TD Bank and strange QFX Format

[Jack]

On 2019.10.15 13:03, Thomas Baumgart wrote:
> On Dienstag, 15. Oktober 2019 06:00:54 CEST Ryan Novosielski wrote:
> 
> > On Mon, Oct 14, 2019, at 11:06 PM, Jack wrote:
> > > Hello, Ryan,
> > >
> > > On 2019.10.14 18:30, Ryan Novosielski wrote:
> > > > Using KMyMoney 5.0.0.
> > > >
> > > > Recently TD Bank changed their website around, and now when I
> > > > download a transaction history for a credit card now, it  
> imports kind
> > > > of screwy. In the native OFX importer, I get every transaction  
> with a
> > > > name of **UNASSIGNED** and the transaction is a credit instead  
> of a
> > > > charge as it should be.
> > > >
> > > > I tried importing via OFX through AqBanking instead. That at  
> least
> > > > gets the name of the retailer correct, but the transactions are  
> still
> > > > credits instead of charges.
> > > >
> > > > Any advice for the best way to make this somewhat less painful?
> > > This is unfortunately, not an uncommon issue.
> > >
> > > I can post more details tomorrow, but there is a configuration  
> choice
> > > which OFX tag is used for selecting the Payee.  You might need to  
> look
> > > into what the bank is actually providing, but you can try  
> different
> > > values for this choice.
> > >
> > > I'll restrain myself from simply suggesting choosing a different  
> bank,
> > > since this type of problem sooner or later hits all providers.   
> The
> > > issue is that while the bank itself might or might not program  
> it's
> > > website, it is almost certain that their OFX interface is most  
> likely
> > > contracted out to a different provider, most likely unconnected to
> > > their basic website.
> >
> > I don't actually use them as a bank, just their credit card, but  
> yes, they are awful in all sorts of ways, so they're probably history  
> in the long run, but that doesn't help this month.
> >
> > > For the setting I referred to, open the account in the ledger,  
> then
> > > edit the account, select the online settings tab, then the Import
> > > Details subtab.  Try the different options for the "Payee's name  
> is
> > > based on contects of the OFS tag" dropdown.  (choices are Payeeid,
> > > Name, and Memo.)
> >
> > Does this account have to be mapped to an online account in order  
> for you to see that option? I went looking in the documentation  
> earlier, found that suggestion, and can't find it anyplace in the  
> actual software. This bank also doesn't support Direct Connect, so  
> far as I know.
> 
> This is probably what you are looking for:
> 
> commit 65e3837697c9378616a7eda2ba3387d1208f288e
> Author: Thomas Baumgart <thb at net-bembel.de>
> Date:   Sun Apr 7 09:12:22 2019 +0200
> 
>     Provide method to construct unique transaction ID for OFX import
> 
>     User reports show, that the FITID attribute provided by some
>     institutions during OFX statement download is not as unique as it  
> is
>     specified. In fact, some banks issue a different FITID every time  
> you
>     download the same statement.
> 
>     This makes the FITID attribute absolutely worthless for duplicate
>     detection.
> 
>     Since asking banks to fix their software has no reasonable chance  
> to
>     lead to success on short call, this change allows to switch the
>     duplicate detection to be based on a method provided by KMyMoney.  
> In
>     fact it uses the same algorithm which works well in the KBanking
>     importer for years already. Note: switching from "OFX FITID" to
>     "KMyMoney Hash" may result in duplicates for one more time.
> 
>     At the same time, the OFX import options are now available during  
> file
>     import and not only for mapped accounts (OFX direct connect).
> 
>     BUG: 390681
>     FIXED-IN: 5.0.4
> 
> 
> Regards
> 
> Thomas
> 
> --
> 
> Regards
> 
> Thomas Baumgart
Thomas,

I think this is about which OFX tag to use for identifying the Payee.   
There was a discussion a while ago that the setting was only available  
for already mapped accounts, and not available (yet) for importing OFX  
files.   In fact, I filed https://bugs.kde.org/show_bug.cgi?id=369438  
about two years ago for exactly this problem.

Jack