[pradeeppeddineni]

Fixes #188.

The failing run was test_time_with_colons_gets_anonymized. The date offset is always a whole number of hours, so whenever a random seed lands on a whole number of days the time of day comes out the same and the new_time != "121558.123456" check fails. I measured it locally and the time came back unchanged in 36 of 1000 runs with a random seed, which lines up with roughly 1 in 24.

I switched the test to Anonymizer(seed=""), same as test_multivalued_date_and_time_pair_gets_anonymized does, and added a comment saying why. With that seed the offset is -6553 hours so the time always comes out as 111558.123456. I also added a check that the minutes and seconds are kept (new_time[2:] == "1558.123456"), which shows the colons got stripped and the value was really parsed rather than just mangled.

test only change, no code touched. uv run test.py passes (ruff format, ruff check, mypy, 328 tests).

[Copilot]

🟢 Approval recommended

The test-only change removes flaky equality outcomes and adds explicit checks that minutes, seconds, and fractional seconds are preserved.

No open issues 🎉

What changed in this PR

This pull request makes the colon-formatted DICOM time anonymization test deterministic and verifies that the time components are preserved.

Changes:

• Uses Anonymizer(seed="") to prevent intermittent equality with the original time.
• Verifies that minutes, seconds, and fractional seconds remain unchanged.
• Updates tests only; no production code is modified.

File	Description
tests/​test_anonymize_single_instance.py	Makes the time anonymization test deterministic and strengthens its output assertions.

_{Review effort: Lite}

[blairconrad]

Hi, @pradeeppeddineni. Thanks for the contribution! Sorry for the delay; I had been travelling and then... well, life.