perf: hoist uniqueQueryTokens out of per-result loop in rankAnswerEvidence

[Copilot]

Summary

• uniqueQueryTokens(query) was called 2× per result in rankAnswerEvidence — once in answerEvidenceScore and again inside phraseScore — producing N×2 identical tokenizations for a constant query. For a typical 15-result ranking pass this meant ~30 redundant calls (each involving normalizedClinicalSearchTokens + regex + Set dedup).

• Tokens are now computed once in rankAnswerEvidence and threaded into answerEvidenceScore and phraseScore as a parameter:

// before: answerEvidenceScore called uniqueQueryTokens, then phraseScore called it again
score: answerEvidenceScore(query, result, queryClass)

// after: tokens computed once, passed through
const tokens = uniqueQueryTokens(query);
score: answerEvidenceScore(tokens, result, queryClass)

Verification

• npm run verify:cheap
• npm run verify:ui when UI, routing, styling, browser behavior, reduced-motion, or forced-colors behavior changed
• npm run verify:release before release or handoff confidence claims
• npm run format:check
• npm run check:production-readiness when clinical workflow, privacy, environment, Supabase, source governance, or deployment behavior changed

Clinical Governance Preflight

Complete this section when the change touches ingestion, answer generation, search/ranking, source rendering, document access, privacy, production env, or clinical output.

• Source-backed claims still require linked source verification before clinical use
• No patient-identifiable document workflow was introduced or expanded without explicit governance approval
• Supabase target remains Clinical KB Database (sjrfecxgysukkwxsowpy)
• Service-role keys and private document access remain server-only
• Demo/synthetic content remains clearly separated from real clinical sources
• Source metadata, review status, and outdated/unknown-source behavior remain conservative
• Deployment classification/TGA SaMD impact was checked when clinical decision-support behavior changed

Notes

• Pure refactor — no behaviour change. Ranking scores are identical; all 7 answer-ranking tests and 14 related tests pass unchanged.
• phraseScore signature changed from (query: string, haystack: string) to (tokens: string[], haystack: string) — private function, no external callers.

[Copilot]

Pull request overview

This PR optimizes answer-evidence ranking by avoiding repeated tokenization of a constant query during scoring, reducing redundant work per result while preserving scoring behavior.

Changes:

• Hoists uniqueQueryTokens(query) to rankAnswerEvidence and reuses the computed tokens for all results.
• Refactors answerEvidenceScore and phraseScore to accept precomputed tokens instead of the raw query string.

---

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.