memind/mindspace-seo-tags.mjs

import { resolveGeoKeywords } from './mindspace-geo-meta.mjs';
import { extractCoverSignals } from './mindspace-thumbnails.mjs';
import { extractSharePreviewMeta } from './mindspace-og-tags.mjs';

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

function hasMeta(html, attr, name) {
  return new RegExp(`<meta[^>]+${attr}=["']${name}["']`, 'i').test(String(html));
}

function hasLinkRel(html, rel) {
  return new RegExp(`<link[^>]+rel=["'][^"']*\\b${rel}\\b[^"']*["']`, 'i').test(String(html));
}

function appendBeforeHeadClose(html, block) {
  const source = String(html ?? '');
  if (!block) return source;
  if (/<\/head>/i.test(source)) {
    return source.replace(/<\/head>/i, `${block}\n</head>`);
  }
  if (/<head[^>]*>/i.test(source)) {
    return source.replace(/(<head[^>]*>)/i, `$1${block}`);
  }
  return `${block}\n${source}`;
}

function rawMetaContent(html, attr, name) {
  const pattern = new RegExp(
    `<meta[^>]+${attr}=["']${name}["'][^>]+content=["']([^"']*)["']|<meta[^>]+content=["']([^"']*)["'][^>]+${attr}=["']${name}["']`,
    'i',
  );
  const match = String(html ?? '').match(pattern);
  return match?.[1] || match?.[2] || '';
}

export function injectRobotsNoindex(html) {
  const source = String(html ?? '');
  if (hasMeta(source, 'name', 'robots')) return source;
  const tags = [
    '<meta name="robots" content="noindex, nofollow">',
    '<meta name="googlebot" content="noindex, nofollow">',
  ]
    .map((tag) => `  ${tag}`)
    .join('\n');
  return appendBeforeHeadClose(source, `\n${tags}\n`);
}

export function resolveCanonicalUrl({
  publication = null,
  pageUrl = '',
  origin = '',
} = {}) {
  const fromPublication = String(publication?.publicUrl ?? '').trim();
  if (fromPublication) {
    if (/^https?:\/\//i.test(fromPublication)) return fromPublication.split('#')[0];
    if (fromPublication.startsWith('/') && origin) {
      return `${origin}${fromPublication}`.split('#')[0];
    }
    return fromPublication.split('#')[0];
  }
  return String(pageUrl ?? '').split('#')[0];
}

export function injectSeoTags(
  html,
  {
    origin = '',
    pageUrl = '',
    pageDirUrl = '',
    publication = null,
    canonicalEnabled = true,
    htmlFilePath = '',
    fileExists = null,
    meta = {},
  } = {},
) {
  const source = String(html ?? '');
  const preview = extractSharePreviewMeta(source, {
    origin,
    pageUrl,
    pageDirUrl,
    meta,
    htmlFilePath,
    fileExists,
  });
  const signals = extractCoverSignals(source, meta);
  const tags = [];
  const links = [];

  if (!hasMeta(source, 'name', 'description') && preview.description) {
    tags.push(
      `<meta name="description" content="${escapeAttr(preview.description)}">`,
    );
  }
  const keywords = resolveGeoKeywords(source);
  for (const item of Array.isArray(meta.keywords) ? meta.keywords : []) {
    const token = String(item ?? '').trim();
    if (token && !keywords.includes(token)) keywords.push(token);
  }
  if (signals.tag && !keywords.includes(signals.tag)) keywords.unshift(signals.tag);
  const keywordLine = keywords.slice(0, 12);
  if (keywordLine.length > 0 && !hasMeta(source, 'name', 'keywords')) {
    tags.push(`<meta name="keywords" content="${escapeAttr(keywordLine.join(', '))}">`);
  }
  if (canonicalEnabled) {
    const canonicalUrl = resolveCanonicalUrl({ publication, pageUrl, origin });
    if (canonicalUrl && !hasLinkRel(source, 'canonical')) {
      links.push(`<link rel="canonical" href="${escapeAttr(canonicalUrl)}">`);
    }
  }
  const existingLang = source.match(/<html[^>]*\blang=["']([^"']+)["']/i)?.[1];
  if (!existingLang) {
    // Leave html lang to author; only add content-language when absent.
    if (!hasMeta(source, 'http-equiv', 'content-language')) {
      tags.push('<meta http-equiv="content-language" content="zh-CN">');
    }
  }

  const block = [...tags, ...links].map((item) => `  ${item}`).join('\n');
  if (!block) return source;
  return appendBeforeHeadClose(source, `\n${block}\n`);
}

export function extractSeoDescription(html) {
  return (
    rawMetaContent(html, 'property', 'og:description') ||
    rawMetaContent(html, 'name', 'description') ||
    extractCoverSignals(String(html ?? '')).subtitle ||
    ''
  );
}