunrise/sunset times to the agenda.

– Nick Dokos

The diary package provides the function diary-sunrise-sunset which can be used in a diary s-expression in some agenda file like this:


Seb Vauban asked if it is possible to put sunrise and sunset in separate lines. Here is a hack to do that. It adds two functions (they have to be available before the agenda is shown, so I add them early in my org-config file which is sourced from .emacs, but you'll have to suit yourself here) that just parse the output of diary-sunrise-sunset, instead of doing the right thing which would be to take advantage of the data structures that diary/solar.el provides. In short, a hack - so perfectly suited for inclusion here :-)

The functions (and latitude/longitude settings which you have to modify for your location) are as follows:

(setq calendar-latitude 48.2)
(setq calendar-longitude 16.4)
(setq calendar-location-name "Vienna, Austria")

(autoload 'solar-sunrise-sunset "solar.el")
(autoload 'solar-time-string "solar.el")
(defun diary-sunrise ()
  "Local time of sunrise as a diary entry.
The diary entry can contain `%s' which will be replaced with
  (let ((l (solar-sunrise-sunset date)))
    (when (car l)
       (if (string= entry "")
         (format entry (eval calendar-location-name))) " "
         (solar-time-string (caar l) nil)))))

(defun diary-sunset ()
  "Local time of sunset as a diary entry.
The diary entry can contain `%s' which will be replaced with
  (let ((l (solar-sunrise-sunset date)))
    (when (cadr l)
       (if (string= entry "")
         (format entry (eval calendar-location-name))) " "
         (solar-time-string (caadr l) nil)))))

You also need to add a couple of diary s-expressions in one of your agenda files:

%%(diary-sunrise)Sunrise in %s

This will show sunrise with the location and sunset without it.

The thread on the mailing list that started this can be found here. In comparison to the version posted on the mailing list, this one gets rid of the timezone information and can show the location.

Add lunar phases to the agenda.

– Rüdiger

Emacs comes with lunar.el to display the lunar phases (M-x lunar-phases). This can be used to display lunar phases in the agenda display with the following function:

(require 'cl-lib)

(org-no-warnings (defvar date))
(defun org-lunar-phases ()
  "Show lunar phase in Agenda buffer."
  (require 'lunar)
  (let* ((phase-list (lunar-phase-list (nth 0 date) (nth 2 date)))
         (phase (cl-find-if (lambda (phase) (equal (car phase) date))
    (when phase
      (setq ret (concat (lunar-phase-name (nth 2 phase)) " "
                        (substring (nth 1 phase) 0 5))))))

Add the following line to an agenda file:

#+beginsrc org-mode

Lunar phase


This should display an entry on new moon, first/last quarter moon, and on full moon. You can customize the entries by customizing lunar-phase-names.

E.g., to add Unicode symbols:

(setq lunar-phase-names
      '("● New Moon" ; Unicode symbol: 🌑 Use full circle as fallback
        "☽ First Quarter Moon"
        "○ Full Moon" ; Unicode symbol: 🌕 Use empty circle as fallback
        "☾ Last Quarter Moon"))

Unicode 6 even provides symbols for the Moon with nice faces. But those symbols are currently barely supported in fonts. See Astronomical symbols on Wikipedia.

Export BBDB contacts to org-contacts.el

Calculating date differences - how to write a simple elisp function

Alexander Wingård asked how to calculate the number of days between a time stamp in his org file and today (see Although the resulting answer is probably not of general interest, the method might be useful to a budding Elisp programmer.

Alexander started from an already existing org function, org-evaluate-time-range. When this function is called in the context of a time range (two time stamps separated by "--"), it calculates the number of days between the two dates and outputs the result in Emacs's echo area. What he wanted was a similar function that, when called from the context of a single time stamp, would calculate the number of days between the date in the time stamp and today. The result should go to the same place: Emacs's echo area.

