test: add unit tests for prepare_benchmark_for_insert field coverage

[anfredette]

Context

PR #190 fixed a bug where model_uri was missing from the _INSERT_QUERY in loader.py. This class of bug (missing column in the INSERT query) is easy to introduce because there are no unit tests that verify prepare_benchmark_for_insert produces a dict with all the keys referenced by _INSERT_QUERY.

Proposal

Add unit tests for prepare_benchmark_for_insert in tests/unit/ that:

1. Assert that calling prepare_benchmark_for_insert with minimal input includes all expected keys (including model_uri)
2. Assert that optional fields default to None when not provided
3. Assert that provided values (like model_uri) are preserved
4. Bonus: Add a structural test that parses _INSERT_QUERY for %(...)s parameter names and asserts they're all present in the output of prepare_benchmark_for_insert — this would catch any future column additions that forget to update the preparation function

Example

def test_prepare_benchmark_defaults_model_uri_to_none():
    bench = {"model_hf_repo": "org/model", "hardware": "A100",
             "hardware_count": 1, "prompt_tokens": 512, "output_tokens": 256}
    result = prepare_benchmark_for_insert(bench)
    assert result["model_uri"] is None

def test_prepare_benchmark_preserves_model_uri():
    bench = {"model_hf_repo": "org/model", "hardware": "A100",
             "hardware_count": 1, "prompt_tokens": 512, "output_tokens": 256,
             "model_uri": "oci://registry.example.com/model:latest"}
    result = prepare_benchmark_for_insert(bench)
    assert result["model_uri"] == "oci://registry.example.com/model:latest"

def test_all_insert_query_params_present_in_prepared_output():
    """Structural test: every %(key)s in _INSERT_QUERY must appear in prepared output."""
    import re
    from planner.knowledge_base.loader import _INSERT_QUERY, prepare_benchmark_for_insert
    
    params = set(re.findall(r'%\((\w+)\)s', _INSERT_QUERY))
    bench = {"model_hf_repo": "org/model", "hardware": "A100",
             "hardware_count": 1, "prompt_tokens": 512, "output_tokens": 256}
    result = prepare_benchmark_for_insert(bench)
    missing = params - set(result.keys())
    assert not missing, f"Keys in _INSERT_QUERY but missing from prepared output: {missing}"

Related

• PR fix: include model_uri column in benchmark data loader INSERT query #190: fix that motivated this issue