Errors from run_query(), not from bean-query

[Eric Altendorf]

I'm getting errors inside beancount/ops/summarize.py while using beanquery's run_query() programmatically.  This is on a beancount ledger that works fine when I run bean-query from the command line.  I'm not sure if I'm failing to use the programmatic API correctly or if there's something else going on.

bean-query access (works):

subprocess.run(["bean-query", ledger, query], check=True, capture_output=True) 

Programmatic access (doesn't work):

entries, errors, options = parser.parse_file(ledger)

run_query(entries, options, query)

The first error I got was this:

  ....

  File "/home/eric/crypto-taxes/beancount/beancount/ops/summarize.py", line 497, in conversions
    narration = 'Conversion for {}'.format(conversion_balance)
  File "/home/eric/crypto-taxes/beancount/beancount/core/inventory.py", line 122, in __str__
    return self.to_string()
  File "/home/eric/crypto-taxes/beancount/beancount/core/inventory.py", line 114, in to_string
    ', '.join(pos.to_string(dformat) for pos in sorted(self)))
  File "/home/eric/crypto-taxes/beancount/beancount/core/position.py", line 246, in __lt__
    return self.sortkey() < other.sortkey()
  File "/home/eric/crypto-taxes/beancount/beancount/core/position.py", line 227, in sortkey
    if self.cost.number:
AttributeError: 'CostSpec' object has no attribute 'number'

This was only a little surprising because I'm using the "{USD}" cost-spec hack, and I've had to elsewhere tweak beancount-core to be robust to such cost specs.  It's still a bit surprising since as I said don't get this error when running the bean-query tool on the same file.  Anyway, I hacked the comparator and got past this point, to:

  ....

  File "/home/eric/crypto-taxes/beancount/beancount/ops/summarize.py", line 480, in conversions
    conversion_balance = interpolate.compute_entries_balance(entries, date=date)
  File "/home/eric/crypto-taxes/beancount/beancount/core/interpolate.py", line 275, in compute_entries_balance
    total_balance.add_position(posting)
  File "/home/eric/crypto-taxes/beancount/beancount/core/inventory.py", line 461, in add_position
    return self.add_amount(position.units, position.cost)
  File "/home/eric/crypto-taxes/beancount/beancount/core/inventory.py", line 416, in add_amount
    key = (units.currency, cost)
AttributeError: type object 'MISSING' has no attribute 'currency'

So somehow I've got a missing value for some units.  I started tweaking this code too to be robust to this case, but there were too many places where it's assumed not-missing, and it seemed sketch to be hacking the code for adding to inventories this way.

Any idea what's going on?

[Martin Blais]

Use bn.loader.load_file(), not just the parser.

Beancount has two stages: parsing, then booking and running all the plugins.

Loading does both.

The query is intended to be run on the output of the loader.

--
You received this message because you are subscribed to the Google Groups "Beancount" group.
To unsubscribe from this group and stop receiving emails from it, send an email to beancount+...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/beancount/CAFXPr0st_mX_NyovM%2BDkr8XJ%3D7Q048kOa8JHhUfCyaAcP9M0Zw%40mail.gmail.com.

[Martin Blais]

Btw, fwiw, the beanlabs repo has a bunch of random scripts you might find useful to eyeball, that use the api

[Eric Altendorf]

On Wed, Aug 2, 2023 at 5:55 PM Martin Blais <bl...@furius.ca> wrote:

On Thu, Aug 3, 2023, 08:54 Martin Blais <bl...@furius.ca> wrote:

Use bn.loader.load_file(), not just the parser.

Beancount has two stages: parsing, then booking and running all the plugins.

Loading does both.

The query is intended to be run on the output of the loader.

Awesome.  I figured it must be something like that.

 

Btw, fwiw, the beanlabs repo has a bunch of random scripts you might find useful to eyeball, that use the api

Thanks, I had looked around for recipes but google was failing me.

Up and running.

 

To view this discussion on the web visit https://groups.google.com/d/msgid/beancount/CAK21%2BhNn8j6t8kcdPUgg3WL0B5WOCkGdWG4LMLcAa3es7QhE_A%40mail.gmail.com.

[Daniele Nicolodi]

On 03/08/23 07:20, Eric Altendorf wrote:
>
> On Wed, Aug 2, 2023 at 5:55 PM Martin Blais <bl...@furius.ca
> <mailto:bl...@furius.ca>> wrote:
>
> On Thu, Aug 3, 2023, 08:54 Martin Blais <bl...@furius.ca
> <mailto:bl...@furius.ca>> wrote:
>
> Use bn.loader.load_file(), not just the parser.
> Beancount has two stages: parsing, then booking and running all
> the plugins.
> Loading does both.
> The query is intended to be run on the output of the loader.
>
> Awesome.  I figured it must be something like that.

BTW, I recently added another main entry point for the beanquery API
that should make its use a bit easier:

import beanquery
connection = beanquery.connect('beancount:' + filename)
types, rows = connection.execute('''SELECT ...''')

It is a first step into providing a Python DB-API interface, thus it
will be further transformed, but if you want to experiment with it, it
works in the form above now.

Cheers,
Dan