[Qrytics]

What

cmd2.utils.get_types documents that it raises ValueError when its input
can't be processed, but an unresolvable string annotation (e.g. a forward
reference under from __future__ import annotations, or a TYPE_CHECKING
-only import) leaks a raw NameError (or SyntaxError for invalid
annotation strings) out of inspect.get_annotations(..., eval_str=True).
That surfaces as a confusing crash when registering hooks/commands instead
of the documented error type.

Repro

from cmd2.utils import get_types

def foo(x: "NotDefinedAnywhere") -> None: ...
get_types(foo)  # NameError: name 'NotDefinedAnywhere' is not defined

Fix

Catch (NameError, SyntaxError) around the get_annotations call and
re-raise as ValueError naming the offending callable, mirroring the
existing TypeError handling in the same function. No success path
changes.

Tests

Added test_get_types_unresolvable_forward_reference in
tests/test_utils.py (fails before with NameError, passes after);
full test_utils.py get_types selection passes, and
ruff check/ruff format are clean.

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 99.75%. Comparing base (4e98a38) to head (619a7e2).

Additional details and impacted files

@@           Coverage Diff           @@
##             main    #1806   +/-   ##
=======================================
  Coverage   99.75%   99.75%           
=======================================
  Files          23       23           
  Lines        6593     6596    +3     
=======================================
+ Hits         6577     6580    +3     
  Misses         16       16           

Flag	Coverage Δ
unittests	99.75% <100.00%> (+<0.01%)	⬆️

Flags with carried forward coverage won't be shown. Click here to find out more.

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

[tleonhardt]

@Qrytics
Thanks for the PR Mario! Wrapping the error and naming the offending callable is the right direction, and much better than leaking a raw NameError. A few changes requested inline. One more that has no line to attach to:

• CHANGELOG: please add a bug-fix entry to CHANGELOG.md under a new in-progress version heading (4.4.1 is already released).

[tleonhardt]

This existing handler also swallows a TypeError raised while evaluating an annotation. For example, def b(x: "int[str]") reports "Argument passed to get_types should be a function or method", which is wrong.

Action: distinguish the two cases, e.g. check callable(func_or_method) up front for the invalid-argument error, then treat any TypeError from get_annotations as a resolution failure.

[tleonhardt]

AttributeError still leaks: def a(x: "typing.NoSuch") raises AttributeError: module 'typing' has no attribute 'NoSuch'.

Action: add AttributeError to this tuple. _resolve_func_hints in cmd2/annotated.py catches (NameError, AttributeError, TypeError) for the same failure.

[tleonhardt]

Consider raising TypeError here instead of ValueError. The only callers of get_types are the hook-registration validators in cmd2/cmd2.py, which raise TypeError for every other annotation problem, and _resolve_func_hints in cmd2/annotated.py already raises TypeError("Failed to resolve type hints for ...") for this exact failure.

Action: raise TypeError for unresolvable annotations and update the :raises docstring to match.

[tleonhardt]

Only the NameError path is tested.

Action: add cases (parametrizing this test would work well) for:

• invalid syntax, e.g. "1 +"
• a bad attribute, e.g. "typing.NoSuch"
• a TypeError during evaluation, e.g. "int[str]"

Please also add one test that registers a hook with an unresolvable annotation (e.g. via register_postparsing_hook) and asserts the resulting error.