feat: improve mindspace asset handling and local runtime paths

2026-06-28 12:18:26 +08:00
parent 4a9bc710f1
commit ea19ffb5fa
36 changed files with 867 additions and 100 deletions
@@ -1,8 +1,186 @@
 import fs from 'node:fs/promises';
 import path from 'node:path';
-import { PUBLISH_ROOT_DIR, PUBLISH_KEY_UUID } from './user-publish.mjs';
+import { PUBLISH_ROOT_DIR, PUBLISH_KEY_UUID, PUBLIC_ZONE_DIR, resolvePublishDir } from './user-publish.mjs';
 import { workspaceThumbnailRelativePath } from './mindspace-workspace-thumbnails.mjs';

+const PRIVATE_ASSET_URL_PATTERN =
+  /(?:https?:\/\/[^/]+)?\/api\/mindspace\/v1\/assets\/([a-z0-9-]+)\/download(?:\?[^"'<>\\\s)]*)?/gi;
+
+const PUBLIC_TEMP_IMAGE_DIR = '.tmp-images';
+
+function extensionForMime(mimeType, filename = '') {
+  const ext = path.extname(String(filename ?? '')).replace(/^\./, '').toLowerCase();
+  if (ext && /^[a-z0-9]{1,8}$/.test(ext)) return ext === 'jpeg' ? 'jpg' : ext;
+  if (mimeType === 'image/jpeg') return 'jpg';
+  if (mimeType === 'image/png') return 'png';
+  if (mimeType === 'image/webp') return 'webp';
+  if (mimeType === 'image/gif') return 'gif';
+  return 'bin';
+}
+
+function publicTempImageFilename(assetId, mimeType, fallbackFilename = '') {
+  return `${assetId}.${extensionForMime(mimeType, fallbackFilename)}`;
+}
+
+function absoluteStoragePath(storageRoot, storageKey) {
+  const resolvedRoot = path.resolve(storageRoot);
+  const resolved = path.resolve(resolvedRoot, storageKey);
+  if (resolved !== resolvedRoot && !resolved.startsWith(`${resolvedRoot}${path.sep}`)) {
+    throw Object.assign(new Error('路径越界'), { code: 'invalid_storage_path' });
+  }
+  return resolved;
+}
+
+function workspaceTempImageRelativeUrl(htmlRelativePath, filename) {
+  const htmlDir = path.posix.dirname(String(htmlRelativePath ?? '').replace(/^\/+/, ''));
+  const tmpDir = path.posix.join(PUBLIC_ZONE_DIR, PUBLIC_TEMP_IMAGE_DIR);
+  const relDir =
+    htmlDir === '.' || !htmlDir
+      ? tmpDir
+      : path.posix.relative(htmlDir, tmpDir).replace(/\\/g, '/');
+  return relDir ? `${relDir}/${filename}` : filename;
+}
+
+export async function materializePrivateAssetsInWorkspaceHtml({
+  pool,
+  storageRoot,
+  h5Root,
+  userId,
+  html,
+  htmlRelativePath,
+  writeBack = false,
+}) {
+  const matches = [...String(html).matchAll(PRIVATE_ASSET_URL_PATTERN)];
+  if (matches.length === 0) {
+    return { html, count: 0, changed: false };
+  }
+
+  const assetIds = [...new Set(matches.map((match) => match[1]).filter(Boolean))];
+  const [assets] = await pool.query(
+    `SELECT a.id, a.mime_type, a.original_filename, v.storage_key
+     FROM h5_assets a
+     JOIN h5_asset_versions v ON v.id = a.current_version_id
+     WHERE a.user_id = ? AND a.id IN (?) AND a.mime_type LIKE 'image/%'
+       AND a.status <> 'deleted'`,
+    [userId, assetIds],
+  );
+
+  const publishDir = resolvePublishDir(h5Root, { id: userId });
+  const tmpDir = path.join(publishDir, PUBLIC_ZONE_DIR, PUBLIC_TEMP_IMAGE_DIR);
+  const urlMap = new Map();
+
+  for (const asset of assets) {
+    const filename = publicTempImageFilename(asset.id, asset.mime_type, asset.original_filename);
+    const target = path.join(tmpDir, filename);
+    try {
+      const sourcePath = absoluteStoragePath(storageRoot, asset.storage_key);
+      await fs.mkdir(tmpDir, { recursive: true });
+      await fs.copyFile(sourcePath, target);
+      urlMap.set(asset.id, workspaceTempImageRelativeUrl(htmlRelativePath, filename));
