A check of the F::Q modules found that the only ones that return a quote
time return a bogus one and do so only to mollify GnuCash.
Since there's no good way to determine the TZ of the exchange originating
the quote there's no good way to decide if the quote is current or from
a previous market session, so we just punt and use a time of 16:00 for
all quotes.
Allows for cleaner code with less state, less coupling of the GncQuotes
class, and better transfer of error messages to client code.
Also translates some error messages for presentation to users.
Provide a specialization GncFQQuoteSource and move the F::Q command
construction and query functions to GncFQQuoteSource.
This allows for dependency injection to provide testing that doesn't
need F::Q to be installed.
- make more use of auto
- mark user visible strings as translatable
- return early on input errors
- fix date conversion fallback to actually fall back to today
The book parameter is only needed while fetching quotes.
In case the user passes one or more commodities to process
the book can be readily derived from the commodity/commodities.
In the other case (fetch all quotes) the user now is
required to pass a book to the call.
This code will convert the json data into GncPrice objects and add them
to the pricedb, effectively doing what price-quotes.scm does.
A few notable remarks:
- still requires plenty of cleaning up. This is the first proof of concept
- like the original scm based code, this parser completely ignores timezone
information. As it wasn't used before and nobody complained, it may not
be that important. Or it can be implemented later.
- price-quotes.scm would first check if a price already existed in the pricedb
and try to update that one instead of adding one (only if the old price's
type is inferior). However that is redundant as gnc_pricedb_add_price does
the same check. So I have omitted this extra check from GncQuotes.
- currency quotes can be inverted. I have slightly changed the way to handle
this. The perl wrapper code will simply set an "inverted" flag in that case,
but will otherwise not swap currency and commodity as it used to be the case.
On parsing, the inversion flag will cause the GncNumeric that's parsed from
the price to be inverted. As it's still a GncNumeric that shouldn't result
in any loss of precision, while keeping prices in the db always in the default
currency.
That allows the private implementation to pass a number of variables
based on various boost libraries. It's better to not have them in
the public interface to keep compilation times down.
For all but the basic check a book is required. Might
as well be able to pass it directly and store a reference
to it. That will simplify member function declarations.
I have been reading on singleton implementations and there appears
to be a lot of pushback against those.
We can revisit this if it turns out performance degrades
significantly by running the F::Q check multiple times.
This rewritten version takes JSON input and spits out JSON.
Additionally inverted currency quotes will only be flagged.
The old code also swapped currencies in the result.
GncQuotes will be written towards these new implementation
choices.
At the same time do an explicit reinstantiation of quotes_cached at first use
to work around what seems to be a race condition between static instantiation
and binreloc.
This library intends to wrap Finance::Quote the way price-quotes.scm currently does.
In this first commit the library replaces price-quotes.scm's library to install
quote sources. In addition it exposes a new command line parameter in gnucash-cli
to show which version of Finance::Quote is installed on the system (if any) and
which quotes sources that verions exposes.
Instead of using static_assert. This prevents the compiler from even
trying and avoids weird compilation errors when testing types for
instantiating other templates.
The compiler complains that there's no matching
gnc_register_number_range_option for GncOptionDB*, which without this
commit is true because the explicit templates are for GncOptionDBPtr&.
Note that the original template definition is for GncOptionDB* and
that the header-defined inlines that take GncOptionDBPtr& call the
GncOptionDB* version.
Instead of using static_assert. This prevents the compiler from even
trying and avoids weird compilation errors when testing types for
instantiating other templates.
The compiler complains that there's no matching
gnc_register_number_range_option for GncOptionDB*, which without this
commit is true because the explicit templates are for GncOptionDBPtr&.
Note that the original template definition is for GncOptionDB* and
that the header-defined inlines that take GncOptionDBPtr& call the
GncOptionDB* version.
Because when loading the value the split in the xml backend doesn't yet
have a parent so the code tried to convert to GNC_COMMODITY_MAX_FRACTION
and if the numerator was larger than 10^10 that would overflow.
To fix it this changes the "don't know" response to get_currency_denom
and get_commodity_denom to GNC_DENOM_AUTO which will normally leave
the denominator alone.
Handle template transactions that don't have any splits with empty credit
and debit strings and those having no set transaction account.
Set the concrete transaction commodity to the first found of:
The template transaction's commodity
The commodity of the first split with a credit or debit string
The commodity of the first split.
Test these scheduled transactions:
* 2 splits with fixed amounts "123"
* 2 splits with fixed amounts "0"
* 2 splits with empty amounts ""
Verify that automatically created scheduled transactions exist.
Enables full functionality for complex-boolean and multichoice-callback
options.
Note that setter-function callback isn't used in any GnuCash code so it
is not implemented and is not present in the register-callback functions.
The root cause of which is that on recent releases of GLib (recent
meaning 2.66 in the current Debian stable!) g_type_instance_get_private
looks in the wrong place for the private data. When running the script
in question it returned NULL and since the code didn't check for a valid
pointer, it crashed.
So this change replaces all calls to g_type_instance_get_private with
the function [type_prefix]_get_instance_private() added in glib-2.36
except for two register2 files that have been removed from master; those
are ignored to avoid unnecessary merge conflicts.
This scrubbing function calls xaccTransRollbackEdit which is
leaky. Instead of trying to fix xaccTransRollbackEdit which requires
superhuman skills, it's easiest to track the changes separately in a
GList, and use xaccTransBeginEdit/xaccTransCommitEdit only if
necessary.
tests TXN_TYPE_NONE in utest-Transaction.c
testing TXN_TYPE_INVOICE, TXN_TYPE_PAYMENT, and TXN_TYPE_LINK will
require valid posted invoices, so, are best tested in utest-Invoice.c