The solution presented in the mail thread is as follows:

(defun aw/org-evaluate-time-range (&optional to-buffer)
  (if (org-at-date-range-p t)
      (org-evaluate-time-range to-buffer)
    ;; otherwise, make a time range in a temp buffer and run o-e-t-r there
    (let ((headline (buffer-substring (point-at-bol) (point-at-eol))))
    (insert headline)
    (goto-char (point-at-bol))
    (re-search-forward org-ts-regexp (point-at-eol) t)
    (if (not (org-at-timestamp-p t))
        (error "No timestamp here"))
    (goto-char (match-beginning 0))
    (org-insert-time-stamp (current-time) nil nil)
    (insert "--")
    (org-evaluate-time-range to-buffer)))))

The function assumes that point is on some line with some time stamp (or a date range) in it. Note that org-evaluate-time-range does not care whether the first date is earlier than the second: it will always output the number of days between the earlier date and the later date.

As stated before, the function itself is of limited interest (although it satisfied Alexander's need).The method used might be of wider interest however, so here is a short explanation.

The idea is that we want org-evaluate-time-range to do all the heavy lifting, but that function requires that it be in a date-range context. So the function first checks whether it's in a date range context already: if so, it calls org-evaluate-time-range directly to do the work. The trick now is to arrange things so we can call this same function in the case where we do not have a date range context. In that case, we manufacture one: we create a temporary buffer, copy the line with the purported time stamp to the temp buffer, find the time stamp (signal an error if no time stamp is found) and insert a new time stamp with the current time before the existing time stamp, followed by "--": voilà, we now have a time range on which we can apply our old friend org-evaluate-time-range to produce the answer. Because of the above-mentioned property of org-evaluate-time-range, it does not matter if the existing time stamp is earlier or later than the current time: the correct number of days is output.

Note that at the end of the call to with-temp-buffer, the temporary buffer goes away. It was just used as a scratch pad for the function to do some figuring.

The idea of using a temp buffer as a scratch pad has wide applicability in Emacs programming. The rest of the work is knowing enough about facilities provided by Emacs (e.g. regexp searching) and by Org (e.g. checking for time stamps and generating a time stamp) so that you don't reinvent the wheel, and impedance-matching between the various pieces.

ibuffer and org files

Neil Smithline posted this snippet to let you browse org files with ibuffer:

(require 'ibuffer)

(defun org-ibuffer ()
  "Open an `ibuffer' window showing only `org-mode' buffers."
  (ibuffer nil "*Org Buffers*" '((used-mode . org-mode))))

Enable org-mode links in other modes

Sean O'Halpin wrote a minor mode for this, please check it here.

See the relevant discussion here.

poporg.el: edit comments in org-mode

poporg.el is a library by François Pinard which lets you edit comments from your code using a separate org-mode buffer.

Convert a .csv file to an Org-mode table

Nicolas Richard has a nice recipe using the pcsv library (available from the Marmelade ELPA repository):

(defun yf/lisp-table-to-org-table (table &optional function)
  "Convert a lisp table to `org-mode' syntax, applying FUNCTION to each of its elements.
The elements should not have any more newlines in them after
applying FUNCTION ; the default converts them to spaces. Return
value is a string containg the unaligned `org-mode' table."
  (unless (functionp function)
    (setq function (lambda (x) (replace-regexp-in-string "\n" " " x))))
  (mapconcat (lambda (x)                ; x is a line.
               (concat "| " (mapconcat function x " | ") " |"))
             table "\n"))

(defun yf/csv-to-table (beg end)
"Convert a csv file to an `org-mode' table."
  (interactive "r")
  (require 'pcsv)
  (insert (yf/lisp-table-to-org-table (pcsv-parse-region beg end)))
  (delete-region beg end)

Hacking Org: Working with Org-mode and External Programs.

Use Org-mode with Screen [Andrew Hyatt]

"The general idea is that you start a task in which all the work will take place in a shell. This usually is not a leaf-task for me, but usually the parent of a leaf task. From a task in your org-file, M-x ash-org-screen will prompt for the name of a session. Give it a name, and it will insert a link. Open the link at any time to go the screen session containing your work!"

