@cervid/decomposer

High-performance binary data decomposer for the Cervid ecosystem.

@cervid/decomposer is the low-level ingestion layer for binary and columnar formats such as:

• NPY
• Parquet
• Arrow
• Avro
• other binary-oriented formats

Its job is to read source files and expose them in a shared-memory-friendly representation that can later be consumed by other Cervid libraries, such as @cervid/data.

Status

This package is currently under construction.

The API may change while the core architecture stabilizes.

At the moment, the main focus is:

• NPY ingestion
• shared-memory output
• stable interchange contract between Decomposer and Data

Goals

• Fast binary ingestion
• SharedArrayBuffer-based output
• Zero-copy or near-zero-copy interoperability
• Clean bridge to @cervid/data
• Future format exporters (toCSV, toJSON, toNPY, toParquet, etc.)

Installation

npm install @cervid/decomposer