Labox

A storage framework for heterogeneous data in Python.

At a Glance

Model your data with Pydantic

import pandas as pd
import plotly.express as px
from labox.extra.pydantic import StorableModel


class ExperimentData(StorableModel, class_id="..."):  # (1)!
    name: str
    params: dict[str, float]
    data: pd.DataFrame
    plot: go.Figure


# Populate it with data
params = {"temperature": 298.15, "ph": 7.4, "concentration": 0.1}
data = pd.DataFrame({"measure_num": [1, 2, 3], "measure_val": [1.23, 4.56, 7.89]})
plot = px.scatter(data)
experiment = ExperimentData(name="Measurement data", params=params, data=data, plot=plot)

1. Models have immutable class_ids you need to define.

Then save and load it to a storage backend of your choice:

import boto3
from sqlalchemy.ext.asyncio import AsyncSession
from labox.core import Registry, save_one, load_one
from labox.extra.aws import S3Storage, simple_s3_router

# With a bit of setup
s3_storage = S3Storage(boto3.client("s3"), simple_s3_router("my-bucket"))
registry = Registry(...)  # (1)!
session = AsyncSession(...)  # (2)!

record = await save_one(experiment, session=session, registry=registry)
loaded_experiment = await load_one(record, ExperimentData, session=session, registry=registry)

1. Labox has a registry of integrations you need to declare
2. Labox uses SQLAlchemy sessions you need to be establish

Installation

To install the core package:

pip install labox

To include the core package and extra integrations:

# all extras
pip install labox[all]

# or specific extras
pip install labox[pydantic,pandas,aws]

Be sure to checkout how Labox works with Pydantic as well as all the other integrations Labox offers.

Next Steps

Check out more usage examples, or dive into the concepts and integrations to learn more about how Labox works.