Record published companion artifacts

2026-07-02 21:47:10 +08:00
parent b38dd5961f
commit c86a3fd15b
2 changed files with 178 additions and 0 deletions
@@ -7,7 +7,9 @@ import { pageInternals } from './mindspace-pages.mjs';
 import { loadMindSpaceConfig } from './mindspace-config.mjs';
 import { buildPublicUrl, resolvePublicBaseUrl, resolvePublishDir } from './user-publish.mjs';
 import {
+  isRelativeDownloadReference,
  loadWorkspaceDownloadLinkIndex,
+  resolveWorkspaceRelativeFilePath,
  rewriteRelativeDownloadLinks,
  inferWorkspaceHtmlRelativePath,
 } from './mindspace-html-download-links.mjs';
@@ -16,6 +18,7 @@ import { createImgproxySigner } from './imgproxy-signer.mjs';
 const SCANNER_VERSION = 'mindspace-content-v1';
 const PRIVATE_ASSET_DOWNLOAD_URL_PATTERN =
  /(?:https?:\/\/[^/]+)?\/api\/mindspace\/v1\/assets\/([a-z0-9-]+)\/download(?:\?[^"'<>\\\s)]*)?/gi;
+const URL_ATTR_PATTERN = /(\b(?:href|src)\s*=\s*["'])([^"']+)(["'])/gi;
 const ACCESS_MODES = new Set([
  'public',
  'password',
@@ -165,6 +168,71 @@ function buildPublicationThumbnailFallback(ownerSlug, urlSlug) {
  return `/u/${encodeURIComponent(ownerSlug)}/pages/${encodeURIComponent(urlSlug)}.thumbnail.png`;
 }

+function companionKindForWorkspacePath(workspacePath) {
+  const normalized = String(workspacePath ?? '').toLowerCase();
+  if (normalized.endsWith('.long.png')) return 'long_image';
+  if (normalized.endsWith('.docx') || normalized.endsWith('.doc')) return 'docx';
+  if (normalized.endsWith('.pdf')) return 'pdf';
+  return 'generated_file';
+}
+
+function companionMimeForWorkspacePath(workspacePath) {
+  const normalized = String(workspacePath ?? '').toLowerCase();
+  if (normalized.endsWith('.long.png') || normalized.endsWith('.png')) return 'image/png';
+  if (normalized.endsWith('.docx')) {
+    return 'application/vnd.openxmlformats-officedocument.wordprocessingml.document';
+  }
+  if (normalized.endsWith('.doc')) return 'application/msword';
+  if (normalized.endsWith('.pdf')) return 'application/pdf';
+  if (normalized.endsWith('.csv')) return 'text/csv';
+  if (normalized.endsWith('.md')) return 'text/markdown';
+  if (normalized.endsWith('.txt')) return 'text/plain';
+  return 'application/octet-stream';
+}
+
+async function statWorkspaceFile(publishDir, workspacePath) {
+  if (!publishDir || !workspacePath) return null;
+  const normalized = String(workspacePath).replace(/^\/+/, '').replace(/\\/g, '/');
+  const root = path.resolve(publishDir);
+  const absolutePath = path.resolve(root, ...normalized.split('/'));
+  if (absolutePath !== root && !absolutePath.startsWith(`${root}${path.sep}`)) return null;
+  try {
+    const stat = await fs.stat(absolutePath);
+    if (!stat.isFile()) return null;
+    return { workspacePath: normalized, sizeBytes: stat.size };
+  } catch (error) {
+    if (error?.code !== 'ENOENT') throw error;
+    return null;
+  }
+}
+
+async function collectPublicationCompanionArtifacts({
+  publishDir,
+  html = '',
+  htmlRelativePath = null,
+}) {
+  if (!publishDir || !htmlRelativePath) return [];
+  const candidates = new Set();
+  const normalizedHtmlPath = String(htmlRelativePath).replace(/^\/+/, '').replace(/\\/g, '/');
+  if (normalizedHtmlPath.toLowerCase().endsWith('.html')) {
+    candidates.add(normalizedHtmlPath.replace(/\.html$/i, '.docx'));
+    candidates.add(normalizedHtmlPath.replace(/\.html$/i, '.long.png'));
+  }
+  let match;
+  while ((match = URL_ATTR_PATTERN.exec(String(html ?? ''))) !== null) {
+    const ref = match[2];
+    if (isRelativeDownloadReference(ref)) {
+      candidates.add(resolveWorkspaceRelativeFilePath(normalizedHtmlPath, ref));
+    }
+  }
+  const artifacts = [];
+  for (const candidate of candidates) {
+    const stat = await statWorkspaceFile(publishDir, candidate);
+    if (stat) artifacts.push(stat);
+  }
+  return artifacts;
+}
+
 const PUBLIC_IMAGE_STORAGE_KEY_PATTERN = /^users\/([^/]+)\/images\/(\d{4}-\d{2}-\d{2}\/[^?#"'<>@\s]+)$/i;

