Howto define formula for table regions

Howto define formula for table regions

[Torsten Wagner]

Hi,

I am just playing around with the table function of org-mode.
I tried to create a little table to track the financial status of a
project. I thought it would be nice to distinguish between entries which
really had billed the account already and entries which will bill the
account in the future. This would allow me to track and compare the
current amount of my project account and in addition makes sure I will
not overdraw my account by missing an outstanding transaction.

Thus, initially my idea was to create a table like this:

|-----+------+-----------------------+--------+------------+------|
| nr. | date | title                 | number |     amount |  sum |
|-----+------+-----------------------+--------+------------+------|
|   1 |      | inital fund           |      1 |       1000 | 1000 |
|   2 |      | 1. payment to crew    |      3 |       -150 | -450 |
|   3 |      | 1. order              |      1 |        -50 |  -50 |
|   4 |      |                       |        |            |      |
|-----+------+-----------------------+--------+------------+------|
|     |      | balance on account    |        |            |  500 |
|-----+------+-----------------------+--------+------------+------|
|   1 |      | outstanding order     |      1 |       -100 | -100 |
|   2 |      | 2. payment to crew    |      3 |       -100 | -300 |
|-----+------+-----------------------+--------+------------+------|
|     |      | balance - outstanding |        |            |  100 |
|-----+------+-----------------------+--------+------------+------|

Now I would like to add formulas for:
1. Increasing the number of the first column ("Nr."), (entry in the FAQ,
I know), whenever I add a new row. However do this for two blocks and
leave row 6 and the last row empty.
2. Place the multiplication of "number" with "amount" in the sum column
3. Add the total sums of the two blocks in row 6 and the last row
(taking into account that the table will grow over time)

I know about column formulas and field formulas but I did not find a way
to do something like "for each field in column X between row a and row b
calculate ..."

I thought I could use the hline notation to encapsulate a undefined
amount of rows in between, however this doesn't work since every formula
does only refer to one field or to an entire column.

What I am looking for is something like:

$6@II+1..$6@III-1 = $4@0 * $5@0

or in text:

for each field in column 6 starting with the first row below the second
hline and end with row above the third hline multiply the column 4 and
column 5 of the current row.

Is there any other way to achieve this?
Or does this make much enough sense to implement it ;)
I guess there might be a lisp-way but I think a more intuitive way for
non-lisper might be useful.

As a kind of extra....for the very org-mode pros on this list
If I fill dates into the date column, a mechanism which moves
automatically (in the same way as I refresh calculation of formulas),
the rows from the second block into the first block when the date passed
would be some nice gadget


Best regards

Torsten

CC. The manual says, the very first hline will not be counted for the
hline referencing notation. However, as far as I can see it does count
in both formulas and by indicating it with C-} (Org-mode version
6.36trans (release_6.36.437.g05b7e))

Re: Howto define formula for table regions

[Łukasz Stelmach]

Torsten Wagner <torsten.wagner@gmail.com> writes:

> I am just playing around with the table function of org-mode.
> I tried to create a little table to track the financial status of a
> project. I thought it would be nice to distinguish between entries which
> really had billed the account already and entries which will bill the
> account in the future. This would allow me to track and compare the
> current amount of my project account and in addition makes sure I will
> not overdraw my account by missing an outstanding transaction.
>
> Thus, initially my idea was to create a table like this:
>
> |-----+------+-----------------------+--------+------------+------|
> | nr. | date | title                 | number |     amount |  sum |
> |-----+------+-----------------------+--------+------------+------|
> |   1 |      | inital fund           |      1 |       1000 | 1000 |

OK. Let's rebuild the table a bit

|---+------+------+-----------------------+--------+--------+--------|
| ! |   nr | date | title                 | number | amount |    sum |
|---+------+------+-----------------------+--------+--------+--------|
| # |    1 |      | inital fund           |      1 |   1000 |   1000 |
| # |    2 |      | 1. payment to crew    |      3 |   -150 |   -450 |
| # |    3 |      | 1. order              |      1 |    -50 |    -50 |
| # |    4 |      |                       |        |        |      0 |
|---+------+------+-----------------------+--------+--------+--------|
|   |      |      | balance on account    |        |        |    500 |
| ^ |      |      |                       |        |        |    bal |
|---+------+------+-----------------------+--------+--------+--------|
| # |    1 |      | outstanding order     |      1 |   -100 |   -100 |
| # |    2 |      | 2. payment to crew    |      3 |   -100 |   -300 |
|---+------+------+-----------------------+--------+--------+--------|
|   |      |      | balance - outstanding |        |        |    100 |
| ^ |      |      |                       |        | moveup | outbal |
|---+------+------+-----------------------+--------+--------+--------|