(require 'term)

(defun ash-org-goto-screen (name)
  "Open the screen with the specified name in the window"
  (interactive "MScreen name: ")
  (let ((screen-buffer-name (ash-org-screen-buffer-name name)))
    (if (member screen-buffer-name
                (mapcar 'buffer-name (buffer-list)))
        (switch-to-buffer screen-buffer-name)
      (switch-to-buffer (ash-org-screen-helper name "-dr")))))

(defun ash-org-screen-buffer-name (name)
  "Returns the buffer name corresponding to the screen name given."
  (concat "*screen " name "*"))

(defun ash-org-screen-helper (name arg)
  ;; Pick the name of the new buffer.
  (let ((term-ansi-buffer-name
      (ash-org-screen-buffer-name name))))
    (setq term-ansi-buffer-name
       term-ansi-buffer-name "/usr/bin/screen" nil arg name))
    (set-buffer term-ansi-buffer-name)
    (term-set-escape-char ?\C-x)

(defun ash-org-screen (name)
  "Start a screen session with name"
  (interactive "MScreen name: ")
    (ash-org-screen-helper name "-S"))
  (insert-string (concat "[[screen:" name "]]")))

;; And don't forget to add ("screen" . "elisp:(ash-org-goto-screen
;; \"%s\")") to org-link-abbrev-alist.

Org Agenda + Appt + Zenity

Russell Adams posted this setup on the list. It makes sure your agenda appointments are known by Emacs, and it displays warnings in a zenity popup window.

; For org appointment reminders

;; Get appointments for today
(defun my-org-agenda-to-appt ()
  (setq appt-time-msg-list nil)
  (let ((org-deadline-warning-days 0))    ;; will be automatic in org 5.23

;; Run once, activate and schedule refresh
(appt-activate t)
(run-at-time "24:01" nil 'my-org-agenda-to-appt)

; 5 minute warnings
(setq appt-message-warning-time 15)
(setq appt-display-interval 5)

; Update appt each time agenda opened.
(add-hook 'org-finalize-agenda-hook 'my-org-agenda-to-appt)

; Setup zenify, we tell appt to use window, and replace default function
(setq appt-display-format 'window)
(setq appt-disp-window-function (function my-appt-disp-window))

(defun my-appt-disp-window (min-to-app new-time msg)
  (save-window-excursion (shell-command (concat
    "/usr/bin/zenity --info --title='Appointment' --text='"
    msg "' &") nil nil)))

Org and appointment notifications on Mac OS 10.8

Sarah Bagby posted some code on how to get appointments notifications on Mac OS 10.8 with terminal-notifier.

Org-Mode + gnome-osd

Richard Riley uses gnome-osd in interaction with Org-Mode to display appointments. You can look at the code on the emacswiki.

txt2org convert text data to org-mode tables

From Eric Schulte

I often find it useful to generate Org-mode tables on the command line from tab-separated data. The following awk script makes this easy to do. Text data is read from STDIN on a pipe and any command line arguments are interpreted as rows at which to insert hlines.

Here are two usage examples.

  1. running the following
    $ cat <<EOF|~/src/config/bin/txt2org
    one 1
    two 2
    three 3
    twenty 20

    results in

    |    one |  1 |
    |    two |  2 |
    |  three |  3 |
    | twenty | 20 |
  2. and the following (notice the command line argument)
    $ cat <<EOF|~/src/config/bin/txt2org 1
    strings numbers                                          
    one 1
    two 2
    three 3
    twenty 20

    results in

    | strings | numbers |
    |     one |       1 |
    |     two |       2 |
    |   three |       3 |
    |  twenty |      20 |

