/**
 * ─────────────────────────────────────────────
 *  SPB Git — Personal Git Platform
 * ─────────────────────────────────────────────
 *  Author  : Simon-Pierre Boucher
 *  Contact : contact@spboucher.ai
 *  File    : src/lib/search.mjs
 *  Purpose : Tiny inverted index over names, topics, descriptions, READMEs
 *  License : MIT © Simon-Pierre Boucher
 * ─────────────────────────────────────────────
 */

import { allOverviews } from './overview.mjs';

/** Strip markdown/code noise from a README for indexing + snippets. */
export function stripMarkdown(source) {
  return String(source)
    .replace(/```[\s\S]*?```/g, ' ')
    .replace(/`[^`]*`/g, ' ')
    .replace(/!\[[^\]]*\]\([^)]*\)/g, ' ')
    .replace(/\[([^\]]*)\]\([^)]*\)/g, '$1')
    .replace(/<[^>]+>/g, ' ')
    .replace(/[#>*_~|-]{1,}/g, ' ')
    .replace(/\s+/g, ' ')
    .trim();
}

/**
 * Build the search index (cached globally, refreshed on push).
 * @param {{config, repos, meta, cache}} ctx
 * @returns {Promise<{entries: object[]}>}
 */
export async function buildSearchIndex(ctx) {
  const cachePath = ctx.cache.path('global', 'search-index.json');
  const cached = ctx.cache.getJSON(cachePath);
  if (cached) return cached;
  const overviews = await allOverviews(ctx);
  const entries = [];
  for (const o of overviews) {
    let readmeText = '';
    if (o.head) {
      const readme = await ctx.repos.readme(o.name, o.head);
      if (readme) readmeText = stripMarkdown(readme.content.toString('utf8')).slice(0, 20000);
    }
    entries.push({
      name: o.name,
      description: o.description,
      topics: o.topics,
      topLanguage: o.topLanguage,
      lastPush: o.lastPush,
      readmeText,
    });
  }
  const index = { entries, builtAt: new Date().toISOString() };
  ctx.cache.setJSON(cachePath, index);
  return index;
}

/**
 * Query the index. Returns grouped, scored results.
 * @param {{config, repos, meta, cache}} ctx
 * @param {string} query
 * @returns {Promise<{repos: object[], readmes: object[]}>}
 */
export async function search(ctx, query) {
  const terms = String(query ?? '')
    .toLowerCase()
    .split(/\s+/)
    .filter((t) => t.length >= 2)
    .slice(0, 8);
  if (terms.length === 0) return { repos: [], readmes: [] };
  const { entries } = await buildSearchIndex(ctx);
  const repoHits = [];
  const readmeHits = [];
  for (const entry of entries) {
    const name = entry.name.toLowerCase();
    const description = (entry.description ?? '').toLowerCase();
    const topics = (entry.topics ?? []).map((t) => t.toLowerCase());
    const readme = (entry.readmeText ?? '').toLowerCase();
    let score = 0;
    let readmeMatch = false;
    for (const term of terms) {
      if (name === term) score += 100;
      else if (name.includes(term)) score += 40;
      if (description.includes(term)) score += 15;
      if (topics.some((t) => t.includes(term))) score += 25;
      if (readme.includes(term)) {
        score += 5;
        readmeMatch = true;
      }
    }
    if (score === 0) continue;
    const hit = { ...entry, score };
    if (name.includes(terms[0]) || description.includes(terms[0]) || topics.some((t) => t.includes(terms[0]))) {
      repoHits.push(hit);
    }
    if (readmeMatch) {
      const idx = readme.indexOf(terms[0]);
      const start = Math.max(0, idx - 80);
      const snippet = entry.readmeText.slice(start, start + 220).trim();
      readmeHits.push({ ...hit, snippet: (start > 0 ? '…' : '') + snippet + '…' });
    }
  }
  repoHits.sort((a, b) => b.score - a.score);
  readmeHits.sort((a, b) => b.score - a.score);
  return { repos: repoHits.slice(0, 20), readmes: readmeHits.slice(0, 20) };
}