Re: [Orgmode] Re: Org-Babel and Ledger

[Eric Schulte]

Hi Seb,

Sébastien Vauban <address@hidden> writes:

> Hi Eric,
>
> Sébastien Vauban wrote:
>> "Eric Schulte" wrote:
>>> Sébastien Vauban writes:
>>>>
>>>>>>> As you can see, the tables are completely wrongly made, because they're
>>>>>>> based on spaces ("à la Awk") and not on fixed position of fields ("à la
>>>>>>> Cut").
>>>>>>>
>>>>>>> What can I do about this?
>>>>>>>
>>>>>>> - Post-process every ledger command with some awk or cut command that
>>>>>>>   will do whatever is needed

This might actually be a good solution for the shell block you mention
below.

>>>>>
>>>>>   (org-table-convert-region (point-min) (point-max))
>
> As you adviced, I can play with the optional parameter in ob-ledger's code.
>
> But what about the results of sh code?
>
> Imagine now that I'm forced to use =sh= instead of =ledger=, in the following
> pratical case:
>
> #+srcname: reg
> #+begin_src sh
> m4 scorpios-ledger.dat | ledger -f - reg unknown
> #+end_src
>
> I'm using =m4= (and thus make some pre-processing) because I need (or want) to
> be able to switch between =ledger= and =hledger= and comment in or out the
> directives that are accepted by one but not by the other.
>
> Then, such a result is still problematic:
>
> 09-Aug-21 CHEQUE : 9953055                    Expenses:Unknown                
>                     166.70 EUR            166.70 EUR
> 09-Sep-17 CHEQUE : 7691785                    Expenses:Unknown                
>                     100.00 EUR            266.70 EUR
> 09-Oct-16 REMISE CHEQUE N 8686318 001 105     Expenses:Unknown                
>                    -525.00 EUR           -258.30 EUR
>
> I've looked in =org-babel-execute:sh= and =org-babel-reassemble-table=, among
> others, to look for similar constructs (calls to =org-table-convert-region=)
> but did not find any.
>
> Is there still a solution to get such results?
>

As an interim solution you could add a :results scalar header argument
to your sh code block, then pass the output of that code block to a
parsing code block (namely the one from my previous email).  That's
certainly not idea, but it should work for the moment.

Alternately if ledger supports a more clearly delimited output format
(e.g. csv) then that would probably work as well.

>
> #+results:
> | 09-Aug-21 CHEQUE : 9953055                | Expenses:Unknown | 166.70 EUR  
> | 166.70 EUR  |
> | 09-Sep-17 CHEQUE : 7691785                | Expenses:Unknown | 100.00 EUR  
> | 266.70 EUR  |
> | 09-Oct-16 REMISE CHEQUE N 8686318 001 105 | Expenses:Unknown | -525.00 EUR 
> | -258.30 EUR |
>
> Couldn't we specify somehow the field separator as an extra header
> argument?  I've looked for such an already existing in
> http://orgmode.org/manual/Specific-header-arguments.html#Specific-header-arguments,
> but that does not seem to be available yet.

This would probably be a good addition to ob-sh (and any other language
which doesn't have a defined list syntax).  I'll add this idea to the
babel dev stack, but I'm swamped for the next week or so.

Best -- Eric

> Though, I do not know if this is the right solution. I'm sure you'll
> tell me... ;-)
>
> Best regards,
>   Seb