// Links in README.md, docs/*.md and the site resolve: relative links and images point at files that exist, #anchors
// at headings (GitHub / Forgejo slugs) or element ids, and links into this repo on git.crt.fyi
// (/raw/branch/main/<path>, /src/branch/main/<path>) at files in the checkout. Other external URLs are not fetched.
import { describe, expect, test } from 'bun:test';
import { existsSync, readdirSync, readFileSync, statSync } from 'node:fs';
import { dirname, join, relative } from 'node:path';

const ROOT = join(import.meta.dir, '..');
const REPO = /^https:\/\/git\.crt\.fyi\/CRT\/storebound\/(?:raw|src)\/branch\/main\/([^?#]+)(#.*)?$/;

/** GitHub-style heading slug: lower case, punctuation dropped, spaces to hyphens. */
const slug = (h: string): string => h.trim().toLowerCase().replace(/<[^>]+>/g, '').replace(/[`*_]/g, '')
  .replace(/[^\p{L}\p{N}\s-]/gu, '').replace(/\s/g, '-');

function anchorsOf(file: string): Set<string> {
  const src = readFileSync(file, 'utf8');
  if (file.endsWith('.html')) return new Set([...src.matchAll(/\sid="([^"]+)"/g)].map(m => m[1]!));
  const out = new Set<string>();
  const seen = new Map<string, number>();
  let fence = false;
  for (const line of src.split('\n')) {
    if (/^```/.test(line)) fence = !fence;
    const m = !fence && /^#{1,6} (.+)$/.exec(line);
    if (!m) continue;
    const s = slug(m[1]!);
    const n = seen.get(s) ?? 0;
    seen.set(s, n + 1);
    out.add(n ? `${s}-${n}` : s);
  }
  return out;
}

function linksOf(file: string): string[] {
  let src = readFileSync(file, 'utf8');
  if (file.endsWith('.md')) src = src.replace(/^```[\s\S]*?^```/gm, '').replace(/`[^`\n]*`/g, '');
  const md = [...src.matchAll(/\]\(([^)\s]+)(?:\s+"[^"]*")?\)/g)].map(m => m[1]!);
  const html = [...src.matchAll(/\s(?:href|src|srcset|poster)="([^"]+)"/g)].flatMap(m => m[1]!.split(',').map(s => s.trim().split(/\s+/)[0]!));
  return [...md, ...html].filter(Boolean);
}

function problems(file: string): string[] {
  const out: string[] = [];
  const base = dirname(file);
  for (const link of linksOf(file)) {
    const repo = REPO.exec(link);
    let target: string, anchor: string | undefined;
    if (repo) { target = join(ROOT, decodeURIComponent(repo[1]!)); anchor = repo[2]?.slice(1); }
    else if (/^[a-z]+:/i.test(link) || link.startsWith('//')) continue;
    // served by the desktop release server, not the site (tools/vds/apache-storebound-desktop.conf)
    else if (/(^|\/)latest\.json$/.test(link)) continue;
    else {
      const [p, a] = link.split('#') as [string, string | undefined];
      target = p ? join(base, decodeURIComponent(p.split('?')[0]!)) : file;
      anchor = a;
    }
    const rel = relative(ROOT, file);
    if (!existsSync(target)) { out.push(`${rel}: ${link} (no such file)`); continue; }
    if (anchor && statSync(target).isFile() && /\.(md|html)$/.test(target) && !anchorsOf(target).has(anchor))
      out.push(`${rel}: ${link} (no #${anchor})`);
  }
  return out;
}

const docs = readdirSync(join(ROOT, 'docs')).filter(n => n.endsWith('.md')).map(n => join(ROOT, 'docs', n));

describe('links resolve', () => {
  test('README.md and docs/*.md', () => {
    expect([join(ROOT, 'README.md'), ...docs].flatMap(problems)).toEqual([]);
  });
  test('site pages', () => {
    const pages = ['site/index.html', 'site/desktop/index.html'].map(p => join(ROOT, p)).filter(existsSync);
    expect(pages.flatMap(problems)).toEqual([]);
  });
  test('the checker catches a missing file, a missing anchor and a stale repo link', () => {
    expect(slug('Install for your AI tool')).toBe('install-for-your-ai-tool');
    expect(slug('Credentials (never in the repo)')).toBe('credentials-never-in-the-repo');
    expect(anchorsOf(join(ROOT, 'docs/getting-started.md')).has('install-for-your-ai-tool')).toBe(true);
    expect(REPO.exec('https://git.crt.fyi/CRT/storebound/raw/branch/main/site/img/showcase.webp')?.[1]).toBe('site/img/showcase.webp');
  });
});