sheafdata 0.1.0a1

Immutable record batches with deterministic selection and projection.

sheafdata

sheafdata is a dependency-free immutable record-batch type for small, inspectable datasets. It validates a fixed column layout and provides stable projection, filtering, and record conversion without pretending to be a full dataframe library.

from sheafdata import RecordBatch

batch = RecordBatch.from_records(({"id": "a", "score": 3}, {"id": "b", "score": 1}))
assert batch.select("id").records() == ({"id": "a"}, {"id": "b"})