Skip to content

CHANGELOG.md

Latest commit

 

History

History
95 lines (75 loc) · 8.72 KB

CHANGELOG.md

File metadata and controls

95 lines (75 loc) · 8.72 KB
Raw

@chestnutlabs/gcode-bgcode

0.4.0

Patch Changes

  • Updated dependencies [1029580, 8fec7c3]:
    • @chestnutlabs/toolpath-core@0.4.0
    • @chestnutlabs/gcode-containers@0.4.0

0.3.0

Minor Changes

  • #238 75f9f2b Thanks @sobechestnut-dev! - Register .bgcode as a container adapter so it flows through the existing parser pipeline (DD-011 phase 4c, #188). A .bgcode file now "just works" through GcodeParseSession with containers: 'auto' — sniffed by magic, decoded to plain G-code, and parsed to the same IR as the plain .gcode (proven by the golden-equivalence test).

    • gcode-bgcode: openBgcodeContainer(bytes) implements the DD-005 §4.4 { id, sniff, open } shape (single plate; openPlate(0) streams the decoded G-code). openBgcode(bytes, { metadata: true }) now also decodes the metadata (INI) and thumbnail blocks, so the adapter surfaces machine geometry from bed_shape, whitelisted slicer settings (feeding dialect detection + provenance), and thumbnails.
    • gcode-parser: the batteries worker registers the bgcode adapter beside gcode-3mf.

    Verified end-to-end: a real Prusa XL cube .bgcode parses through the session to 11,417 segments with a 360×360 bed and printer_model metadata.

  • #233 83f0336 Thanks @sobechestnut-dev! - New package @chestnutlabs/gcode-bgcode — binary G-code (.bgcode) decode, phase 1 (DD-011, epic #188). A license-clean, in-memory block walker that decodes Prusa .bgcode to plain G-code for the existing parser/dialect/IR/renderer pipeline. Decode-only.

    Phase 1 ships the block walker + per-block CRC32 verification + sniffBgcode/openBgcode, with None and DEFLATE compression and None encoding decoded end-to-end. MeatPack (phase 2) and heatshrink (phase 3) return honest, structured ContainerErrors (E_BGCODE_UNSUPPORTED_ENCODING / E_BGCODE_UNSUPPORTED_COMPRESSION) until then. All failures — bad magic/version, CRC mismatch, truncation, decompression bomb — are bounded structured errors.

    Clean-room from the published spec (RR-003): no AGPL libbgcode/MeatPack code. Depends only on @chestnutlabs/toolpath-core and @chestnutlabs/gcode-containers (for crc32/ContainerError); no three, framework, filesystem, or network. Container-adapter + worker registration + metadata/ thumbnail surfacing + the plain-vs-.bgcode golden-equivalence test land in phase 4.

  • #235 bb3085a Thanks @sobechestnut-dev! - Binary G-code decode phase 3 (DD-011, #188): heatshrink decompression (windows 11 & 12, lookahead 4). With this, all four .bgcode compression codecs and all encodings the spec defines are decoded — a .bgcode GCode block compressed with heatshrink now decodes end-to-end to plain G-code.

    The decoder is a TypeScript port of the LZSS decoder from the ISC atomicobject/heatshrink (© 2013–2015 Scott Vokes) — attribution preserved, no AGPL libbgcode (RR-003 §8). It is validated against vectors built by an independent MSB-first bit-packer from the wire format (literal, single and multi-byte back-references, self-referential runs, window 11 & 12, a realistic repeated G-code fragment) plus block-level integration through openBgcode. Output is bounded (decompression-bomb defense). Container-adapter integration + real PrusaSlicer-file golden-equivalence follow in phase 4.

  • #234 8c0ee6e Thanks @sobechestnut-dev! - Binary G-code decode phase 2 (DD-011, #188): the MeatPack G-code encoding (both variants — MeatPack and MeatPack (comments preserved)). A .bgcode GCode block encoded with MeatPack now decodes end-to-end (optionally after DEFLATE) to plain G-code.

    The decoder is a faithful TypeScript port of the MIT jamesgopsill/meatpack unpacker (© 2025 James Gopsill), which is itself derived from the published Prusa spec — attribution preserved, and no AGPL libbgcode/OctoPrint-MeatPack code (RR-003 §8). It is validated against hand-computed vectors (the nibble table applied by hand as an independent oracle: packing, left/right/double full-width escapes, the newline special case, and the no-spaces + disable-packing commands), plus block-level integration through openBgcode. Output is bounded (decompression-bomb defense) and invalid command bytes are structured errors. heatshrink compression remains phase 3.

Patch Changes

  • #236 852db93 Thanks @sobechestnut-dev! - Fix the .bgcode DEFLATE flavor: it is zlib-wrapped, not raw (DD-011 phase 4 confirmation, #188). Verified against a real Prusa XL .bgcode file — its DEFLATE-compressed Slicer/Print metadata blocks decode only with the zlib header and fail as raw. (GCode blocks use heatshrink, so this was invisible until a real file's metadata was exercised.) The decoder now uses DecompressionStream('deflate'), and a flavor-lock test asserts a raw-DEFLATE block is rejected so this can't regress.

  • #237 f2e79e4 Thanks @sobechestnut-dev! - Add the .bgcode golden-equivalence killer test (DD-011 §D6, #188): a PrusaSlicer 2.9.6 primitive cube, committed in both .gcode and .bgcode, is decoded and parsed, and its IR is asserted byte-identical to the IR of the plain .gcode across every geometry channel (positions, extrusion, kind, tool, layer). This pins decode correctness against the already-trusted plain-G-code path and exercises the real Prusa codec stack end-to-end (heatshrink-12 + MeatPack comments/no-spaces, DEFLATE metadata, thumbnails, per-block CRC32). Test/fixtures only — no code or public-API change.

  • #239 b0ef69f Thanks @sobechestnut-dev! - .bgcode security hardening (DD-011 §7 + phase 5, #188): a deterministic adversarial fuzz corpus (adversarial.test.ts) — ~1600 inputs across pure-random bytes, GCDE-prefixed random, bit-flip mutations of a real file, and garbage payloads through every codec — asserting that only bounded, structured ContainerErrors ever escape (never a crash, unbounded allocation, or hang). Plus edge cases: an oversized declared size can't balloon memory, and unknown block types are walked past.

    Adds the §7 security-review record (docs/design/SECURITY-REVIEW-DD-011-bgcode.md, prepared for maintainer sign-off) and a decode benchmark (tools/benchmark/results/) with real numbers (cube: 15.7 ms; 21 MB Prusa XL ColorMix: 3.3 s decode → 51 MB, all 856 CRC32s verified). Tests + docs only.

  • Updated dependencies [39348de, d161e80, 82bd7ae]:

    • @chestnutlabs/toolpath-core@0.3.0
    • @chestnutlabs/gcode-containers@0.3.0