chore: Exporters refactor

[marta-sd]

No description provided.

[copy-pr-bot]

This pull request requires additional validation before any workflows can run on NVIDIA's runners.

Pull request vetters can view their responsibilities here.

Contributors can view more details about this message here.

[pribalta]

Power Review

Reviewed 30 files, left 21 suggestions: 2 errors, 13 warnings, 6 suggestions.

Dimensions Analyzed

Dimension	🔴	⚠️	💡
anti_patterns	0	4	1
architecture	1	3	1
patterns	1	1	1
security-risks	0	3	0
testing-quality	0	2	3
Total	2	13	6

---

A pull request for exporters refactor arrives,
Our review engine carefully contrives.
Still, only you can see the bigger view,
So weigh these findings before you approve.

---

Automated review by Power Review — findings are advisory.

[pribalta]

[Power Review] 💡 Identical artifact file setup code duplicated across all TestLocalExporterManualScenarios methods · 88% confidence

Every test in TestLocalExporterManualScenarios (test_successful_export_multiple_jobs, test_partial_failure_missing_artifacts, test_non_existing_job_id, test_remote_job_export_with_copy, test_json_format_updates_existing_file) copies the same two write_text calls verbatim for run_config.yml and results.yml. This is the Duplicated Test Setup anti-pattern: shared setup should be extracted into a fixture or helper to reduce maintenance burden and the risk of individual tests drifting out of sync.

💡 Suggestion: Extract the artifact-file creation into a local helper, e.g.:

def _write_standard_artifacts(artifacts_dir: Path) -> None:
    (artifacts_dir / 'run_config.yml').write_text(
        'framework_name: test-harness\nconfig:\n  type: test_task\n'
        'target:\n  api_endpoint:\n    model_id: test-model\n'
    )
    (artifacts_dir / 'results.yml').write_text(
        'results:\n  tasks:\n    test_task:\n      metrics:\n'
        '        accuracy:\n          scores:\n            accuracy:\n              value: 0.9\nconfig: {}\n'
    )

Or add a pytest fixture to conftest.py.

[marta-sd]

TODO

[pribalta]

[Power Review] 💡 Identical artifact directory setup duplicated across all MLflow exporter test methods · 85% confidence

Five tests (test_export_job_ok, test_export_with_skip_existing, test_export_with_update_existing, test_log_config_params_flattens_config, test_log_config_params_with_max_depth) each copy the same three-step pattern: mkdir the artifacts directory, write run_config.yml, write results.yml with identical content. This is textbook Duplicated Test Setup. Extracting it into a fixture would shorten each test and ensure the artifact format is consistent.

💡 Suggestion: Add a pytest fixture (possibly extending the existing prepare_local_job fixture in conftest.py) that creates and returns a pre-populated artifacts directory. Each test can then request this fixture instead of repeating the setup inline.