 function storageKeyToPublicStandardImageUrl(storageKey, publicBaseUrl) {
@@ -755,6 +823,8 @@ export function createPublicationService(pool, options = {}) {
        bundleAssetId,
        storageKey,
        htmlBytes,
+        html: page.content,
+        publishDir: h5Root ? resolvePublishDir(h5Root, { id: userId }) : null,
        now,
      });
      return publication;
@@ -1067,6 +1137,8 @@ async function registerPublicationArtifactForConversation({
  bundleAssetId,
  storageKey,
  htmlBytes,
+  html,
+  publishDir,
  now,
 }) {
  const sessionId = String(page?.source_session_id ?? '').trim();
@@ -1095,6 +1167,38 @@ async function registerPublicationArtifactForConversation({
      sortOrder: Number.isFinite(Number(publication.publishedAt)) ? Number(publication.publishedAt) : 0,
      now,
    });
+    const htmlRelativePath = page.source_relative_path ?? null;
+    const companionArtifacts = await collectPublicationCompanionArtifacts({
+      publishDir,
+      html,
+      htmlRelativePath,
+    });
+    await Promise.all(
+      companionArtifacts.map((companion, index) =>
+        registry.recordArtifact({
+          id: `ca_${publication.id}_${crypto
+            .createHash('sha256')
+            .update(companion.workspacePath)
+            .digest('hex')
+            .slice(0, 12)}`,
+          packageId: packageRecord.id,
+          artifactKind: companionKindForWorkspacePath(companion.workspacePath),
+          role: 'assistant',
+          pageId: page.page_id,
+          publicationId: publication.id,
+          messageId: page.source_message_id ?? null,
+          displayName: path.posix.basename(companion.workspacePath),
+          mimeType: companionMimeForWorkspacePath(companion.workspacePath),
+          sizeBytes: companion.sizeBytes,
+          canonicalUrl: buildPublicUrl(resolvePublicBaseUrl(), userId, companion.workspacePath),
+          sortOrder:
+            (Number.isFinite(Number(publication.publishedAt)) ? Number(publication.publishedAt) : 0) +
+            index +
+            1,
+          now,
+        }),
+      ),
+    );
    await registry.writeManifestForSession({ userId, sessionId });
    return artifact;
  } catch (error) {
@@ -1121,6 +1225,7 @@ export const publicationInternals = {
  workspacePublicAssetRelativeUrl,
  rewriteWorkspacePublicAssetReferences,
  localizePrivateImageReferences,
+  collectPublicationCompanionArtifacts,
  prepareHtmlPublishContent,
  registerPublicationArtifactForConversation,
  scanContent, // re-exported from mindspace-content-scan.mjs
@@ -275,6 +275,79 @@ test('registerPublicationArtifactForConversation records public html artifact',
  ]);
 });

+test('collectPublicationCompanionArtifacts finds real download and long-image files', async () => {
+  const publishDir = await fs.mkdtemp(path.join(os.tmpdir(), 'publication-companions-'));
+  await fs.mkdir(path.join(publishDir, 'public'), { recursive: true });
+  await fs.writeFile(path.join(publishDir, 'public', 'report.docx'), 'docx');
+  await fs.writeFile(path.join(publishDir, 'public', 'report.long.png'), 'png');
+  await fs.writeFile(path.join(publishDir, 'public', 'appendix.pdf'), 'pdf');
+
+  const artifacts = await publicationInternals.collectPublicationCompanionArtifacts({
+    publishDir,
+    htmlRelativePath: 'public/report.html',
+    html: '<a href="appendix.pdf" download>PDF</a>',
+  });
+
+  assert.deepEqual(
+    artifacts.map((item) => item.workspacePath).sort(),
+    ['public/appendix.pdf', 'public/report.docx', 'public/report.long.png'],
+  );
+});
+
+test('registerPublicationArtifactForConversation records public companions', async () => {
+  const publishDir = await fs.mkdtemp(path.join(os.tmpdir(), 'publication-companions-'));
+  await fs.mkdir(path.join(publishDir, 'public'), { recursive: true });
+  await fs.writeFile(path.join(publishDir, 'public', 'report.docx'), 'docx');
+  await fs.writeFile(path.join(publishDir, 'public', 'report.long.png'), 'png');
+  const calls = [];
+  const registry = {
+    async ensurePackage(input) {
+      calls.push(['ensurePackage', input]);
+      return { id: 'cp_session-1' };
+    },
+    async recordArtifact(input) {
+      calls.push(['recordArtifact', input]);
+      return input;
+    },
+    async writeManifestForSession(input) {
+      calls.push(['writeManifestForSession', input]);
+      return { storageKey: 'users/user-1/conversations/session-1/manifest.json' };
+    },
+  };
+
+  await publicationInternals.registerPublicationArtifactForConversation({
+    registry,
+    userId: 'user-1',
+    page: {
+      page_id: 'page-1',
+      title: '发布页',
+      source_session_id: 'session-1',
+      source_message_id: 'message-1',
+      source_relative_path: 'public/report.html',
+    },
+    publication: {
+      id: 'pub-1',
+      publicUrl: 'https://m.tkmind.cn/u/john/pages/published',
+      publishedAt: 3000,
+    },
+    bundleAssetId: 'asset-pub',
+    storageKey: 'users/user-1/publications/pub-1/index.html',
+    htmlBytes: 256,
+    html: '<!doctype html><title>发布页</title>',
+    publishDir,
+    now: 3000,
+  });
+
+  const artifactKinds = calls
+    .filter(([name]) => name === 'recordArtifact')
+    .map(([, input]) => input.artifactKind)
+    .sort();
+  assert.deepEqual(artifactKinds, ['docx', 'long_image', 'public_html']);
+  const docx = calls.find(([, input]) => input?.artifactKind === 'docx')?.[1];
+  assert.equal(docx.displayName, 'report.docx');
+  assert.match(docx.canonicalUrl, /\/MindSpace\/user-1\/public\/report\.docx$/);
+});
+
 test('registerPublicationArtifactForConversation skips pages without source session', async () => {
  assert.equal(
    await publicationInternals.registerPublicationArtifactForConversation({