memind/mindspace-geo-backfill.mjs

import fs from 'node:fs';
import path from 'node:path';
import { injectBeforeDocumentClosingBody } from './html-document-injection.mjs';
import { parseMindspaceCoverMeta, upsertMindspaceCoverMeta } from './mindspace-cover-meta.mjs';
import {
  extractFaqFromHeadings,
  parseMindspaceGeoMeta,
  resolveGeoFaq,
  resolveGeoKeywords,
} from './mindspace-geo-meta.mjs';

function htmlWithoutEmbeddedCode(html) {
  return String(html ?? '')
    .replace(/<script\b[\s\S]*?<\/script>/gi, ' ')
    .replace(/<style\b[\s\S]*?<\/style>/gi, ' ');
}
import { ensurePlatformBrandFooter, injectPlatformBrandVisibilityStyle } from './mindspace-page-tag.mjs';
import { PLATFORM_BRAND_ICON_PATH } from './mindspace-og-tags.mjs';
import { extractCoverSignals } from './mindspace-thumbnails.mjs';
import { extractSeoDescription } from './mindspace-seo-tags.mjs';
import { hasRasterShareImageHint } from './wechat/verify/share-preview.mjs';

export const DEFAULT_RASTER_COVER_URL = 'https://m.tkmind.cn/brand/tkmind-icon.png';

function escapeMetaAttribute(value) {
  return String(value ?? '')
    .replaceAll('&', '&amp;')
    .replaceAll('"', '&quot;')
    .replaceAll('<', '&lt;');
}

function stripHtml(value) {
  return String(value ?? '')
    .replace(/<[^>]+>/g, ' ')
    .replace(/\s+/g, ' ')
    .trim();
}

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);
  return match ? stripHtml(match[1]) : '';
}

function rawFirstParagraph(html) {
  const match = String(html ?? '').match(/<p[^>]*>([\s\S]*?)<\/p>/i);
  return match ? stripHtml(match[1]) : '';
}

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

function escapeHtml(value) {
  return String(value ?? '')
    .replace(/&/g, '&amp;')
    .replace(/</g, '&lt;')
    .replace(/>/g, '&gt;')
    .replace(/"/g, '&quot;');
}

function countH2(html) {
  return (String(html ?? '').match(/<h2\b/gi) ?? []).length;
}

function hasShareDescription(html) {
  return Boolean(extractSeoDescription(html));
}

export function repairMindspaceGeoMeta(html) {
  const source = String(html ?? '');
  const parsed = parseMindspaceGeoMeta(source);
  const hasValidSummary = String(parsed.summary ?? '').trim().length > 0;
  const hasTag = /<meta[^>]*name=["']mindspace-geo["']/i.test(source);
  if (hasTag && hasValidSummary) return source;
  return upsertMindspaceGeoMeta(source, buildMindspaceGeoPayload(source), { forceReplace: true });
}

export function buildMindspaceGeoPayload(html) {
  const source = htmlWithoutEmbeddedCode(html);
  const title = rawH1FromHtml(source) || rawTitleFromHtml(source) || 'TKMind 页面';
  const cover = extractCoverSignals(source);
  const summary =
    String(parseMindspaceGeoMeta(source).summary ?? '').trim() ||
    extractSeoDescription(source) ||
    String(cover.subtitle ?? '').trim() ||
    rawFirstParagraph(source) ||
    title;

  let keywords = resolveGeoKeywords(source);
  if (keywords.length < 2) {
    keywords = [title, cover.tag, 'TKMind'].map((item) => String(item ?? '').trim()).filter(Boolean);
  }
  keywords = [...new Set(keywords)].slice(0, 8);

  let faq = resolveGeoFaq(source);
  if (faq.length < 1) {
    faq = extractFaqFromHeadings(source);
  }
  if (faq.length < 1 && summary) {
    faq = [{ q: `${title}讲什么？`, a: summary.slice(0, 220) }];
  }

