feat: add support for use in importer hooks

[blaggacao]

Context

CONFIG = [
  apply_hooks(bank_account, [
    ReconcileExpected("expected-incoming.beancount"),
    PredictPostings(),
  ])
]
script_utils.ingest(
  CONFIG,
  hooks=[
    ReconcileExpected.adapted_find_duplicate_entries,
  ],
)

[blaggacao]

@maread99 The third commit is just necessary a necessary fix so that I can operate beanahead in my setting. Against all odds, I decided to use global state, because the call stack is deep and the main ledger is part of a global context during runtime.

[maread99]

Hi @blaggacao. Thank you for this PR. Much appreciated!

I don't think I'm not going to get a chance to have a good look before next week. I will do then if not before.

I ran the tests although it's being pulled up on the linting.

Cheers
Marcus

[blaggacao]

@maread99 Ok, I ensured linting, cleaned up the history and use this -as is- in my own production, now. Thanks for considering this contribution!

You could also cherry-pick the first commit as an independent fix.

[maread99]

Thanks again for this @blaggacao. I've had a look, am grateful for the PR and in principle more than happy to merge the new functionality.

I've made a few comments and requests for info / changes. A few of the tests are failing although at least some should come good if the points I've raised are addressed.

Cheers

[maread99]

Would you mind adding a simple test to verify that all's working as expected when users are using non-default category names?

[maread99]

Could a simple test be added to verify the hook is working as intended?

[blaggacao]

For the tests:

• How do I install the environment locally (project-level, not global)?
• How do I run the tests? (just pytest, correct?)

[maread99]

To a fresh (or existing) python environment (>=3.9) you could install the current version with the optional 'dev' dependencies:

>>> pip install beanahead[dev]

Then uninstall beanahead:

>>> pip uninstall beanahead

And install your fork. If you have your fork installed locally then from the project directory (containing pyproject.toml):

>>> pip install -e .

Otherwise:

>>> pip install git+https://github.com/blaggacao/beanahead.git#egg=beanahead

Or, I'd have thought you might(?) be able to do the whole lot with either:

>>> pip install -e .[dev]

or

>>> pip install git+https://github.com/blaggacao/beanahead.git#egg=beanahead[dev]

As for the tests, yup, pytest.

(I use VSCode which nicely integrates both the creation of environments and testing with pytest.)

[maread99]

Hi @blaggacao, I've added a commit which gets all the existing tests passing.

If, when you can, you could add tests to cover the new functionality and then I think we'll be good to merge.

Cheers
Marcus

[blaggacao]

Thank you for the additions!

I've high load, atm, and will ll try to revisit next week.

[maread99]

Hi @blaggacao

I've made some changes to make beanahead compatible with beancount v3. As it is this PR isn't compatible with v3 but I suspect it would just be a matter of changing the dependency on beancount.ingest to beangulp. Although, I'm not sure if smart-importer is v3 compatible(?).

In any event, I think I'm leaning towards not implementing the smart-importer hook directly in beanahead. (This is principally because I don't use smart-importer and would be unable to provide any ongoing maintenance to the hook.) What I have done is implemented the following aspects of this PR:

• Support for custom account root name options (Add support for custom name options #19)
• An option to stream print to stderr (Include option to stream print to stderr #18)

These changes are included in the new v0.3 release. The new Options section of the README provides the documentation together with the note at the end of the Making beanahead files section.

I was thinking it might be best if you publish the hook yourself and we include a reference and link on the README here? With the changes above now made I'd have thought this could be pretty simple?

Let me know what you think.

Cheers
Marcus

[blaggacao]

Hey, this sounds great!

For expectation management, my setup is currently working and I won't touch it in a while (maybe a year or so).

But I've taken note and will seek an occasion to conform to your suggestion, which I agree with and find reasonable.

Let's maybe keep this open just for reference, in the meantime? (Just for the sake of not burying it inside the "closed" stack of PRs).

[maread99]

Great, let me know if/when you do set something up and we can add a reference and link to the README here.

I'll leave this PR open in the meantime.