Here is the script itself

#!/usr/bin/gawk -f
# Read tab separated data from STDIN and output an Org-mode table.
# Optional command line arguments specify row numbers at which to
# insert hlines.
    for(i=1; i<ARGC; i++){
        hlines[ARGV[i]+1]=1; ARGV[i] = "-"; } }

    if(NF > max_nf){ max_nf = NF; };
    for(f=1; f<=NF; f++){
        if(length($f) > lengths[f]){ lengths[f] = length($f); };
        row[NR][f]=$f; } }

    for(f=1; f<=max_nf; f++){
        for(i=0; i<(lengths[f] + 2); i++){ hline_str=hline_str "-"; }
        if( f != max_nf){ hline_str=hline_str "+"; }
        else            { hline_str=hline_str "|"; } }

    for(r=1; r<=NR; r++){ # rows
        if(hlines[r] == 1){ print hline_str; }
        printf "|";
        for(f=1; f<=max_nf; f++){ # columns
            cell=row[r][f]; padding=""
            for(i=0; i<(lengths[f] - length(cell)); i++){ padding=padding " "; }
            # for now just print everything right-aligned
            # if(cell ~ /[0-9.]/){ printf " %s%s |", cell, padding; }
            # else{                printf " %s%s |", padding, cell; }
            printf " %s%s |", padding, cell; }
        printf "\n"; }

    if(hlines[NR+1]){ print hline_str; } }


From Detlef Steuer

Remind ( is a very powerful command line calendaring program. Its features supersede the possibilities of orgmode in the area of date specifying, so that I want to use it combined with orgmode.

Using the script below I'm able use remind and incorporate its output in my agenda views. The default of using 13 months look ahead is easily changed. It just happens I sometimes like to look a year into the future. :-)

Useful webjumps for conkeror

If you are using the conkeror browser, maybe you want to put this into your ~/.conkerorrc file:

define_webjump("orglist", "");
define_webjump("worg", "");

It creates two webjumps for easily searching the Worg website and the Org-mode mailing list.

Use MathJax for HTML export without requiring JavaScript

As of 2010-08-14, MathJax is the default method used to export math to HTML.

If you like the results but do not want JavaScript in the exported pages, check out Static MathJax, a XULRunner application which generates a static HTML file from the exported version. It can also embed all referenced fonts within the HTML file itself, so there are no dependencies to external files.

The download archive contains an elisp file which integrates it into the Org export process (configurable per file with a "#+StaticMathJax:" line).

Read and the comments in org-static-mathjax.el for usage instructions.

Search Org files using lgrep

Matt Lundin suggests this:

  (defun my-org-grep (search &optional context)
    "Search for word in org files.

Prefix argument determines number of lines."
    (interactive "sSearch for: \nP")
    (let ((grep-find-ignored-files '("#*" ".#*"))
      (grep-template (concat "grep <X> -i -nH "
                 (when context
                   (concat "-C" (number-to-string context)))
                 " -e <R> <F>")))
      (lgrep search "*org*" "/home/matt/org/")))

  (global-set-key (kbd "<f8>") 'my-org-grep)

Automatic screenshot insertion

Suggested by Russell Adams

(defun my-org-screenshot ()
  "Take a screenshot into a time stamped unique-named file in the
same directory as the org-buffer and insert a link to this file."
  (setq filename
          (concat (buffer-file-name)
                  (format-time-string "%Y%m%d_%H%M%S_")) ) ".png"))
  (call-process "import" nil nil nil filename)
  (insert (concat "[[" filename "]]"))

Capture invitations/appointments from MS Exchange emails

Dirk-Jan C.Binnema provided code to do this. Please check org-exchange-capture.el

Audio/video file playback within org mode

Paul Sexton provided code that makes file: links to audio or video files (MP3, WAV, OGG, AVI, MPG, et cetera) play those files using the Bongo Emacs media player library. The user can pause, skip forward and backward in the track, and so on from without leaving Emacs. Links can also contain a time after a double colon – when this is present, playback will begin at that position in the track.

See the file org-player.el

Under X11 Keep a window with the current agenda items at all time

I struggle to keep (in emacs) a window with the agenda at all times. For a long time I have wanted a sticky window that keeps this information, and then use my window manager to place it and remove its decorations (I can also force its placement in the stack: top always, for example).

I wrote a small program in qt that simply monitors an HTML file and displays it. Nothing more. It does the work for me, and maybe somebody else will find it useful. It relies on exporting the agenda as HTML every time the org file is saved, and then this little program displays the html file. The window manager is responsible of removing decorations, making it sticky, and placing it in same place always.

