roboto.formats

Format-specific decoding of topic data files into rows.

This package decodes MCAP and Parquet topic data into Roboto’s nested row / Arrow representation: per-format parsing and read planning, field projection, and timestamp extraction. The byte-transport it builds on (HTTP range reads, local disk caching, sparse buffering) lives in roboto.storage; this package depends on the topic record and message-path types in roboto.domain.topics without depending back on the topic readers, so the readers in both roboto.domain.topics and roboto.experimental.topics build on it.

Submodules

• roboto.formats.fields
• roboto.formats.mcap
• roboto.formats.parquet

Package Contents

class roboto.formats.FieldSelection

A field to read out of a data file, identified by its path through the schema.

The roboto.formats readers take the fields to read as FieldSelection values; to_field_selection() converts a topic’s message path into one.

path_in_schema: tuple[str, ...]

Path components locating this field in the source data schema, root to leaf.

A component may contain dots, so ("header", "stamp") (a stamp field nested in a header struct) and ("header.stamp",) (a top-level column named header.stamp) are different fields.