utils

module of dascore.io.arrow source

Encode patches as Arrow IPC and decode them back.

One patch is one IPC stream or file whose single column, data, holds the patch’s values flattened in C order. Dims, shape, coordinates and attrs are a JSON document in the schema metadata under dascore:patch. The format is specified in dastor’s docs/arrow-patch-format.md.

Functions

Name Description
arrow_type Return the Arrow type for a data dtype, refusing unsupported ones.
batch_values Return a checked batch’s values as a numpy array (a view where possible).
coord_to_json Write a coordinate as the format’s coordinate document.
data_batches Yield the record batches holding a patch’s values, in first-dim slices.
document_meta Return the data-less patch a checked document describes.
json_to_coord Rebuild a coordinate from its document.
patch_document Return the dascore:patch document for a patch.
read_document Return the checked dascore:patch document from an Arrow schema.