Milestone 7 of 10

Wrap and validate versioned codec files

Apply the LMP version and payload-length boundary independently to every document and reject each outer-container failure before payload decoding.

Each document gets its own versioned codec file. Validate the outer boundary before interpreting the inner payload.

Goal

Wrap each payload in the accepted FILE-15 container, save independent per-file artifacts, and test every outer magic, version, length, truncation, and trailing byte failure.

Inputs

Use valid DTC1 payloads, accepted container version, and positive payload bound. The outer layout is:

bytes 0..2   magic b"LMP"
byte 3       container version, from 0 through 255
bytes 4..7   payload length as an unsigned four-byte big-endian integer
remaining    exact codec payload bytes

Read checks truncated header, magic, accepted version, truncated payload, and trailing container bytes in that order before parsing codec fields. The configuration imposes a smaller positive payload bound before allocation.

Deliverables

Implement src/container.py. Write one exact .lmpc binary file per source document, with digest and container metadata containing magic, version, declared and actual payload lengths, document identity, and codec identity.

Checks

Check exact header bytes, unsigned big-endian length, accepted version, deterministic independent files, payload bound, and source identity. Exercise truncated header, changed magic, unsupported version, truncated payload, declared-length mismatch, trailing container bytes, and excessive payload.

Check outer failures occur before payload decoding, payload bytes pass through unchanged, and source records remain immutable. Do not infer a version or layout from content.

Workspace

Keep outer wrapping and validation in src/container.py. Write per-document binary files and rejection evidence. Do not perform the complete corpus recovery yet.

Hints

HintKeep boundaries separate

The outer header bounds the payload; the payload parser bounds its own fields. Validate the first before the second.

HintUse the declared length

Read bytes 4 through 7 as unsigned big-endian. Remaining file length is a check, not the source of truth.

HintOne file per document

Reset output path and container metadata per source ID. Never let one document's payload share a container with another.

Review

Take a valid payload and append one byte to its container. Which outer check rejects it, and why must the payload parser never see that byte?

How to check your work

Checks compare per-document .lmpc bytes, digests, metadata, and failure reasons with the fixtures. The supplied fixture applies one exact versioned outer format.