[GNC-dev] [GNC] Failed - import of QIF files from Quicken

[Christopher Lam]

I think I've nailed this bug. Apply this modification to qif-to-gnc.scm and
it should handle empty S fields. C

modified   gnucash/import-export/qif-imp/qif-to-gnc.scm
@@ -584,13 +584,11 @@
                                         (hash-ref qif-memo-map qif-memo)))
                                (and (string? memo)
                                     (not (string=? memo ""))
-                                    (hash-ref qif-memo-map memo))))
-                     (if (not far-acct-info)
-                                (set! far-acct-info
-                                      (hash-ref qif-memo-map
-                                                (default-unspec-acct))))))
+                                    (hash-ref qif-memo-map memo))))))

-                   (set! far-acct-name (qif-map-entry:gnc-name
far-acct-info))
+                   (set! far-acct-name (if far-acct-info
+                                           (qif-map-entry:gnc-name
far-acct-info)
+                                           (default-unspec-acct)))
                    (set! far-acct (hash-ref gnc-acct-hash far-acct-name))

On Fri, 28 Feb 2020 at 15:41, Christopher Lam <christopher.lck at gmail.com>
wrote:

> (attachment snipped)
>
> To help you import your old quicken datafile, you would best revert to
> 3.1, import completely then upgrade to 3.8. The offending code has been
> identified but is *very* difficult to understand because the author left
> nearly 20 years ago. Meanwhile the code base and the language have both
> moved on.
>
> On Tue, 25 Feb 2020 at 18:34, Tom Hatzigeorgiou <tomhatz at hotmail.com>
> wrote:
>
>> Hi I have attached the first year of my data that fails as I try to
>> import into GC.
>> I have data since '89, the first year that I have an issue with is 1997.
>> It was about the time that I started trading, so maybe the problem has
>> something to so with my trading account.
>>
>> Again, the QIF file loads with no problem using GC (for windows) version
>> 3.1 (gnucash-3.1-2.setup.exe), but it fails with the importer of version
>> 3.8 (gnucash-3.8.setup.exe).
>>
>> Thank you.
>>
>>
>>
>> ------------------------------
>> *From:* Tom Hatzigeorgiou <tomhatz at hotmail.com>
>> *Sent:* Tuesday, February 18, 2020 11:18 PM
>> *To:* Frank H. Ellenberger <frank.h.ellenberger at gmail.com>
>> *Cc:* Christopher Lam <christopher.lck at gmail.com>; GnuCash users group <
>> gnucash-user at gnucash.org>; gnucash-devel <gnucash-devel at gnucash.org>
>> *Subject:* Re: [GNC-dev] [GNC] Failed - import of QIF files from Quicken
>>
>> If what I asked cannot be done i will try to produce a file with the
>> errorsome other way
>> I would suggest to take these emails offline.  I think they confuse the
>> user community
>>
>> I will be in touch
>>
>> Thanks
>>
>>
>> On Feb 18, 2020 5:34 PM, "Frank H. Ellenberger" <
>> frank.h.ellenberger at gmail.com> wrote:
>>
>>
>>
>> Am 18.02.20 um 22:52 schrieb Christopher Lam:
>> > Sorry. Please ask other devs on public mailing list.
>> >
>> > On Tue, 18 Feb 2020 at 15:51, Tom Hatzigeorgiou <tomhatz at hotmail.com>
>> wrote:
>> >
>> >> To do this someone will need to help me.
>> >> I don't know how projects like this work.
>> >> I have no idea where the program source exists and the tools I will
>> need
>> >> to manipulate it.
>> >>
>> >> I was thinking that a very simple way to debug the code is, for
>> someone to
>> >> place a display on a version of the new importer where every record
>> that it
>> >> processes gets displayed (in a log file) before it gets imported.
>> >> This way by running the importer I will see at what record(s) it
>> failed.
>> >> I would run it multiple times until I find and eliminate ALL the
>> records
>> >> of my file that it fails on.
>> >> When all bad records get eliminated and my file gets loaded correctly
>> by
>> >> the current importer, I don't think I would have a problem sending you
>> all
>> >> those bad records for you to analyze and see where the new importer
>> has
>> >> gone bad.
>> >>
>> >> DO you think something like the above can be done?
>>
>> At least it should - instead of "Failed" - say "Failed to read line nr
>> %d".
>>
>> That would reduce the burden of bisect.
>>
>> ~Frank
>>
>>
>>