memind/mindspace-seo-discovery-service.mjs

import { isPublicationIndexable, normalizePublicationSnapshot } from './mindspace-index-policy.mjs';
import { PLATFORM_STATIC_DISCOVERY_ENTRIES } from './platform-seo-html.mjs';

function escapeXml(value) {
  return String(value ?? '')
    .replaceAll('&', '&')
    .replaceAll('<', '&lt;')
    .replaceAll('>', '&gt;')
    .replaceAll('"', '&quot;')
    .replaceAll("'", '&apos;');
}

function toAbsoluteUrl(origin, value) {
  const raw = String(value ?? '').trim();
  if (!raw) return '';
  if (/^https?:\/\//i.test(raw)) return raw.split('#')[0];
  if (raw.startsWith('/') && origin) return `${origin}${raw}`.split('#')[0];
  return raw.split('#')[0];
}

export async function listIndexablePublications(
  pool,
  { limit = 5000, offset = 0 } = {},
) {
  if (!pool) return [];
  const safeLimit = Math.min(Math.max(Number(limit) || 5000, 1), 10000);
  const safeOffset = Math.max(Number(offset) || 0, 0);
  const [rows] = await pool.query(
    `SELECT pr.id, pr.page_id, pr.public_url, pr.access_mode, pr.status,
            pr.user_confirmed_at, pr.expires_at, pr.updated_at, pr.published_at,
            p.title, p.summary
     FROM h5_publish_records pr
     LEFT JOIN h5_page_records p ON p.id = pr.page_id
     WHERE pr.status = 'online'
       AND pr.access_mode = 'public'
       AND (pr.expires_at IS NULL OR pr.expires_at > ?)
     ORDER BY pr.published_at DESC
     LIMIT ? OFFSET ?`,
    [Date.now(), safeLimit, safeOffset],
  );
  return rows
    .map((row) =>
      normalizePublicationSnapshot({
        id: row.id,
        pageId: row.page_id,
        publicUrl: row.public_url,
        accessMode: row.access_mode,
        status: row.status,
        userConfirmedAt: row.user_confirmed_at,
        expiresAt: row.expires_at,
        updatedAt: row.updated_at ?? row.published_at,
        title: row.title,
        summary: row.summary,
      }),
    )
    .filter((entry) => isPublicationIndexable(entry));
}

export async function resolvePublicationIndexSnapshot(
  pool,
  {
    publicationId = null,
    pageId = null,
    userId = null,
  } = {},
) {
  if (!pool) return null;
  const id = String(publicationId ?? '').trim();
  const page = String(pageId ?? '').trim();
  const owner = String(userId ?? '').trim();
  if (!id && !page) return null;

  const params = [];
  let sql = `SELECT pr.id, pr.page_id, pr.user_id, pr.public_url, pr.access_mode, pr.status,
                    pr.user_confirmed_at, pr.expires_at, pr.updated_at, pr.published_at,
                    p.title, p.summary
             FROM h5_publish_records pr
             LEFT JOIN h5_page_records p ON p.id = pr.page_id
             WHERE pr.status = 'online'`;
  if (id) {
    sql += ' AND pr.id = ?';
    params.push(id);
  } else {
    sql += ' AND pr.page_id = ?';
    params.push(page);
    if (owner) {
      sql += ' AND pr.user_id = ?';
      params.push(owner);
    }
  }
  sql += ' ORDER BY pr.published_at DESC LIMIT 1';
  const [rows] = await pool.query(sql, params);
  const row = rows[0];
  if (!row) return null;
  return normalizePublicationSnapshot({
    id: row.id,
    pageId: row.page_id,
    publicUrl: row.public_url,
    accessMode: row.access_mode,
    status: row.status,
    userConfirmedAt: row.user_confirmed_at,
    expiresAt: row.expires_at,
    updatedAt: row.updated_at ?? row.published_at,
    title: row.title,
    summary: row.summary,
  });
}

