pymat.search(query) — fuzzy search over the domain library (incl. grades)

[gerchowl]

Summary

pymat.search(query: str, *, limit: int = 10) -> list[Material] as a top-level fuzzy-find over the loaded library. Complements pymat.vis.search(...) (visual catalog) with a symmetric domain-side verb.

Design

Tokenized + weighted-target matching over a flat registry — grades, variants, vendors are already first-class:

• registry key: weight 10
• Material.name / grade: weight 5
• hierarchy parent names: weight 3

All tokens must match somewhere (conjunctive). Ties broken by shorter key.

Triggers load_all() so results are exhaustive. Case-insensitive.

Examples

pymat.search("stainless")         # → [stainless, s304, s316L, ...]
pymat.search("316")               # → [s316L, s316, ...]
pymat.search("stainless 316")     # → [s316L]   # all tokens must match
pymat.search("lyso ce saint")     # → [prelude420, ...]   # deep hierarchy

Out of scope for v1

• Scoped search (kind="grade" / kind="parent") — optional follow-up.
• Aliases / synonyms — punt.
• Levenshtein via rapidfuzz — substring + token is enough. Optional extra later if needed.
• Property filters ("density > 8") — different API.

Related

• build123d#1270 — motivates the shape-side lookup shorthand.
• Vis.to_gltf() / .to_threejs() / .export_mtlx() method sugar #78 — Vis adapter-method sugar (same discoverability-without-architectural-change pattern).
• mat-vis#93 — visual-side architectural endgame; orthogonal.