const MAX_EXPANSION_CACHE = 512;
const expansionCache = new Map<string, string[]>();

const CONJUNCT_RE = /\b(?:and|or|vs|versus)\b/i;

function normalizeQuery(query: string): string {
  return query.trim().replace(/\s+/g, " ");
}

function splitClauses(query: string): string[] {
  // Split on commas, semicolons, and conjunctions to get independent clauses.
  const parts = query.split(/[,;]|\b(?:and|or|vs|versus)\b/i);
  return parts.map((p) => p.trim()).filter(Boolean);
}

function unique(values: string[]): string[] {
  const seen = new Set<string>();
  const result: string[] = [];
  for (const v of values) {
    const lower = v.toLowerCase();
    if (seen.has(lower)) continue;
    seen.add(lower);
    result.push(v);
  }
  return result;
}

/**
 * Expand a query into multiple variants for broader retrieval coverage.
 *
 * OPTIMIZATIONS over the original:
 *  1. Simplified clause splitting — removed the duplicate second-half push
 *     that produced near-duplicate variants wasting retrieval calls.
 *  2. Added length-based pruning — variants shorter than 3 chars are dropped.
 *  3. Cache key uses a stronger separator (U+0001 instead of U+0000) to
 *     prevent edge-case collisions with queries containing null bytes.
 */
export function expandQueries(query: string, maxCount: number): string[] {
  const base = normalizeQuery(query);

  if (base.length < 15) {
    return [base];
  }

  const cacheKey = base + "\u0001" + maxCount;

  const cached = expansionCache.get(cacheKey);
  if (cached !== undefined) return cached;

  const words = base.split(/\W+/).filter(Boolean);
  const focus = words.slice(0, 6).join(" ");
  const clauseVariants = splitClauses(base).slice(0, 2);

  const v0 = words.length > 5 ? words.slice(0, 5).join(" ") : "";
  const v1 = focus && focus !== base ? focus : "";
  const v2 = clauseVariants[0] ? "tell me about " + clauseVariants[0] : "";
  const v3 = clauseVariants[1] ? "explain " + clauseVariants[1] : "";
  const v4 = focus && focus.length < 30 ? focus + " meaning" : "";

  const limited = unique([base, v0, v1, v2, v3, v4])
    .filter((v) => v.length >= 3)
    .slice(0, Math.max(1, maxCount));

  if (limited.length === 0) {
    expansionCache.set(cacheKey, [base]);
    return [base];
  }

  if (expansionCache.size >= MAX_EXPANSION_CACHE) {
    const oldest = expansionCache.keys().next().value;
    if (oldest !== undefined) expansionCache.delete(oldest);
  }
  expansionCache.set(cacheKey, limited);

  return limited;
}

const MAX_EXPANSION_CACHE = 512;
const expansionCache = new Map<string, string[]>();

const CONJUNCT_RE = /\b(?:and|or|vs|versus)\b/i;

function normalizeQuery(query: string): string {
  return query.trim().replace(/\s+/g, " ");
}

function splitClauses(query: string): string[] {
  // Split on commas, semicolons, and conjunctions to get independent clauses.
  const parts = query.split(/[,;]|\b(?:and|or|vs|versus)\b/i);
  return parts.map((p) => p.trim()).filter(Boolean);
}

function unique(values: string[]): string[] {
  const seen = new Set<string>();
  const result: string[] = [];
  for (const v of values) {
    const lower = v.toLowerCase();
    if (seen.has(lower)) continue;
    seen.add(lower);
    result.push(v);
  }
  return result;
}

/**
 * Expand a query into multiple variants for broader retrieval coverage.
 *
 * OPTIMIZATIONS over the original:
 *  1. Simplified clause splitting — removed the duplicate second-half push
 *     that produced near-duplicate variants wasting retrieval calls.
 *  2. Added length-based pruning — variants shorter than 3 chars are dropped.
 *  3. Cache key uses a stronger separator (U+0001 instead of U+0000) to
 *     prevent edge-case collisions with queries containing null bytes.
 */
export function expandQueries(query: string, maxCount: number): string[] {
  const base = normalizeQuery(query);

  if (base.length < 15) {
    return [base];
  }

  const cacheKey = base + "\u0001" + maxCount;

  const cached = expansionCache.get(cacheKey);
  if (cached !== undefined) return cached;

  const words = base.split(/\W+/).filter(Boolean);
  const focus = words.slice(0, 6).join(" ");
  const clauseVariants = splitClauses(base).slice(0, 2);

  const v0 = words.length > 5 ? words.slice(0, 5).join(" ") : "";
  const v1 = focus && focus !== base ? focus : "";
  const v2 = clauseVariants[0] ? "tell me about " + clauseVariants[0] : "";
  const v3 = clauseVariants[1] ? "explain " + clauseVariants[1] : "";
  const v4 = focus && focus.length < 30 ? focus + " meaning" : "";

  const limited = unique([base, v0, v1, v2, v3, v4])
    .filter((v) => v.length >= 3)
    .slice(0, Math.max(1, maxCount));

  if (limited.length === 0) {
    expansionCache.set(cacheKey, [base]);
    return [base];
  }

  if (expansionCache.size >= MAX_EXPANSION_CACHE) {
    const oldest = expansionCache.keys().next().value;
    if (oldest !== undefined) expansionCache.delete(oldest);
  }
  expansionCache.set(cacheKey, limited);

  return limited;
}