The special characters in the first columns make some difference. With
`!' you get names for columns you may use in your formulae (you'll see
below). `#' forces recalculation of selected rows upon any Tab or Return
keypress, and `^' gives names to cells in the row above. You can find
more about the special characters for the first column in
<info:(org)Advanced features> chapter. These changes aren't a must but
they make formulae a bit more readable.


> Now I would like to add formulas for:
> 1. Increasing the number of the first column ("Nr."), (entry in the FAQ,
> I know), whenever I add a new row. However do this for two blocks and
> leave row 6 and the last row empty.

Be not afraid, it's easier than you think. Simply do what FAQ says but
enter :=1 in first lines of every block. If we consider the table above:

B2 = 1
B8 = 1
B& = @-1 + 1

is everything you need[*].

> 2. Place the multiplication of "number" with "amount" in the sum
> column

This is the easiest, just follow the manual and write
$7 = $number * $amount

> 3. Add the total sums of the two blocks in row 6 and the last row
> (taking into account that the table will grow over time)

This isn't as hard as it seems too:

bal = vsum(@-II..@-I)
outbal = bal + vsum

(the numbers suggest outbal includes bal am I right).

> I know about column formulas and field formulas but I did not find a way
> to do something like "for each field in column X between row a and row b
> calculate ..."

There is no need for such a thing since cell formulae "cover" column
ones.

> As a kind of extra....for the very org-mode pros on this list
> If I fill dates into the date column, a mechanism which moves
> automatically (in the same way as I refresh calculation of formulas),
> the rows from the second block into the first block when the date passed
> would be some nice gadget

This one *is* tricky. You won't achieve it without lisp (at least
IMHO). OK, let's try:

+ I'd use an empty cell to put a Lisp formula in it: moveup.

+ The value it returnes isn't as important as its side effects: table
  rearrangement.

+ Let's use org timestamps: <2010-09-22 śro> (C-c C-.), so we can use
  org-parse-time-string function.

[...]

Phew, that was a *thing* but here you are:

--8<---------------cut here---------------start------------->8---
(save-excursion
    (let* (
           (hlineA (1+ (org-table-get-descriptor-line "-II")))
           (hlineB (org-table-get-descriptor-line "-I") )
           (curline hlineA) theline)
      (while (< curline hlineB)
        (org-goto-line curline)
        (org-table-goto-column 3)
        (when (time-less-p
               (apply
                'encode-time
                (org-parse-time-string
                 (org-no-properties (org-table-get-field))))
               (current-time) )
          (dotimes (i 4)
            (setq theline (append theline (list (org-table-get-field))))
            (org-table-next-field))
          (org-table-kill-row)
          (org-goto-line (1- (org-table-get-descriptor-line "III")))
          (org-table-goto-column 3)
          (org-table-next-row)
          (dolist (col theline)
            (insert col)
            (org-table-next-field))
          (setq curline (1- curline))
          (setq theline nil)
          )
        (setq curline (1+ curline))))
    "")
--8<---------------cut here---------------end--------------->8---


and put this as:

$moveup = '(save-excursion .... blah blah ..)

It should be possible to cut it down by about 20% but I leave it as an
exercise to someone else ;-)

You could also try the <info:(org)Column view> feature which enables
simple calculations too plus you can keep your numbers in the tree
structure with more elaborate info on them. You could then use agenda
views to select and sum up entries by dates too. 

[*] I've copied all the formulae from the formula buffer which you
activate with C-' while cursor is in a table. This is very convenient
way of entering formulae since you get the cells you reference
highlighted.

-- 
Miłego dnia,
Łukasz Stelmach