function scoreDiscoveryEntry(entry) {
  const url = String(entry?.publicUrl ?? entry?.public_url ?? '').trim();
  if (!url) return -100;
  if (/\/_archived/i.test(url)) return -50;
  if (url.includes('/u/')) return 100;
  if (url.includes('/MindSpace/')) return 10;
  return 20;
}

export function dedupeDiscoveryEntriesByPage(publications = []) {
  const byPage = new Map();
  for (const entry of publications) {
    const snapshot = normalizePublicationSnapshot(entry);
    if (!snapshot) continue;
    const pageKey =
      String(snapshot.pageId ?? '').trim() ||
      String(snapshot.publicUrl ?? '').trim();
    if (!pageKey) continue;
    const existing = byPage.get(pageKey);
    if (
      !existing ||
      scoreDiscoveryEntry(snapshot) > scoreDiscoveryEntry(existing)
    ) {
      byPage.set(pageKey, snapshot);
    }
  }
  return [...byPage.values()];
}

export function mergeDiscoveryEntries(publications = [], staticEntries = PLATFORM_STATIC_DISCOVERY_ENTRIES) {
  const merged = [];
  const seen = new Set();
  const dedupedPublications = dedupeDiscoveryEntriesByPage(publications);
  for (const entry of [...dedupedPublications, ...staticEntries]) {
    const key = String(entry.publicUrl ?? entry.public_url ?? '').trim();
    if (!key || seen.has(key)) continue;
    seen.add(key);
    merged.push(normalizePublicationSnapshot(entry) ?? entry);
  }
  return merged;
}

export function renderSitemapXml(entries, { origin = '' } = {}) {
  const urls = entries
    .map((entry) => {
      const loc = toAbsoluteUrl(origin, entry.publicUrl);
      if (!loc) return '';
      const lastmod = entry.updatedAt
        ? new Date(Number(entry.updatedAt)).toISOString()
        : new Date().toISOString();
      return `  <url>
    <loc>${escapeXml(loc)}</loc>
    <lastmod>${escapeXml(lastmod)}</lastmod>
  </url>`;
    })
    .filter(Boolean)
    .join('\n');
  return `<?xml version="1.0" encoding="UTF-8"?>
<urlset xmlns="http://www.sitemaps.org/schemas/sitemap/0.9">
${urls}
</urlset>
`;
}

export function renderRobotsTxt({
  origin = '',
  sitemapEnabled = false,
  llmsTxtEnabled = false,
} = {}) {
  const lines = [
    'User-agent: *',
    'Allow: /u/',
    'Allow: /learn/',
    'Disallow: /api/',
    'Disallow: /admin-api/',
    'Disallow: /mindspace/',
  ];
  if (sitemapEnabled && origin) {
    lines.push(`Sitemap: ${origin}/sitemap.xml`);
  }
  if (llmsTxtEnabled && origin) {
    lines.push(`# AI discovery: ${origin}/llms.txt`);
  }
  lines.push('');
  return `${lines.join('\n')}`;
}

export function renderLlmsTxt(entries, { origin = '' } = {}) {
  const header = [
    '# TKMind MindSpace public pages',
    '# Public online publications are listed here.',
    '',
  ];
  const body = entries
    .map((entry) => {
      const url = toAbsoluteUrl(origin, entry.publicUrl);
      if (!url) return '';
      const summary = String(entry.summary ?? entry.title ?? '').trim();
      return summary ? `- [${summary}](${url})` : `- ${url}`;
    })
    .filter(Boolean);
  return `${header.concat(body).join('\n')}\n`;
}

export function createMindspaceSeoDiscoveryService(pool) {
  return {
    listIndexablePublications: (options) => listIndexablePublications(pool, options),
    resolvePublicationIndexSnapshot: (options) =>
      resolvePublicationIndexSnapshot(pool, options),
    dedupeDiscoveryEntriesByPage,
    mergeDiscoveryEntries,
    renderSitemapXml,
    renderRobotsTxt,
    renderLlmsTxt,
  };
}