lassen 0.1.1

Common webapp scaffolding.

lassen

40.4881° N, 121.5049° W

Core utilities for MonkeySee web applications.

Not guaranteed to be backwards compatible, use at your own risk.

Structure

Stores: Each model is expected to have its own store. Base classes that provide standard logic are provided by lassen.store

• StoreBase: Base class for all stores
• StoreFilterMixin: Mixin for filtering stores that specify an additional schema to use to filter

Datasets: Optional huggingface datasets processing utilities. Only installed under the lassen[datasets] extra. These provide support for:

• batch_to_examples: Iterate and manipulate each example separately, versus over nested key-based lists.
• examples_to_batch: Takes the output of a typehinted element-wise batch and converts into the format needed for dataset insertion. If datasets can't automatically interpret the type of the fields, also provide automatic casting based on the typehinted dataclass.

from lassen.datasets import batch_to_examples, examples_to_batch
import pandas as pd

@dataclass
class BatchInsertion:
    texts: list[str]

def batch_process(examples):
    new_examples : list[BatchInsertion] = []
    for example in batch_to_examples(examples):
        new_examples.append(
            BatchInsertion(
                example["raw_text"].split()
            )
        )

    # datasets won't be able to typehint a dataset that starts with an empty example, so we use our explicit schema to cast the data
    return examples_to_batch(new_examples, BatchInsertion, explicit_schema=True)

df = pd.DataFrame(
    [
        {"raw_text": ""},
        {"raw_text": "This is a test"},
        {"raw_text": "This is another test"},
    ]
)

dataset = Dataset.from_pandas(df)

dataset = dataset.map(
    batch_process,
    batched=True,
    batch_size=1,
    num_proc=1,
    remove_columns=dataset.column_names,
)

Migrations: Lassen includes a templated alembic.init and env.py file. Client applications just need to have a migrations folder within their project root. After this you can swap poetry run alembic with poetry run migrate.

poetry run migrate upgrade head

Settings: Application settings should subclass our core settings. This provides a standard way to load settings from environment variables and includes common database keys.

from lassen.core.config import CoreSettings, register_settings

@register_settings
class ClientSettings(CoreSettings):
    pass

Schemas: For helper schemas when returning results via API, see lassen.schema.

Development

poetry install --extras "datasets"

createuser lassen
createdb -O lassen lassen_db
createdb -O lassen lassen_test_db

Unit Tests:

poetry run pytest