Here is a screenshot (see window to the bottom right). The decorations are removed by the window manager:


Here is the code. As I said, very, very simple, but maybe somebody will find if useful.

–daniel german

Script (thru procmail) to output emails to an Org file

Tycho Garen sent this:

I've [...] created some procmail and shell glue that takes emails and
inserts them into an org-file so that I can capture stuff on the go using
the email program.

Everything is documented here.

Save File With Different Format for Headings (fileconversion)

Using hooks and on the fly

  • when writing a buffer to the file replace the leading stars from headings with a file char
  • when reading a file into the buffer replace the file chars with leading stars for headings

To change to save an Org file in one of the formats or back just add or remove the keyword in the STARTUP line and save.

Now you can also change to Fundamental mode to see how the file looks like on the level of the file, go back to Org mode, reenter Org mode or change to any other major mode and the conversion gets done whenever necessary.

Headings Without Leading Stars (hidestarsfile and nbspstarsfile)

This is like "a cleaner outline view":

Example of the file content first with leading stars as usual and below without leading stars through "#+STARTUP: odd hidestars hidestarsfile":

#+STARTUP: odd hidestars
***** TODO section
******* subsection
********* subsubsec
          - bla bla
***** section
      - bla bla
******* subsection
#+STARTUP: odd hidestars hidestarsfile
    * TODO section
      * subsection
        * subsubsec
          - bla bla
    * section
      - bla bla
      * subsection

The latter is convenient for better human readability when an Org file, additionally to Emacs, is read with a file viewer or, for smaller edits, with an editor not capable of the Org file format.

hidestarsfile is a hack and can not become part of the Org core:

  • An Org file with hidestarsfile can not contain list items with a star as bullet due to the syntax conflict at read time. Mark E. Shoulson suggested to use the non-breaking space which is now implemented in fileconversion as nbspstarsfile as an alternative for hidestarsfile. Although I don't recommend it because an editor like typically e. g. Emacs may render the non-breaking space differently from the space 0x20.
  • An Org file with hidestarsfile can almost not be edited with an Org mode without added functionality of hidestarsfile as long as the file is not converted back.

Headings in Markdown Format (markdownstarsfile)

For "oddeven" you can use markdownstarsfile to be readable or even basically editable with Markdown (does not make much sense with "odd", see org-convert-to-odd-levels and org-convert-to-oddeven-levels for how to convert).

Example of the file content:

#+STARTUP: oddeven markdownstarsfile
# section level 1
  1. first item of numbered list (same format in Org and Markdown)
## section level 2
   - first item of unordered list (same format in Org and Markdown)
### section level 3
    + first item of unordered list (same format in Org and Markdown)
#### section level 4
     * first item of unordered list (same format in Org and Markdown)
     * avoid this item type to be compatible with Org hidestarsfile

An Org file with markdownstarsfile can not contain code comment lines prefixed with "#", even not when within source blocks.

emacs-lisp code

;; - fileconversion version 0.7
;; - DISCLAIMER: Make a backup of your Org files before using
;;   my-org-fileconv-*.
;; - supported formats: hidestarsfile, markdownstarsfile

;; design summary: fileconversion is a round robin of two states
;; linked by two actions:
;; - state v-org-fileconv-level-org-p is nil: the level is “file”
;;   (encoded)
;; - action f-org-fileconv-decode: replace file char with “*”
;; - state v-org-fileconv-level-org-p is t: the level is “Org”
;;   (decoded)
;; - action f-org-fileconv-encode: replace “*” with file char
;; naming convention of prefix:
;; - f-[...]: “my function”, instead of the unspecific prefix “my-”
;; - v-[...]: “my variable”, instead of the unspecific prefix “my-”

(defvar v-org-fileconv-level-org-p nil
  "Whether level of buffer is Org or only file.
nil means the level is file (encoded), non-nil means the level is Org
(make-variable-buffer-local 'v-org-fileconv-level-org-p)
;; survive a change of major mode that does kill-all-local-variables,
;; e. g. when reentering Org mode through “C-c C-c” on a STARTUP line
(put 'v-org-fileconv-level-org-p 'permanent-local t)

