More pdf2kmymoney

[Jack]

On 2020.12.30 17:36, Aaron Mehl wrote:
>     On Wednesday, December 30, 2020, 02:45:40 PM EST, Jochen Schmidt  
> <jochen at r-ci.de> wrote:
>  Am 30.12.2020 um 12:22 schrieb Aaron Mehl:
> > I need a little help,
> > I tried the different ideas for moving pdf to kmymoney.
> >
> > Okular - I used the table select, but couldn't figure out how to  
> export
> > csv. Plus I want something more automatic, if possible
> 
> not automatic, but: Select table in Okular, add vertical delimiters by
> clicking on the top border. Then just type Ctr-C.
> 
> Change to a spreadsheet application (i.e. Libreoffice Calc) and type  
> Ctrl-V
> in an empty table. You are asked for formatting details, just  
> accepting the
> defaults does work in most cases.
> 
> > pdftotext - produced a nice text file, but since the columns were
> > separated, by commas and the dates also use commas I need to get the
> > columns separated by tabs? Also I found that in the final text file  
> there
> > was extra rows that was overflow from the same column but the row  
> above.
> 
> You can use awk to output it to useful format. Years ago I wrote a  
> perl
> script to reformat such files according to a config  
> file.---------------------------------------
> 
> 
> You have to help them by providing a suitable script (awk / perl   
> whatever)
> to get this done.-----------------------------------
> I unfortunately, have no such script. I am assuming it would need to  
> be configurable.As I said if anyone has such a script, bring it  
> on.....Aaron 
I'm working on a longer response, but saying it would have to be  
"configurable" is a massive understatement.  Every bank has its own  
format for PDF statements, and even those are subject to change with no  
warning, notice, or logic.  You would likely have to have a separate  
script/program (or at least a module, plugin, or sub-script) for each  
bank.  I suppose some amount of interactive configurability might be  
possible (similar to what KMM already does for importing csv files) to  
find where the data begins and ends and what the columns are.  However,  
designing it would require samples of many banks' statements - and I  
suspect those are pretty hard to obtain from users, as they always  
contain personal information.

Jack