memind/mindspace-public-delivery.mjs

import crypto from 'node:crypto';
import fs from 'node:fs';
import path from 'node:path';
import { injectMindSpacePageDataContext } from './mindspace-public-page-context.mjs';
import { preparePublishedPlatformBrand } from './mindspace-page-tag.mjs';
import { injectPublicImageRetryScript } from './mindspace-public-image-retry.mjs';
import { applyWechatSurveyCompat } from './mindspace-page-data-wechat-survey-compat.mjs';
import { stripPublicationHtmlCspMeta } from './plaza-embed.mjs';
import {
  decorateMindSpaceSeoGeoHtml,
} from './mindspace-seo-geo-delivery.mjs';
import { stripPublicFileShareWidget } from './mindspace-public-share-widget.mjs';

const INLINE_SCRIPT_PATTERN = /<script\b(?![^>]*\bsrc\b)[^>]*>([\s\S]*?)<\/script>/gi;

export function collectInlineScriptHashes(html) {
  const hashes = [];
  const seen = new Set();
  for (const match of String(html ?? '').matchAll(INLINE_SCRIPT_PATTERN)) {
    const hash = crypto.createHash('sha256').update(match[1] ?? '').digest('base64');
    if (seen.has(hash)) continue;
    seen.add(hash);
    hashes.push(hash);
  }
  return hashes;
}

export function shouldSkipPublicShareWidget({
  htmlFilePath = '',
  pageUrl = '',
  relativePath = '',
  requestPath = '',
} = {}) {
  const haystack = `${htmlFilePath}\n${pageUrl}\n${relativePath}\n${requestPath}`
    .replace(/\\/g, '/')
    .toLowerCase();
  return (
    haystack.includes('learning-assistant') ||
    haystack.includes('/public/learn/') ||
    /\/learn(?:\/|$|[?#])/.test(haystack)
  );
}

export async function handleMindSpaceLongImageDownload({
  query,
  filePath,
  pageUrl = null,
  res,
  isLongImageDownloadRequest,
  longImagePathForHtml,
  renderLongImage,
} = {}) {
  if (!isLongImageDownloadRequest?.(query)) return false;
  const longImagePath = longImagePathForHtml(filePath);
  try {
    await renderLongImage(pageUrl
      ? { url: pageUrl, outputPath: longImagePath }
      : { htmlPath: filePath, outputPath: longImagePath });
    res.set('Cache-Control', 'no-store');
    res.download(longImagePath, path.basename(longImagePath), (err) => {
      if (err && !res.headersSent) res.status(404).json({ message: '长图文件不存在' });
    });
  } catch (error) {
    res
      .status(500)
      .type('text/plain; charset=utf-8')
      .send(`长图生成失败：${error?.message || '未知错误'}`);
  }
  return true;
}

export function decorateMindSpacePublishedHtml({
  html,
  embed = false,
  isOwner = true,
  pageDataContext = null,
  context,
  htmlFilePath = '',
  relativePath = '',
  requestPath = '',
  fileExists = fs.existsSync,
  userAgent = '',
  preparePublicationHtmlForEmbed,
  injectOgTags,
  injectWechatShareBridge,
  injectPublicFileShareButton,
  publishedPageCsp,
  isWechatUserAgent,
  seoGeoConfig = null,
  publicationSnapshot = null,
} = {}) {
  // Preview HTML carries a restrictive inline CSP (often script-src 'none').
  // Published delivery sets the authoritative CSP response header below; keep
  // the preview meta out of the delivered document so Page Data and other
  // same-origin scripts are governed by that header instead of being blocked
  // by a stale preview policy.
  let nextHtml = stripPublicationHtmlCspMeta(html);
  let allowEmbedFrame = false;

  if (embed) {
    nextHtml = preparePublicationHtmlForEmbed(nextHtml);
    allowEmbedFrame = true;
  }

  try {
    nextHtml = preparePublishedPlatformBrand(nextHtml);
    nextHtml = injectOgTags(nextHtml, {
      origin: context.origin,
      pageUrl: context.pageUrl,
      pageDirUrl: context.pageDirUrl,
      fallbackImageUrl: context.fallbackImageUrl,
      htmlFilePath,
      fileExists,
    });
    const wechatShare = !embed && isWechatUserAgent(userAgent || '');
    if (wechatShare) {
      nextHtml = injectWechatShareBridge(nextHtml, { pageUrl: context.pageUrl });
    }
    const skipShareWidget = !embed && shouldSkipPublicShareWidget({
      htmlFilePath,
      pageUrl: context?.pageUrl ?? '',
      relativePath,
      requestPath,
    });
    if (skipShareWidget) {
      nextHtml = stripPublicFileShareWidget(nextHtml);
    } else if (!embed) {
      nextHtml = injectPublicFileShareButton(nextHtml, { isOwner }).html;
    }
    // Retries transient asset-download failures client-side before a page's own onerror
    // permanently swaps the <img> for a placeholder. See mindspace-public-image-retry.mjs.
    if (!embed) {
      nextHtml = injectPublicImageRetryScript(nextHtml).html;
    }
    if (pageDataContext?.pageId) {
      nextHtml = injectMindSpacePageDataContext(nextHtml, pageDataContext);
    }
    if (!embed && isWechatUserAgent(userAgent || '')) {
      nextHtml = applyWechatSurveyCompat(nextHtml).html;
    }
    const seoGeoDecorated = decorateMindSpaceSeoGeoHtml(nextHtml, {
      seoGeoConfig,
      publicationSnapshot,
      pageDataContext,
      embed,
      context,
      htmlFilePath,
      fileExists,
    });
    nextHtml = seoGeoDecorated.html;
    const scriptHashes = collectInlineScriptHashes(nextHtml);
    return {
      html: nextHtml,
      csp: publishedPageCsp(nextHtml, {
        embed,
        wechatShare,
        scriptHashes,
      }),
      allowEmbedFrame,
      robotsHeader: seoGeoDecorated.robotsHeader,
    };
  } catch {
    return {
      html: nextHtml,
      csp: publishedPageCsp(nextHtml, { embed }),
      allowEmbedFrame,
      robotsHeader: null,
    };
  }
}