(add-hook 'org-mode-hook 'f-org-fileconv-init
          ;; _append_ to hook to have a higher chance that a message
          ;; from this function will be visible as the last message in
          ;; the minibuffer
          ;; hook addition globally

(defun f-org-fileconv-init ()
  ;; instrument only when converting really from/to an Org _file_, not
  ;; e. g. for a temp Org buffer unrelated to a file like used e. g.
  ;; when calling the old Org exporter
  (when (buffer-file-name)
    (message "INF: f-org-fileconv-init, buffer: %s" (buffer-name))
    ;; the hooks are not permanent-local, this way and as needed they
    ;; will disappear when the major mode of the buffer changes
    (add-hook 'change-major-mode-hook 'f-org-fileconv-encode nil
              ;; hook addition limited to buffer locally
    (add-hook 'before-save-hook 'f-org-fileconv-encode nil
              ;; hook addition limited to buffer locally
    (add-hook 'after-save-hook 'f-org-fileconv-decode nil
              ;; hook addition limited to buffer locally

(defun f-org-fileconv-re ()
  "Check whether there is a STARTUP line for fileconversion.
If found then return the expressions required for the conversion."
    (goto-char (point-min))  ;; beginning-of-buffer not allowed
    (let (re-list (count 0))
      (while (re-search-forward "^#\\+STARTUP:" nil t)
        ;; #+STARTUP: hidestarsfile
        (when (string-match-p "\\bhidestarsfile\\b"
                              (thing-at-point 'line))
          ;; exclude e. g.:
          ;; - line starting with star for bold emphasis
          ;; - line of stars to underline section title in loosely
          ;;   quoted ASCII style (star at end of line)
          (setq re-list '("\\(\\* \\)"  ; common-re
                          ?\ ))         ; file-char
          (setq count (1+ count)))
        ;; #+STARTUP: nbspstarsfile
        (when (string-match-p "\\bnbspstarsfile\\b"
                              (thing-at-point 'line))
          (setq re-list '("\\(\\* \\)"  ; common-re
                          ?\xa0))       ; file-char non-breaking space
          (setq count (1+ count)))
        ;; #+STARTUP: markdownstarsfile
        (when (string-match-p "\\bmarkdownstarsfile\\b"
                              (thing-at-point 'line))
          ;; exclude e. g.:
          ;; - #STARTUP:
          (setq re-list '("\\( \\)"  ; common-re
                          ?#))       ; file-char
          (setq count (1+ count))))
      (when (> count 1)
        (error "More than one fileconversion found"))

(defun f-org-fileconv-decode ()
  "In headings replace file char with '*'."
  (let ((re-list (f-org-fileconv-re)))
    (when (and re-list (not v-org-fileconv-level-org-p))
      ;; no `save-excursion' to be able to keep point in case of error
      (let* ((common-re (nth 0 re-list))
             (file-char (nth 1 re-list))
             (file-re   (concat "^" (string file-char) "+" common-re))
             (org-re    (concat "^\\*+" common-re))
             (p         (point)))
        (goto-char (point-min))  ;; beginning-of-buffer not allowed
        ;; syntax check
        (when (re-search-forward org-re nil t)
          (goto-char (match-beginning 0))
          (error "Org fileconversion dec: syntax conflict at point"))
        (goto-char (point-min))  ;; beginning-of-buffer not allowed
        ;; substitution
          (while (re-search-forward file-re nil t)
            (goto-char (match-beginning 0))
            ;; faster than a lisp call of insert and delete on each
            ;; single char
            (setq len (- (match-beginning 1) (match-beginning 0)))
            (insert-char ?* len)
            (delete-char len)))
        (goto-char p))))

        ;; notes for ediff when only one file has fileconversion:
        ;; - The changes to the buffer with fileconversion until here
        ;;   are not regarded by ediff-files because the first call to
        ;;   diff is made with the bare files directly. Only
        ;;   ediff-update-diffs and ediff-buffers write the decoded
        ;;   buffers to temp files and then call diff with them.
        ;; - Workarounds (choose one):
        ;;   - after ediff-files first do a "!" (ediff-update-diffs)
        ;;     in the "*Ediff Control Panel*"
        ;;   - instead of using ediff-files first open the files and
        ;;     then run ediff-buffers (better for e. g. a script that
        ;;     takes two files as arguments and uses "emacs --eval")

