feat(tensor): bit-granular read cursor for byteConversionAppend + pub…

[LeoBuron]

…lic unpackSignExtend

The append loop could start WRITING at any bit position (dstStartBit) but always started READING at bit 0 of the source byte pointer — decoding a packed mixed-width stream (DeltaSym: segments rarely start byte-aligned) was impossible.

• byteConversionAppend gains a trailing srcStartBit: the input cursor is seeded the same way as the output cursor (index srcStartBit/8, phase srcStartBit%8). Reads exactly numValues*dataInBits bits, so segments may also END mid-byte. byteConversion delegates with (0, 0).
• unpackSignExtend is now public (TensorConversion.h) with a srcStartBit parameter, so DeltaSym-style decoders can sign-extend a segment that starts mid-byte; internally switched from byteConversion to byteConversionAppend (clear-then-set defines all widened bits, no memset).

[coderabbitai]

No actionable comments were generated in the recent review. 🎉

ℹ️ Recent review info

⚙️ Run configuration

Configuration used: Organization UI

Review profile: CHILL

Plan: Pro Plus

Run ID: d7820690-45c4-48bf-a9a9-bf5b04d42b67

📥 Commits

Reviewing files that changed from the base of the PR and between 9de065f and 927db2b.

📒 Files selected for processing (7)

• docs/conventions/tensor.md
• src/tensor/Tensor.c
• src/tensor/TensorConversion.c
• src/tensor/include/Tensor.h
• src/tensor/include/TensorConversion.h
• test/unit/tensor/UnitTestTensor.c
• test/unit/tensor/UnitTestTensorConversion.c

---

📝 Walkthrough

Walkthrough

The tensor conversion bridge now accepts independent source and destination bit offsets. Signed unpacking exposes srcStartBit, uses append-based widening, and has expanded unit coverage and documentation.

Changes

Bit-Offset Conversion

Layer / File(s)	Summary
Conversion contracts src/tensor/include/Tensor.h, src/tensor/include/TensorConversion.h, docs/conventions/tensor.md	byteConversionAppend documents and accepts srcStartBit; unpackSignExtend is declared with mid-byte decoding semantics and documented sign-extension behavior.
Conversion implementation src/tensor/Tensor.c, src/tensor/TensorConversion.c	Packing and unpacking derive source and destination bit positions, while tensor conversion callers pass the updated arguments.
Offset conversion validation test/unit/tensor/UnitTestTensor.c, test/unit/tensor/UnitTestTensorConversion.c	Tests cover offset reads, offset transcoding, narrowing, mixed-width round trips, and signed decoding at nonzero offsets.

Estimated code review effort: 3 (Moderate) | ~20 minutes

Sequence Diagram(s)

sequenceDiagram
  participant TensorConversion
  participant unpackSignExtend
  participant byteConversionAppend
  TensorConversion->>unpackSignExtend: decode packed signed values
  unpackSignExtend->>byteConversionAppend: read from srcStartBit
  byteConversionAppend-->>unpackSignExtend: widened values
  unpackSignExtend-->>TensorConversion: sign-extended int32 outputs

Loading

Poem

I’m a rabbit with bits in a row,
Hopping where source and destinations go.
Mid-byte I munch, then widen with care,
Signed little values bloom in the air.
Tests thump their paws: the offsets align!

🚥 Pre-merge checks | ✅ 4 | ❌ 1

❌ Failed checks (1 warning)

Check name	Status	Explanation	Resolution
Docstring Coverage	⚠️ Warning	Docstring coverage is 76.47% which is insufficient. The required threshold is 80.00%.	Write docstrings for the functions missing them to satisfy the coverage threshold.

✅ Passed checks (4 passed)

Check name	Status	Explanation
Title check	✅ Passed	The title clearly points to the main changes: bit-granular reads for byteConversionAppend and making unpackSignExtend public.
Description check	✅ Passed	The description is directly related to the PR and accurately summarizes the byteConversionAppend and unpackSignExtend changes.
Linked Issues check	✅ Passed	Check skipped because no linked issues were found for this pull request.
Out of Scope Changes check	✅ Passed	Check skipped because no linked issues were found for this pull request.

✨ Finishing Touches

📝 Generate docstrings

• Create stacked PR
• Commit on current branch

🧪 Generate unit tests (beta)

• Create PR with unit tests
• Commit unit tests in branch feat/bit-granular-read-cursor

---

Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out.

❤️ Share

• X
• Mastodon
• Reddit
• LinkedIn

_{Comment @coderabbitai help to get the list of available commands.}