Fix AttributeError on Python 3.9 when generating invalid attribute error message

[Jgprog117]

Summary

Fixes #4821

On Python 3.9, when an invalid attribute value is passed that cannot be converted to a string, the code raises AttributeError instead of the intended TypeError. This occurs because the error message generation attempts to access __name__ on all types in _VALID_ANY_VALUE_TYPES, but typing.Mapping (and other generic types) don't have a __name__ attribute in Python 3.9.

Changes

• Updated _clean_extended_attribute_value() in opentelemetry-api/src/opentelemetry/attributes/init.py:193 to use getattr(valid_type, '__name__', getattr(valid_type, '_name', None)) for safe attribute access
• Added test case test_invalid_type_error_message() to verify that invalid types that cannot be stringified return None without raising AttributeError

Implementation

The fix follows the suggestion from @xrmx in the issue discussion, using nested getattr() calls to:

1. First try to get __name__ attribute
2. Fall back to _name attribute if __name__ doesn't exist
3. Return None if neither attribute exists

This ensures that the error message generation works correctly across all Python versions, including Python 3.9 where generic types from typing module don't have __name__.

Test plan

• Added test case that creates an object which cannot be converted to string
• This triggers the TypeError path where the error message is generated
• On Python 3.9 with the old code, this would fail with AttributeError
• With the fix, it correctly returns None and logs a warning

The CI tests will run this on Python 3.9-3.14 to verify the fix works across all supported versions.

[linux-foundation-easycla]

The committers listed above are authorized under a signed CLA.

• ✅ login: Jgprog117 / name: Jgprog117 (5491829)

[xrmx]

Thanks for the PR but there is already a PR for that issue in #4876