+    } catch {
+      // skip unreadable assets
+    }
+  }
+
+  if (urlMap.size === 0) {
+    return { html, count: 0, changed: false };
+  }
+
+  let count = 0;
+  const result = String(html).replace(PRIVATE_ASSET_URL_PATTERN, (value, assetId) => {
+    const nextUrl = urlMap.get(assetId);
+    if (!nextUrl) return value;
+    count += 1;
+    return nextUrl;
+  });
+
+  if (count === 0 || result === html) {
+    return { html, count: 0, changed: false };
+  }
+
+  if (writeBack) {
+    const htmlPath = path.join(publishDir, htmlRelativePath);
+    await fs.writeFile(htmlPath, result, 'utf8');
+  }
+
+  return { html: result, count, changed: true };
+}
+
+export async function repairMissingHtmlAssetReferences({ pool, userId, sourceContent, html }) {
+  const collectIds = (text) => [
+    ...new Set([...String(text).matchAll(PRIVATE_ASSET_URL_PATTERN)].map((match) => match[1])),
+  ];
+  const htmlIds = collectIds(html);
+  const messageIds = collectIds(sourceContent);
+  if (htmlIds.length === 0 || messageIds.length === 0) {
+    return { html, changed: false };
+  }
+
+  const [rows] = await pool.query(
+    `SELECT id FROM h5_assets
+     WHERE user_id = ? AND id IN (?) AND mime_type LIKE 'image/%' AND status <> 'deleted'`,
+    [userId, [...new Set([...htmlIds, ...messageIds])]],
+  );
+  const existing = new Set(rows.map((row) => row.id));
+  const invalidHtmlIds = htmlIds.filter((id) => !existing.has(id));
+  const validMessageIds = messageIds.filter((id) => existing.has(id));
+  if (invalidHtmlIds.length === 0 || validMessageIds.length === 0) {
+    return { html, changed: false };
+  }
+
+  const idMap = new Map();
+  for (let index = 0; index < invalidHtmlIds.length; index += 1) {
+    const replacement = validMessageIds[index];
+    if (!replacement) break;
+    idMap.set(invalidHtmlIds[index], replacement);
+  }
+  if (idMap.size === 0) {
+    return { html, changed: false };
+  }
+
+  let changed = false;
+  const nextHtml = String(html).replace(PRIVATE_ASSET_URL_PATTERN, (value, assetId) => {
+    const replacement = idMap.get(assetId);
+    if (!replacement) return value;
+    changed = true;
+    return value.replace(assetId, replacement);
+  });
+  return { html: nextHtml, changed };
+}
+
+export function createAssetDataUriResolver(pool, storageRoot, userId) {
+  const cache = new Map();
+  return async (assetId) => {
+    const key = String(assetId ?? '').trim();
+    if (!key) return null;
+    if (cache.has(key)) return cache.get(key);
+    const [rows] = await pool.query(
+      `SELECT a.mime_type, v.storage_key
+       FROM h5_assets a
+       JOIN h5_asset_versions v ON v.id = a.current_version_id
+       WHERE a.user_id = ? AND a.id = ? AND a.mime_type LIKE 'image/%'
+         AND a.status <> 'deleted'
+       LIMIT 1`,
+      [userId, key],
+    );
+    const asset = rows[0];
+    if (!asset) {
+      cache.set(key, null);
+      return null;
+    }
+    try {
+      const buffer = await fs.readFile(absoluteStoragePath(storageRoot, asset.storage_key));
+      const mimeType = String(asset.mime_type || 'image/jpeg');
+      const dataUri = `data:${mimeType};base64,${buffer.toString('base64')}`;
+      cache.set(key, dataUri);
+      return dataUri;
+    } catch {
+      cache.set(key, null);
+      return null;
+    }
+  };
+}
+
 const URL_PATTERN =
  /https?:\/\/[^\s<>"')\]]+\/(?:MindSpace|temp)\/([0-9a-f-]{36}|[a-z0-9._-]+)\/([^\s<>"')\]]+\.html)/gi;