Today I have to clear over a quarter worth of transactions on my current accout. With a bit of help from, I was able to get the importing script (in the repository: doc/samples/recon_csv.pl
) adapted to my bank's download (tabbed text file).
The resulting script has to be stored in LedgerSMB/Reconciliation/CSV/Formats as recon_tab.pl
because my extensions are .tab.
The result is much better than going through the entire list by hand, but I've found some problems as well:
1. If I make a mistake and make a reverse posting to undo the effect, I typically put a prefix "REVERSE:" in the description field. However, this seems to have a counter-productive effect, since these are not matched to be cleared before any of the others
2. If I want to make the matching algorithm match, I have to pass empty "SCN" fields (basically a description), however, if there are any unmatched transactions, this leaves me with a (possibly long) list of hard to recognize transactions.
3. Lines are added to the report's third section if, e.g., I enter cash receipts. If said cash receipt was in the upload file, it's not being matched when I click Update. (Re-uploading the file doesn't work as it simply adds all previously matched items in the file to the unmatched list.)
I'd like to change at least (2) and (3). But in my ideal solution, the reconciliation tool would help generate the right transactions for the unmatched lines from the upload, e.g. by matching the description field for open AR/AP invoice numbers or if those are not available by building up history, learning the usual accounts against which lines with specific tokens are usually posted.
So, basically, my question would be: where to start?