fix(viewer): support cbor2 6.x deserialization (frozendict + tuple)

[Bahtya]

Summary

Fixes #2375

cbor2 6.0 introduces two breaking changes:

• Nested dicts return as cbor2.frozendict — a Mapping, but not a dict subclass
• Arrays return as tuple instead of list

The isinstance(data, dict) guards in viewer_file.py silently short-circuit on frozendict, so deserialization does nothing and recorder round-trip tests fail.

Changes

File	Change
newton/_src/viewer/viewer_file.py	Replace isinstance(x, dict) → isinstance(x, Mapping) in deserialize(), transfer_to_model(), the warp callback, and _resolve_cache_refs(). Also update _resolve_cache_refs to handle tuple the same as list (cbor2 6.x arrays), always returning a plain list.
pyproject.toml	Remove the <6 upper bound added as a workaround in #2373.

Test plan

• pytest newton/tests/ -k recorder passes with cbor2>=5.7.0,<6
• pytest newton/tests/ -k recorder passes with cbor2>=6.0 (once 6.0 is released)
• pip install 'cbor2>=5.7.0' installs without version conflict

Notes

Mapping is already imported from collections.abc at the top of the file, so no new imports are needed.

Bahtya

Summary by CodeRabbit

• Bug Fixes

  • Clarified install hint for a binary serialization dependency.

• Improvements

  • Broadened compatibility with newer versions of the binary serialization library.
  • Improved handling of varied mapping types during binary deserialization and data transfer, increasing robustness across frozen/immutable mappings.

[linux-foundation-easycla]

• ❌ The email address for the commit (21de760, fd4853c) is not linked to the GitHub account, preventing the EasyCLA check. Consult this Help Article and GitHub Help to resolve. (To view the commit's email address, add .patch at the end of this PR page's URL.) For further assistance with EasyCLA, please submit a support request ticket.

[coderabbitai]

No actionable comments were generated in the recent review. 🎉

ℹ️ Recent review info

⚙️ Run configuration

Configuration used: Path: .coderabbit.yml

Review profile: CHILL

Plan: Pro

Run ID: 7ef22f66-88b9-4bec-9f9b-ed0dd2d5af97

📥 Commits

Reviewing files that changed from the base of the PR and between fd4853c and 21de760.

📒 Files selected for processing (1)

• newton/_src/viewer/viewer_file.py

🚧 Files skipped from review as they are similar to previous changes (1)

• newton/_src/viewer/viewer_file.py

---

📝 Walkthrough

Walkthrough

Updated viewer deserialization to accept generic Mapping types (e.g., frozendict) and tuples from cbor2 6.x; resolver now returns plain dict for resolved mappings. The cbor2 example dependency upper-version bound was removed.

Changes

Cohort / File(s)	Summary
Viewer deserialization newton/_src/viewer/viewer_file.py	Generalized checks from dict to typing.Mapping across transfer_to_model, deserialize, depointer_as_key, and _resolve_cache_refs; treat cache-ref containers as any Mapping, recurse mappings, and return plain dict for resolved mappings. Adjusted missing-cbor2 install hint for .bin to pip install 'cbor2>=5.7.0'.
Dependency constraint pyproject.toml	Removed the upper-bound <6 from the examples optional dependency for cbor2, changing cbor2>=5.7.0, <6 → cbor2>=5.7.0.

Estimated code review effort

🎯 3 (Moderate) | ⏱️ ~20 minutes

🚥 Pre-merge checks | ✅ 4 | ❌ 1

❌ Failed checks (1 warning)

Check name	Status	Explanation	Resolution
Docstring Coverage	⚠️ Warning	Docstring coverage is 66.67% 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
Description Check	✅ Passed	Check skipped - CodeRabbit’s high-level summary is enabled.
Title check	✅ Passed	The title accurately and specifically describes the main change: supporting cbor2 6.x deserialization by handling frozendict and tuple returns.
Linked Issues check	✅ Passed	The PR successfully addresses all coding requirements from #2375: replaces isinstance(x, dict) with Mapping checks, handles tuple arrays, and removes the cbor2 <6 constraint.
Out of Scope Changes check	✅ Passed	All changes are directly scoped to the linked issue requirements: viewer_file.py refactoring for Mapping support and pyproject.toml dependency constraint removal.

_{✏️ Tip: You can configure your own custom pre-merge checks in the settings.}

✨ Finishing Touches

🧪 Generate unit tests (beta)

• Create PR with unit tests

---

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 and usage tips.}

[coderabbitai]

Actionable comments posted: 1

🤖 Prompt for all review comments with AI agents

Verify each finding against the current code and only fix it if needed.

Inline comments:
In `@newton/_src/viewer/viewer_file.py`:
- Around line 1014-1016: The current _resolve_cache_refs post-pass
unconditionally converts tuples to lists; change it to preserve real tuples by
branching on type: if obj is a list, return a new list with _resolve_cache_refs
applied to each element; if obj is a tuple, return a tuple constructed from
applying _resolve_cache_refs to each element (so genuine tuple-valued fields
keep their type). Keep the existing behavior for other iterables unchanged and
reference the _resolve_cache_refs function (and the surrounding deserialize
flow) when making the change.

🪄 Autofix (Beta)

Fix all unresolved CodeRabbit comments on this PR:

• Push a commit to this branch (recommended)
• Create a new PR with the fixes

---

ℹ️ Review info

⚙️ Run configuration

Configuration used: defaults

Review profile: CHILL

Plan: Pro

Run ID: aa50d953-209f-471d-997c-cdf2f1e10542

📥 Commits

Reviewing files that changed from the base of the PR and between 8d7a6e2 and fd4853c.

📒 Files selected for processing (2)

• newton/_src/viewer/viewer_file.py
• pyproject.toml

[Bahtya]

Thanks for the review! I've addressed the CodeRabbit feedback:

Preserve real tuples in _resolve_cache_refs: The original code converted both list and tuple to a list, which meant genuine tuple-valued fields would round-trip as a different type. I've now split the check so that list objects resolve to list and tuple objects resolve to tuple, preserving the original container type through the post-pass.

[Bahtya]

Friendly ping 👋 This PR adds support for cbor2 6.x deserialization by handling frozendict and other type changes. CodeRabbit review found no issues. Would appreciate a review.