Building a user interface

sci_etl_core.discovery holds the read-model a presentation layer renders: DiscoveryResult (the query text and chips, fused hits, graph, facets, matched count, elapsed time, and degraded and skipped legs) and Facet. Both are frozen dataclasses, and importing the module loads no store, no event-loop machinery, and no optional dependency.

Render FusedHit.snippets with your own markup from their highlights, one line per field, and label a hit whose lexical_rank is None as a match by meaning. Build date or year histograms from range_counts, which leaves every bucket visible while one is selected.

Keep parsing on the keystroke and make only retrieval asynchronous: debounce it, cancel a search when a newer one starts, and drop any result that arrives after a newer search began.

Before sharing stores between a UI and ingest runs, read Store ownership.