memind/mindspace-geo-tags.mjs

import { extractCoverSignals } from './mindspace-thumbnails.mjs';
import { extractSharePreviewMeta, PLATFORM_SITE_NAME } from './mindspace-og-tags.mjs';
import { extractSeoDescription } from './mindspace-seo-tags.mjs';

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

function rawTitleFromHtml(html) {
  return String(html ?? '').match(/<title[^>]*>([^<]*)<\/title>/i)?.[1]?.trim() ?? '';
}

function rawH1FromHtml(html) {
  const match = String(html ?? '').match(/<h1[^>]*>([\s\S]*?)<\/h1>/i);
  if (!match) return '';
  return match[1].replace(/<[^>]+>/g, '').replace(/\s+/g, ' ').trim();
}

function hasJsonLd(html) {
  return /<script[^>]+type=["']application\/ld\+json["']/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 parseMindspaceGeoMeta(html) {
  const tag = String(html ?? '').match(/<meta[^>]*name=["']mindspace-geo["'][^>]*>/i)?.[0];
  if (!tag) return {};
  const contentMatch =
    tag.match(/content=(["'])([\s\S]*?)\1/i) ?? tag.match(/content=["']([^"']+)["']/i);
  const raw = contentMatch?.[2] ?? contentMatch?.[1];
  if (!raw) return {};
  try {
    return JSON.parse(raw.replaceAll('&quot;', '"'));
  } catch {
    return {};
  }
}

function buildArticleJsonLd({
  title,
  description,
  imageUrl,
  pageUrl,
  siteName = PLATFORM_SITE_NAME,
  faq = [],
}) {
  const graph = {
    '@context': 'https://schema.org',
    '@type': 'Article',
    headline: title,
    description,
    inLanguage: 'zh-CN',
    isAccessibleForFree: true,
    publisher: {
      '@type': 'Organization',
      name: siteName,
    },
  };
  if (pageUrl) graph.mainEntityOfPage = pageUrl;
  if (imageUrl) graph.image = [imageUrl];
  if (Array.isArray(faq) && faq.length > 0) {
    graph.hasPart = {
      '@context': 'https://schema.org',
      '@type': 'FAQPage',
      mainEntity: faq
        .map((item) => ({
          '@type': 'Question',
          name: String(item?.q ?? item?.question ?? '').trim(),
          acceptedAnswer: {
            '@type': 'Answer',
            text: String(item?.a ?? item?.answer ?? '').trim(),
          },
        }))
        .filter((item) => item.name && item.acceptedAnswer.text),
    };
  }
  return graph;
}

export function buildGeoStructuredData(
  html,
  {
    origin = '',
    pageUrl = '',
    pageDirUrl = '',
    publication = null,
    siteName = PLATFORM_SITE_NAME,
    htmlFilePath = '',
    fileExists = null,
    meta = {},
  } = {},
) {
  const source = String(html ?? '');
  const authorMeta = parseMindspaceGeoMeta(source);
  const preview = extractSharePreviewMeta(source, {
    origin,
    pageUrl,
    pageDirUrl,
    siteName,
    meta,
    htmlFilePath,
    fileExists,
  });
  const title =
    rawMetaOgTitle(source) ||
    rawTitleFromHtml(source) ||
    rawH1FromHtml(source) ||
    'MindSpace 页面';
  const description =
    String(authorMeta.summary ?? '').trim() ||
    extractSeoDescription(source) ||
    preview.description ||
    '';
  const canonical =
    String(publication?.publicUrl ?? pageUrl ?? '').trim() || pageUrl;
  const absoluteUrl = canonical.startsWith('http')
    ? canonical
    : origin && canonical.startsWith('/')
      ? `${origin}${canonical}`
      : pageUrl;
  return buildArticleJsonLd({
    title,
    description,
    imageUrl: preview.imageUrl,
    pageUrl: absoluteUrl,
    siteName: preview.siteName || siteName,
    faq: authorMeta.faq,
  });
}

function rawMetaOgTitle(html) {
  return (
    String(html ?? '').match(
      /<meta[^>]+property=["']og:title["'][^>]+content=["']([^"']+)["']/i,
    )?.[1] ||
    String(html ?? '').match(
      /<meta[^>]+content=["']([^"']+)["'][^>]+property=["']og:title["']/i,
    )?.[1] ||
    ''
  );
}

export function injectGeoTags(
  html,
  ctx = {},
) {
  const sourceHtml = String(ctx.sourceHtml ?? html ?? '');
  const targetHtml = String(html ?? '');
  if (hasJsonLd(targetHtml)) return targetHtml;
  const payload = buildGeoStructuredData(sourceHtml, ctx);
  const serialized = JSON.stringify(payload).replace(/</g, '\\u003c');
  const block = `\n  <script type="application/ld+json">${serialized}</script>\n`;
  return appendBeforeHeadClose(targetHtml, block);
}