This reduces parsing time in the optimized build by 25%, and was a safe,
easy patch. If the "quick predicate evaluator" fails, we disable it
from that point on go back to what the standard code does.
This lets you run standard report queries against a Ledger journal and
get back a collection of postings, for example:
import ledger
journal = ledger.Journal("sample.dat")
posts = journal.collect("-M assets")
for post in posts:
print post.account
However, this code is not really working yet for a large variety of
reasons, having to do with object life-time, shallow copying, and lack
of reference counting where it should be. For instance, calling
ledger.Journal().collect() fails because the temporary journal object is
destroyed and the collection now has a host of bad pointers. Using "for
post in journal.collect()" fails too with other bad pointers. And the
whole lot of it crashes on exit at the moment.
This allows for value expressions to be used which reference the
incoming posting, for example:
= Income:Clients:
(Liabilities:Taxes:VAT1) (floor(amount) * 1)
(Liabilities:Taxes:VAT2) 0.19
2009/07/27 * Invoice
Assets:Bank:Checking $1,190.45
Income:Clients:ACME_Inc
The automated posting for VAT1 will use the floored amount multiplied by
a factor, while the posting for VAT2 multiples the whole amount as
before.
This lets you, for example, debug registers that cull data from many
different sources, without having to change the basic formatting
string. You can locate each posting's location with this:
ledger reg --prepend-format='%-25(filename + ":" + beg_line)'
Example of a tagged entry:
i 2009/11/01 12:00:00 Account Payee ; :Foo:
o 2009/11/01 13:00:00
Two spaces or a tab must separate account from payee, and payee from
note.
