[codex] Support POD struct array fields

[isVoid]

Summary

Adds support for POD struct data-model fields with fixed-size arrays, multidimensional arrays, nested POD structs, and arrays of nested POD structs. Fixes float[2][12] so it models two rows of twelve floats.

Also keeps nested struct binding deduplication scoped to fully qualified record names, so same short names in different nested scopes do not collide.

Validation

• pre-commit run --all-files passed.
• Focused struct/model tests passed: 5 passed.
• python -m compileall -q numbast/src/numbast/struct.py numbast/tests/test_struct.py passed.
• git diff --check passed.

Full numbast/tests/test_struct.py cannot complete here because CUDA kernel launch tests fail with CUDA_ERROR_NO_DEVICE.

Summary by CodeRabbit

• New Features

  • Multidimensional fixed-size C/C++ arrays now map correctly to nested Numba tuple types; recursive binding of nested POD structs improved with duplicate-name disambiguation.

• Documentation

  • Added guidance and examples showing POD struct field mappings, array/nested-struct cases, and C declaration row/column ordering.

• Tests

  • New tests validating multidimensional array conversions, POD struct layouts, nested fields, and name-disambiguation.

[coderabbitai]

No actionable comments were generated in the recent review. 🎉

ℹ️ Recent review info

⚙️ Run configuration

Configuration used: Path: .coderabbit.yaml

Review profile: ASSERTIVE

Plan: Enterprise

Run ID: c77756bf-9e89-4bce-85a0-3426344c0129

📥 Commits

Reviewing files that changed from the base of the PR and between 6a8ba9a and 13a8501.

📒 Files selected for processing (1)

• numbast/tests/test_struct.py

---

📝 Walkthrough

Walkthrough

Adds multi-dimensional fixed-size array parsing and mapping to nested Numba UniTuple types, recursive nested POD struct binding with duplicate-name deduplication, tests validating type/LLVM layouts, and documentation describing POD struct field modeling and multidimensional ordering.

Changes

Multi-dimensional Fixed-size Array Support for POD Struct Fields

Layer / File(s)	Summary
Multi-dimensional array parsing and type conversion numbast/src/numbast/types.py, numbast/src/numbast/static/types.py	Adds regexes and parse_fixed_size_array_type to extract base type and all extents. Updates to_numba_type to construct nested nbtypes.UniTuple layers and updates to_numba_type_str to rebuild nested UniTuple(...) strings for multi-dimensional fixed-size arrays.
Nested struct binding and deduplication numbast/src/numbast/struct.py	Adds helpers to detect injected-class-name records and normalize parsed records, tracks _bound_struct_names to avoid duplicate bindings, registers both name and qual_name in C2N, broadens alias application keys, and invokes recursive nested-struct binding after registration.
Test coverage for array types and POD struct layouts numbast/tests/test_struct.py	Adds imports, _bind_pod_structs_from_source helper, and tests asserting multi-dimensional UniTuple preservation, correct LLVM array element shapes for POD fields (1D and multi-D), nested struct layouts including arrays of nested structs, and qualified-name disambiguation for duplicate short names.
Documentation of POD struct field support docs/source/supported_declarations.rst	Documents that StructModel supports POD fields (fixed-size arrays, multi-dimensional arrays, nested POD structs, arrays of nested POD structs), provides a C++→Numba field mapping table, and clarifies C declaration ordering for multidimensional arrays (e.g., float transform[2][12]).

Estimated code review effort

🎯 3 (Moderate) | ⏱️ ~25 minutes

Poem

🐰 In structs I hop through arrays and rows,
Count dimensions where the UniTuple grows,
I bind each nested leaf and name each kind,
So C++ layouts map true in my mind,
Hopping tests and docs—clear paths I compose.

🚥 Pre-merge checks | ✅ 4 | ❌ 1

❌ Failed checks (1 warning)

Check name	Status	Explanation	Resolution
Docstring Coverage	⚠️ Warning	Docstring coverage is 33.33% 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 pull request title directly and specifically describes the main change: adding support for POD struct array fields, which is the core objective across all modified files.
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.

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

✨ 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 codex/pod-struct-binding-gaps

---

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

[github-actions]

Doc Preview CI
🚀 View pre-built docs at
https://NVIDIA.github.io/numbast/pr-preview/pr-343/
Preview will be ready when GitHub Pages deployment finishes.

[coderabbitai]

Actionable comments posted: 2

🤖 Prompt for all review comments with AI agents

Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.

Inline comments:
In `@numbast/src/numbast/struct.py`:
- Around line 462-463: The code currently deduplicates nested struct bindings
using short names stored in bound_struct_names/_bound_struct_names which causes
incorrect skips for same short names in different scopes; update all checks and
mutations that refer to nested_names, bound_struct_names, and
_bound_struct_names to operate on fully qualified names only (compute or use the
qualified name set instead of short names when testing with nested_names and
when adding entries), and replace the places that currently store short names
(the block that populates _bound_struct_names / bound_struct_names around the
later storage) so they insert qualified names; apply the same change to the
other similar checks mentioned (the blocks around the nested binding checks and
the storage of bound names).

In `@numbast/tests/test_struct.py`:
- Around line 89-99: The test currently picks the bound struct by positional
index from _bind_pod_structs_from_source (e.g., PodArrayFields =
_bind_pod_structs_from_source(...)[0]), which relies on parse ordering; change
this to lookup by struct name instead: call _bind_pod_structs_from_source(...)
and find the struct whose name equals "PodArrayFieldsForNumbast" (or the
intended binding name) and assign that to PodArrayFields; apply the same
name-based selection for the other occurrences around lines 110–124 to make the
tests deterministic.

🪄 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: Path: .coderabbit.yaml

Review profile: ASSERTIVE

Plan: Enterprise

Run ID: 871c04ef-e06c-4010-b99d-5aae3af8b3be

📥 Commits

Reviewing files that changed from the base of the PR and between 266b9df and af7e7ea.

📒 Files selected for processing (5)

• docs/source/supported_declarations.rst
• numbast/src/numbast/static/types.py
• numbast/src/numbast/struct.py
• numbast/src/numbast/types.py
• numbast/tests/test_struct.py

[coderabbitai]

Actionable comments posted: 1

🤖 Prompt for all review comments with AI agents

Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.

Inline comments:
In `@numbast/tests/test_struct.py`:
- Around line 107-111: The current assertion comparing LLVM IR types by string
in the test (assert [str(ty) for ty in llvm_ty.elements] == [...]) is fragile;
replace it with direct structural comparisons using llvmlite.ir type
constructors: build expected types with ir.ArrayType(ir.IntType(32), 20),
ir.ArrayType(ir.ArrayType(ir.FloatType(), 12), 2), and
ir.ArrayType(ir.FloatType(), 12) and assert equality against llvm_ty.elements
(e.g., assert llvm_ty.elements[0] == ir.ArrayType(ir.IntType(32), 20), etc.),
doing the same for the other occurrences at the noted assertions so tests
compare types structurally instead of by string.

🪄 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: Path: .coderabbit.yaml

Review profile: ASSERTIVE

Plan: Enterprise

Run ID: 7bc4acf4-c3db-428b-8ea3-49ab4a3c7918

📥 Commits

Reviewing files that changed from the base of the PR and between f301a90 and 6a8ba9a.

📒 Files selected for processing (1)

• numbast/tests/test_struct.py