  return {
    summary: summary.slice(0, 320),
    keywords,
    faq: faq.slice(0, 6),
  };
}

export function upsertMindspaceGeoMeta(html, geoPayload, { forceReplace = false } = {}) {
  const payload = forceReplace
    ? { ...(geoPayload ?? {}) }
    : {
        ...buildMindspaceGeoPayload(html),
        ...(geoPayload && typeof geoPayload === 'object' ? geoPayload : {}),
      };
  payload.summary = String(payload.summary ?? '').trim().slice(0, 320);
  payload.keywords = Array.isArray(payload.keywords)
    ? payload.keywords.map((item) => String(item ?? '').trim()).filter(Boolean).slice(0, 12)
    : [];
  payload.faq = Array.isArray(payload.faq)
    ? payload.faq
        .map((item) => ({
          q: String(item?.q ?? '').trim().slice(0, 120),
          a: String(item?.a ?? '').trim().slice(0, 500),
        }))
        .filter((item) => item.q && item.a && item.a.length >= 4)
        .slice(0, 6)
    : [];
  if (payload.faq.length < 1 && payload.summary) {
    const topic = String(rawH1FromHtml(html) || rawTitleFromHtml(html) || payload.summary).trim();
    payload.faq = [{ q: `${topic}讲什么？`, a: payload.summary.slice(0, 220) }];
  }
  const metaTag = `<meta name="mindspace-geo" content="${escapeMetaAttribute(JSON.stringify(payload))}">`;
  const source = String(html ?? '');

  if (hasMeta(source, 'mindspace-geo')) {
    return source.replace(/<meta[^>]*name=["']mindspace-geo["'][^>]*>\s*/i, `${metaTag.trim()}\n  `);
  }
  if (/<meta[^>]*name=["']mindspace-cover["'][^>]*>/i.test(source)) {
    return source.replace(
      /(<meta[^>]*name=["']mindspace-cover["'][^>]*>)/i,
      `$1\n  ${metaTag.trim()}`,
    );
  }
  if (/<meta[^>]*name=["']description["'][^>]*>/i.test(source)) {
    return source.replace(
      /(<meta[^>]*name=["']description["'][^>]*>)/i,
      `$1\n  ${metaTag.trim()}`,
    );
  }
  if (/<head[^>]*>/i.test(source)) {
    return source.replace(/<head([^>]*)>/i, `<head$1>\n  ${metaTag.trim()}`);
  }
  return `${metaTag.trim()}\n${source}`;
}

export function upsertDescriptionMeta(html, description) {
  const text = String(description ?? '').trim();
  if (!text || hasMeta(html, 'description')) return String(html ?? '');
  const metaTag = `<meta name="description" content="${escapeMetaAttribute(text)}">`;
  const source = String(html ?? '');
  if (/<title[^>]*>/i.test(source)) {
    return source.replace(/(<title[^>]*>[^<]*<\/title>)/i, `$1\n  ${metaTag}`);
  }
  if (/<head[^>]*>/i.test(source)) {
    return source.replace(/<head([^>]*)>/i, `<head$1>\n  ${metaTag}`);
  }
  return `${metaTag}\n${source}`;
}

export function buildMinimalCoverMeta(html, { title = '' } = {}) {
  const cover = parseMindspaceCoverMeta(html);
  if (cover) return cover;
  const signals = extractCoverSignals(String(html ?? ''));
  const resolvedTitle = title || rawH1FromHtml(html) || rawTitleFromHtml(html) || 'TKMind 页面';
  const description = extractSeoDescription(html) || rawFirstParagraph(html) || resolvedTitle;
  return {
    tag: signals.tag || '报告',
    emoji: '📄',
    accent: signals.accent || '#3b82f6',
    accent2: signals.accent2 || '#1e293b',
    subtitle: signals.subtitle || description.slice(0, 80),
  };
}

