Skip to main content
Glama

loadBackup

Restore specific Heptabase MCP backups by specifying the backup path and ID, with optional extraction to access data directly.

Input Schema

TableJSON Schema
NameRequiredDescriptionDefault
backupIdNo
backupPathNo
extractNo

Implementation Reference

  • MCP tool handler for 'loadBackup'. Resolves backup path (from ID if provided), calls BackupManager.loadBackup, initializes/updates HeptabaseDataService with extracted data, returns success message.
    this.tools.loadBackup = { inputSchema: loadBackupSchema, handler: async (params) => { if (!this.backupManager) { this.backupManager = new BackupManager({ backupPath: this.config.backupPath, extractionPath: this.config.extractionPath, autoExtract: params.extract ?? this.config.autoExtract, watchDirectory: false, keepExtracted: this.config.keepExtracted }); } let backupPath = params.backupPath; // If backupId is provided, find the backup path if (params.backupId) { const backups = await this.backupManager.listBackups(); const backup = backups.find(b => b.backupId === params.backupId); if (!backup) { throw new Error('Backup not found'); } backupPath = backup.backupPath; } const metadata = await this.backupManager.loadBackup(backupPath!); // Initialize data service with the extracted path const dataPath = metadata.extractedPath || path.dirname(metadata.backupPath); // Update existing data service if it exists, otherwise create new one if (this.dataService) { // Update the data path and reload (this.dataService as any).config.dataPath = dataPath; await this.dataService.loadData(); } else { this.dataService = new HeptabaseDataService({ dataPath, cacheEnabled: this.config.cacheEnabled, cacheTTL: this.config.cacheTTL }); await this.dataService.loadData(); } return { content: [{ type: 'text', text: `Backup loaded successfully: ${metadata.backupId}` }] }; } };
  • Zod input schema for the loadBackup tool, requiring either backupPath or backupId.
    const loadBackupSchema = z.object({ backupPath: z.string().optional(), backupId: z.string().optional(), extract: z.boolean().optional() }).refine(data => data.backupPath || data.backupId, { message: 'Either backupPath or backupId must be provided' });
  • src/server.ts:273-280 (registration)
    MCP server tool registration for 'loadBackup', with inline schema and handler wrapper that validates input using loadBackupSchema.
    this.server.tool('loadBackup', { backupPath: z.string().optional(), backupId: z.string().optional(), extract: z.boolean().optional() }, async (params) => { const validated = loadBackupSchema.parse(params); return this.tools.loadBackup.handler(validated); });
  • Core implementation in BackupManager that loads backup metadata, extracts ZIP if needed and autoExtract is true, caches metadata, emits events.
    async loadBackup(backupPath: string): Promise<BackupMetadata> { this.emit('backup:load:start', backupPath); try { const stats = await fs.stat(backupPath); const fileName = path.basename(backupPath); const backupInfo = this.getBackupInfo(fileName); const isCompressed = fileName.endsWith('.zip'); let extractedPath: string | undefined; if (isCompressed && this.config.autoExtract) { const backupName = path.basename(backupPath, '.zip'); extractedPath = path.join(this.config.extractionPath, backupName); // Check if already extracted if (!(await fs.pathExists(extractedPath))) { await this.extractBackup(backupPath, extractedPath); } } const metadata: BackupMetadata = { backupId: path.basename(backupPath, path.extname(backupPath)), backupPath, createdDate: backupInfo.date, fileSize: stats.size, isCompressed, extractedPath, version: backupInfo.version }; this.loadedBackups.set(metadata.backupId, metadata); this.emit('backup:load:complete', metadata); return metadata; } catch (error) { this.emit('backup:load:error', error); throw error; } }

Other Tools

Related Tools

Latest Blog Posts

MCP directory API

We provide all the information about MCP servers via our MCP API.

curl -X GET 'https://glama.ai/api/mcp/v1/servers/LarryStanley/heptabse-mcp'

If you have feedback or need assistance with the MCP directory API, please join our Discord server