Transaction-Science/open-standards
Watch
1
0
Fork
You've already forked open-standards
 0
Code Issues Pull requests Projects Releases 15 Packages Wiki Activity Actions

wai-rs: jpegai_bitstream — codestream container + MSB-first header bit reader #16

Merged
dcharlot merged 1 commit from feat/jpegai-bitstream into main 2026-07-16 04:52:46 -04:00
Conversation 0 Commits 1 Files changed 3 +294
dcharlot commented 2026-07-16 04:52:45 -04:00
Owner
Copy link

The decoder entry point. A codestream is a sequence of marker segments: marker(2B BE) + expgolomb_k0(size) + byte-align + payload. Header substreams (PIH/TON) are plain bit-packed, read MSB-first — headers are NOT arithmetic-coded (only latent substreams are, → jpegai_mans).

  • BitReader — faithful port of the reference EcLibDirect::read_bits (MSB-first, bit-granular cursor)
  • read_expgolomb_k0 — segment payload lengths
  • parse_container — marker walk → [Segment]

Validated on the real test256.bin codestream: full layout SOC/PIH(49)/TON(63)/RDI(1)/SORP(97)/SORS(47)/SOZ(83)/EOC, picture-header prefix → 256x256, 4:4:4, colour_transform_idx=1. Cross-check: SORP/SORS/SOZ sizes (97/47/83) equal the mANS mem lengths — container parse and byte-exact entropy decode agree on the same bytes.

The decoder entry point. A codestream is a sequence of marker segments: `marker(2B BE) + expgolomb_k0(size) + byte-align + payload`. Header substreams (PIH/TON) are plain bit-packed, read MSB-first — headers are NOT arithmetic-coded (only latent substreams are, → jpegai_mans). - `BitReader` — faithful port of the reference `EcLibDirect::read_bits` (MSB-first, bit-granular cursor) - `read_expgolomb_k0` — segment payload lengths - `parse_container` — marker walk → `[Segment]` **Validated on the real test256.bin codestream:** full layout SOC/PIH(49)/TON(63)/RDI(1)/SORP(97)/SORS(47)/SOZ(83)/EOC, picture-header prefix → 256x256, 4:4:4, colour_transform_idx=1. **Cross-check: SORP/SORS/SOZ sizes (97/47/83) equal the mANS `mem` lengths** — container parse and byte-exact entropy decode agree on the same bytes.
The decoder's entry point. A codestream is a sequence of marker segments; each (except
the marker-only SOC/EOC) is `marker(2B big-endian) + expgolomb_k0(size) + byte-align +
payload`. The header substreams (PIH/TON) are plain bit-packed and read MSB-first — the
headers are NOT arithmetic-coded (only the latent substreams are, and those go to
jpegai_mans).

  BitReader        faithful port of the reference EcLibDirect::read_bits — MSB-first
                   within each byte, accumulated MSB-first, bit-granular cursor, reads
                   past end return the available bits (matching the bounds check)
  read_expgolomb_k0  unsigned exp-Golomb k=0, for the segment payload lengths
  parse_container  walk the markers -> [Segment{marker, start, len}]

Validated against the real test256.bin codestream: full segment layout
SOC/PIH(49)/TON(63)/RDI(1)/SORP(97)/SORS(47)/SOZ(83)/EOC, and the picture-header prefix
decodes to 256x256, 4:4:4, colour_transform_idx=1, level 52. Cross-check: the
SORP/SORS/SOZ payload sizes (97/47/83) equal the `mem` lengths the byte-exact mANS
decoder consumes — the container parse and the entropy decode agree on the same bytes.

Co-Authored-By: Claude Opus 4.8 <[email protected]>
Sign in to join this conversation.
Reviewers
No reviewers
Labels
Clear labels
No items
No labels
Milestone
Clear milestone
No items
No milestone
Projects
Clear projects
No items
No project
Assignees
Clear assignees
No assignees
1 participant
Notifications
Due date
The due date is invalid or out of range. Please use the format "yyyy-mm-dd".

No due date set.

Dependencies

No dependencies set

Reference
Transaction-Science/open-standards!16
No description provided.