function humanizeFilename(filePath) {
  const base = path.basename(String(filePath ?? ''), '.html');
  const decoded = decodeURIComponent(base);
  return decoded
    .replace(/[-_]+/g, ' ')
    .replace(/\s+/g, ' ')
    .trim();
}

function isWeakTitle(title) {
  const value = String(title ?? '').trim();
  if (!value || value.length < 2) return true;
  if (/^(页面|我的页面|未命名|demo|test|index|MindSpace 页面)$/i.test(value)) return true;
  if (/^[\x00-\x7F]+$/.test(value) && value.length < 4) return true;
  return false;
}

function buildFallbackTitle(html, htmlFilePath = '') {
  const cover = parseMindspaceCoverMeta(html);
  const fromCover = [cover?.tag, cover?.subtitle]
    .map((item) => String(item ?? '').trim())
    .filter(Boolean)
    .join(' · ');
  if (fromCover && !isWeakTitle(fromCover)) return fromCover;
  const fromH1 = rawH1FromHtml(html);
  if (fromH1 && !isWeakTitle(fromH1)) return fromH1;
  const fromFile = humanizeFilename(htmlFilePath);
  if (fromFile && !isWeakTitle(fromFile)) return fromFile;
  if (fromFile) return `TKMind · ${fromFile}`;
  return 'TKMind 精选页面';
}

export function improveTitleFromContext(html, htmlFilePath = '') {
  const source = String(html ?? '');
  const current = rawTitleFromHtml(source);
  if (!isWeakTitle(current)) return source;
  const candidate =
    [rawH1FromHtml(source), parseMindspaceCoverMeta(source)?.subtitle, buildFallbackTitle(source, htmlFilePath)]
      .map((item) => String(item ?? '').trim())
      .find((item) => item && !isWeakTitle(item)) ?? '';
  if (!candidate) return source;
  if (/<title[^>]*>[^<]*<\/title>/i.test(source)) {
    return source.replace(/<title[^>]*>[^<]*<\/title>/i, `<title>${candidate}</title>`);
  }
  if (/<head[^>]*>/i.test(source)) {
    return source.replace(/<head([^>]*)>/i, `<head$1>\n  <title>${candidate}</title>`);
  }
  return source;
}

export function ensureH1FromTitle(html) {
  const source = String(html ?? '');
  if (/<h1\b/i.test(source)) return source;
  const title = rawTitleFromHtml(source);
  if (!title) return source;
  if (/<body[^>]*>/i.test(source)) {
    return source.replace(/<body([^>]*)>/i, `<body$1>\n  <h1>${title}</h1>`);
  }
  return `${source}\n<h1>${title}</h1>`;
}

export function fixBrokenLocalCover(html, htmlFilePath = '') {
  const coverMeta = parseMindspaceCoverMeta(html);
  if (!coverMeta || !htmlFilePath) return String(html ?? '');
  const cover = String(coverMeta.cover ?? coverMeta.image ?? '').trim();
  if (!cover || /^https?:\/\//i.test(cover) || cover.startsWith('data:')) {
    return String(html ?? '');
  }
  const clean = cover.split('?')[0].split('#')[0].replace(/^\.\//, '');
  const base = path.dirname(htmlFilePath);
  const target = path.resolve(base, clean);
  if (target.startsWith(`${base}${path.sep}`) && fs.existsSync(target)) {
    return String(html ?? '');
  }
  const nextMeta = { ...coverMeta };
  delete nextMeta.cover;
  delete nextMeta.image;
  return upsertMindspaceCoverMeta(html, nextMeta, { replace: true });
}

export function ensureDefaultRasterCover(html, { coverUrl = DEFAULT_RASTER_COVER_URL } = {}) {
  const source = String(html ?? '');
  if (hasRasterShareImageHint(source)) return source;
  const coverMeta = parseMindspaceCoverMeta(source) ?? buildMinimalCoverMeta(source);
  return upsertMindspaceCoverMeta(source, {
    ...coverMeta,
    cover: coverUrl || DEFAULT_RASTER_COVER_URL,
  });
}

