Debian bug: QIF import causes data loss

David Hampton hampton-gnucash at rainbolthampton.net
Mon Jun 5 20:28:35 EDT 2006


On Mon, 2006-06-05 at 20:13 -0400, Derek Atkins wrote:
> Quoting David Hampton <hampton-gnucash at rainbolthampton.net>:
> 
> > If nothing else, the various gnucash importers should check that the
> > data is valid utf8 during import.
> 
> And what if they are not?  If you're running a non-UTF8 locale the
> imported data could be ISO or something else..  QIF for example has
> nothing to specify the encoding.  I suppose we could add a page to
> the importer druid that asks the user for the input encoding, but most
> likely they'd have no clue what it is.

I don't know.  Ask for the encoding?  Drop the invalid character?
Refuse the data entirely?  Something else?  Anything other than
accepting invalid utf8 would be preferable.

If invalid utf8 is imported and the data saved to an xml file, there is
know way I know to recover except by hand editing the file.  As far as I
can tell libxml2 simply stops parsing when it encounters a invalid utf8
character.  Period.  I can't find any available exception handler that
could be used for automated recovery. 

David




More information about the gnucash-devel mailing list