A smart importer for beancount and fava, currently in development as a suggestion for #579 Import: Intelligent suggestions for account names

Prototype, work in progress.

pip install smart_importer

# or, to install from local git clone:
pip install --editable .

When writing a beancount importer, users can apply decorators to their importer classes in order to benefit from smart predictions and suggestions provided by machine learning.

For example:

@PredictPostings(training_data="trainingdata.beancount")
@PredictPayees(training_data="trainingdata.beancount")
class MyImporter(ImporterProtocol):
    def extract(file):
      # do the import, e.g., from a csv file

If you don't want to modify your importers you can also manually instantiate this e.g. in our foo.import

from smart_importer.predict_postings import PredictPostings

MyImporter = PredictPostings(suggest_accounts=False)(MyImporter)
CONFIG = [
   MyImporter("someconfig")
]

# for nicer test output:
pip install coloredlogs

# to run unittests:
make test