export function buildSeoOutlineSection(html) {
  const geo = parseMindspaceGeoMeta(html);
  const title = rawH1FromHtml(html) || rawTitleFromHtml(html) || '本页';
  const summary = String(geo.summary ?? extractSeoDescription(html) ?? title).trim();
  const faq = resolveGeoFaq(html, geo);
  const blocks = [];
  if (summary) {
    blocks.push(
      `<h2 style="font-size:17px;color:#334155;margin:0 0 8px">内容概要</h2>`,
      `<p style="margin:0 0 16px">${escapeHtml(summary)}</p>`,
    );
  }
  for (const item of faq.slice(0, 2)) {
    blocks.push(
      `<h2 style="font-size:17px;color:#334155;margin:16px 0 8px">${escapeHtml(item.q)}</h2>`,
      `<p style="margin:0 0 12px">${escapeHtml(item.a)}</p>`,
    );
  }
  if (countH2(blocks.join('\n')) < 2) {
    blocks.push(
      `<h2 style="font-size:17px;color:#334155;margin:16px 0 8px">关于${escapeHtml(title)}</h2>`,
      `<p style="margin:0 0 12px">${escapeHtml(summary || `${title} 由 TKMind 生成，可在浏览器中直接阅读与分享。`)}</p>`,
    );
  }
  if (blocks.length === 0) return '';
  return `<section data-mindspace-seo-outline="1" aria-label="页面提纲" style="margin:32px auto 20px;padding:20px 16px;max-width:920px;border-top:1px solid rgba(15,23,42,.08);color:#64748b;font-size:14px;line-height:1.75">${blocks.join('\n  ')}</section>`;
}

export function ensureSeoHeadingStructure(html) {
  const source = String(html ?? '');
  if (countH2(source) >= 2 || /data-mindspace-seo-outline=/i.test(source)) return source;
  const outline = buildSeoOutlineSection(source);
  if (!outline) return source;
  return injectBeforeDocumentClosingBody(source, `\n${outline}\n`);
}

export function ensureSeoGeoWarningsFixed(html, options = {}) {
  let next = String(html ?? '');
  if (options.fixPlatformBrand !== false) {
    next = injectPlatformBrandVisibilityStyle(ensurePlatformBrandFooter(next));
  }
  if (options.fixRasterCover !== false) {
    next = ensureDefaultRasterCover(next, { coverUrl: options.coverUrl });
  }
  if (options.fixHeadingStructure !== false) {
    next = ensureSeoHeadingStructure(next);
  }
  return next;
}

export function backfillMindspaceSeoGeoHtml(html, options = {}) {
  const htmlFilePath = String(options.htmlFilePath ?? '');
  let next = String(html ?? '');
  next = improveTitleFromContext(next, htmlFilePath);
  next = fixBrokenLocalCover(next, htmlFilePath);
  next = ensureH1FromTitle(next);

  const title = rawH1FromHtml(next) || rawTitleFromHtml(next);
  let geoPayload = buildMindspaceGeoPayload(next);

  if (!parseMindspaceCoverMeta(next) && options.ensureCover !== false) {
    next = upsertMindspaceCoverMeta(next, buildMinimalCoverMeta(next, { title }));
  }
  next = upsertDescriptionMeta(next, geoPayload.summary);
  const existingGeo = parseMindspaceGeoMeta(next);
  const needsGeoUpsert =
    !hasMeta(next, 'mindspace-geo') ||
    options.forceGeo ||
    !String(existingGeo.summary ?? '').trim();
  if (needsGeoUpsert) {
    next = upsertMindspaceGeoMeta(next, geoPayload, {
      forceReplace: options.forceGeo || !String(existingGeo.summary ?? '').trim(),
    });
  }
  next = repairMindspaceGeoMeta(next);
  if (options.fixWarnings !== false) {
    next = ensureSeoGeoWarningsFixed(next, options);
  }
  return next;
}