Optimize slow code paths in AI service, database, and React components

[Copilot]

Performance audit identified several inefficient patterns: O(n) priority queue insertion, repeated toLowerCase() calls in hot paths, sequential awaits where parallelism is possible, and missing React memoization.

Backend

• AIService.js: Cache lowercase keywords in WeakMap instead of mutating trigger objects. Pre-filter events by chapter range before iteration.

  // Before: O(n) toLowerCase on every call
  trigger.keywords?.some(kw => lowerInput.includes(kw.toLowerCase()))
  
  // After: Cache in WeakMap, compute once
  let lowerKeywords = keywordCache.get(trigger);
  if (!lowerKeywords) {
    lowerKeywords = trigger.keywords.map(kw => kw.toLowerCase());
    keywordCache.set(trigger, lowerKeywords);
  }

• RequestQueue.js: Replace O(n) linear search with O(log n) binary search for priority queue insertion.

• MemoryManager.js: Single-pass sentence processing with Set-based keyword lookup. Remove unused emotion pattern.

• database.js: Wrap deleteStory in transaction, parallelize independent deletions with Promise.all.

Frontend

• StoryPanel.jsx:
  • Memoize displayMessages filtering with useMemo
  • Memoize formatTimestamp with useCallback
  • Hoist static NPC regex outside render path, reset lastIndex before iteration

Original prompt

Identify and suggest improvements to slow or inefficient code

---

💬 We'd love your input! Share your thoughts on Copilot coding agent in our 2 minute survey.