Sanitised traceability pipeline

The internal traceability scripts contain rich examples of DuckDB syntax, but those flows reference tables and paths that cannot be published verbatim. This walkthrough mirrors the original data movements with anonymised relations and is backed by executable tests so the documentation stays in sync with the code.

Tip

Each code listing below is pulled directly from tests/test_examples_traceability_pipeline.py. The tests double as a smoke suite while guaranteeing the documentation never diverges from reality.

Ranking anonymised programs

The first helper demonstrates how to combine catalogue fragments with recent log entries. It uses CTEs to capture the probe barcode, correlate matching fragments, and prioritise results using ROW_NUMBER.

Collecting companion barcodes

The panel lookup mirrors the private alternate search logic with neutral column names. It performs a panel-based join and unions results from an alternate capture stream when available.

Repairing unit costs

The final example rebuilds the rename-and-aggregate pipeline that restores missing costs. It aggregates the latest price snapshots, computes fallbacks per item, and merges the healed rows with the untouched events.