bagx.eval

bagx.eval

Single bag quality evaluation engine.

Evaluates GNSS, IMU, and inter-topic sync quality, producing a structured quality report with a composite score.

EvalConfig dataclass

Configurable scoring thresholds for evaluation.

Users can tune these for their sensor suite via the Python API.

detect_domain_names(topics, custom_domains=None)

Return built-in plus custom domain names inferred from a bag report.

evaluate_bag(bag_path, output_json=None, config=None, custom_rules_path=None, include_anomaly=False)

Run full quality evaluation on a bag file.

When include_anomaly is True, anomaly detection runs alongside the quality eval and its temporal findings (fix-lost segments, IMU spike clusters, rate gaps) are merged into report.findings. This is the surface CI gates with bagx diff and bagx benchmark use to reason about segment-level readiness.

generate_recommendations(report)

Generate actionable recommendations based on eval results.

print_eval_report(report, console=None, *, findings_only=False, severity_min=None)

Pretty-print an evaluation report.

findings_only: when True, suppress sensor tables and recommendations and render the structured findings list instead. severity_min: when set, filter findings to this severity or higher (only affects the findings_only rendering).