  ;; the level is Org most of all when no fileconversion is in effect
  (setq v-org-fileconv-level-org-p t))

(defun f-org-fileconv-encode ()
  "In headings replace '*' with file char."
  (let ((re-list (f-org-fileconv-re)))
    (when (and re-list v-org-fileconv-level-org-p)
      ;; no `save-excursion' to be able to keep point in case of error
      (let* ((common-re (nth 0 re-list))
             (file-char (nth 1 re-list))
             (file-re   (concat "^" (string file-char) "+" common-re))
             (org-re    (concat "^\\*+" common-re))
             (p         (point)))
        (goto-char (point-min))  ;; beginning-of-buffer not allowed
        ;; syntax check
        (when (re-search-forward file-re nil t)
          (goto-char (match-beginning 0))
          (error "Org fileconversion enc: syntax conflict at point"))
        (goto-char (point-min))  ;; beginning-of-buffer not allowed
        ;; substitution
          (while (re-search-forward org-re nil t)
            (goto-char (match-beginning 0))
            ;; faster than a lisp call of insert and delete on each
            ;; single char
            (setq len (- (match-beginning 1) (match-beginning 0)))
            (insert-char file-char len)
            (delete-char len)))
        (goto-char p)
        (setq v-org-fileconv-level-org-p nil))))
  nil)  ;; for the hook

Michael Brand

Meaningful diff for org files in a git repository

Since most diff utilities are primarily meant for source code, it is difficult to read diffs of text files like .org files easily. If you version your org directory with a SCM like git you will know what I mean. However for git, there is a way around. You can use gitattributes to define a custom diff driver for org files. Then a regular expression can be used to configure how the diff driver recognises a "function".

Put the following in your <org_dir>/.gitattributes.

*.org   diff=org

Then put the following lines in <org_dir>/.git/config

[diff "org"]
    xfuncname = "^(\\*+ [a-zA-Z0-9]+.+)$"

This will let you see diffs for org files with each hunk identified by the unmodified headline closest to the changes. After the configuration a diff should look something like the example below.

diff --git a/ b/
index a0672ea..92a08f7 100644
--- a/
+++ b/
@@ -2495,6 +2495,22 @@ ** Script (thru procmail) to output emails to an Org file
 Everything is documented [[][here]].
+** Meaningful diff for org files in a git repository
+Since most diff utilities are primarily meant for source code, it is
+difficult to read diffs of text files like files easily. If you
+version your org directory with a SCM like git you will know what I
+mean. However for git, there is a way around. You can use
+=gitattributes= to define a custom diff driver for org files. Then a
+regular expression can be used to configure how the diff driver
+recognises a "function".
+Put the following in your =<org_dir>/.gitattributes=.
+: *.org	diff=org
+Then put the following lines in =<org_dir>/.git/config=
+: [diff "org"]
+: 	xfuncname = "^(\\*+ [a-zA-Z0-9]+.+)$"
 * Musings
 ** Cooking?  Brewing?

Opening devonthink links

John Wiegley wrote org-devonthink.el, which lets you handle devonthink links from org-mode.


Cooking? Brewing?

See this message from Erik Hetzner:

It currently does metric/english conversion, and a few other tricks. Basically I just use calc’s units code. I think scaling recipes, or turning percentages into weights would be pretty easy.

There is also, for those interested:

for brewing beer. This is again, mostly just calc functions, including hydrometer correction, abv calculation, priming sugar for a given CO2 volume, etc. More integration with org-mode should be possible: for instance it would be nice to be able to use a lookup table (of ingredients) to calculate target original gravity, IBUs, etc.

Author: Worg people

Created: 2013-08-08 tor 11:56

Emacs 24.3.50 (Org mode 8.0.7)

Validate XHTML 1.0