lm-compressors

Purpose

The shared byte-workhorse of the file pipeline: assembly, decompression,
ZIP parsing, output chunking, and staging. Stateless — no files or
store clients inside; it returns bytes and cache references, persistence
is the workflow's job.

Mental model

Workflow hands chunk refs + operation (unpack, assemble, chunk) → lambda
does pure byte work → hands staged bytes/cache refs back. It never decides
what a file means and never stores anything.

Ecosystem value

One place where archive bytes are touched:

• Compressed chunks in, staged entries out — one unpack shape for any caller.
• Any future archive or compression format lands here once and upgrades every intake at once.

Non-goals

• Not file storage or classification.
• Not model conversion or preview rendering.

Responsibility boundary

Owns byte assembly, decompression, archive parsing, output chunking, and
staging; does not own file records or persistence.

Direct module dependencies

• None

Solution membership

• requests

Source

modules/lambdas/data/lm-compressors