[GNC-dev] [GNC] ofxtools - a Python OFX library

[Christopher Singley]

On 5/30/19 10:23 PM, Christopher Lam wrote:
> Not sure how  it works internally.
>
> Perfect matches whereby fitid=online_id are hidden. I'd prefer not to 
> hide, but just grayed out... Because it's disconcerting downloading an 
> ofx and find that the import list is very sparse.
> Unmatched ofx txns attempt to match register txns by multiple 
> approaches: I think amount, date, description-parsing into words then 
> Bayesian matching using internal probability list. It's imperfect.
>
> My major issue is the current importer does *not* reveal its best 
> guessed register-txn therefore I have no idea what it matched against...
>
> And also it does not reveal which register-txns were left unmatched.

This never happens to me, because I avoid manual journal entries as much 
as I can - a habit shared by every accountant with whom I'm acquainted.  
That's the entire point of transaction import... don't make accountants 
do clerks' work; let computers handle that as much as possible.  In that 
workflow, transaction import is a data entry step, not a validation step.

It's entirely possible to use ofxtools to develop an alternative 
importer that better matches your workflow.  However, the infrastructure 
is somewhat lacking in GnuCash python space.  The guys who wrote the 
classes made a good start, but there's a fair amount of work to be done 
before that package is really useful for random-access lookups of the 
existing data set.

In particular, the 5000-line generated file of Python bindings... it 
mockingly includes the function signature as a documentation string.  Is 
there any way to get that file generated with the actual source 
docstrings?  Or else some sort of lookup index of function name to 
source file?