fix: Synthesize category names for nested columns

[Oliver Borchert (borchero)]

Motivation

Nested categoricals specified with a physical dtype do not receive a schema or column name. For example, dy.List(dy.Categorical(pl.UInt16)) raises an assertion when its dtype is accessed; UInt32 silently uses global categories.

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 100.00%. Comparing base (9271774) to head (cda0aec).

Additional details and impacted files

@@            Coverage Diff            @@
##              main      #403   +/-   ##
=========================================
  Coverage   100.00%   100.00%           
=========================================
  Files           46        46           
  Lines         2598      2615   +17     
=========================================
+ Hits          2598      2615   +17     

☔ View full report in Codecov by Harness.
📢 Have feedback on the report? Share it here.

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.

[Copilot]

🔵 Needs a closer look

The categorical test has unresolved assertion-coverage and testing-convention issues.

Pull request overview

Fixes schema-scoped names for nested physical categorical dtypes through recursive column binding.

Changes:

• Recursively binds nested list, array, and struct columns.
• Copies nested definitions to isolate metadata.
• Expands categorical naming tests and documentation.

File summaries

File	Summary
tests/column_types/test_categorical.py	Adds nested naming coverage; test setup and accessor assertions need refinement.
dataframely/columns/struct.py	Recursively binds struct fields.
dataframely/columns/list.py	Recursively binds list elements.
dataframely/columns/categorical.py	Documents nested categorical naming.
dataframely/columns/array.py	Recursively binds array elements.
dataframely/columns/_base.py	Defines nested binding behavior.
dataframely/_base_schema.py	Uses recursive column binding.

Review details

Suppressed comments (2)

tests/column_types/test_categorical.py:39

• Because both tuple entries reference the same mutable top-level Column, TestSchema.columns() rebinds the object before the loop starts. The assertions therefore only exercise the final binding and cannot detect a regression in either access path independently; invoke and assert each accessor before invoking the other (for example, iterate over lambdas).

    for bound_column in (TestSchema.a, TestSchema.columns()["a"]):
        dtype: PolarsDataType = bound_column.dtype
        while isinstance(dtype, pl.List | pl.Array | pl.Struct):
            dtype = (
                dtype.fields[0].dtype if isinstance(dtype, pl.Struct) else dtype.inner

tests/column_types/test_categorical.py:35

• This new test does not separate its setup, dtype evaluation, and assertions into the required Arrange/Act/Assert blocks. Please mark those phases explicitly and keep the assertions in the Assert block so the test follows the repository testing convention.

def test_synthesized_categories_name(column: dy.Column, expected_name: str) -> None:
    class TestSchema(dy.Schema):
        a = column

    for bound_column in (TestSchema.a, TestSchema.columns()["a"]):

• Files reviewed: 7/7 changed files
• Comments generated: 0
• Review effort level: Lite

---

💡 Add a code-review agent skill or configure MCP servers for context-aware, tailored reviews. Learn more in the docs.