Merge pull request #29 from Patrick-Ehimen/feat/dataset-management

feat: implement comprehensive dataset management functionality

Author: Patrick-Ehimen

apps/mcp-server/src/__tests__/dataset-tools.test.ts

@@ -0,0 +1,254 @@
+/**
+ * Dataset Tools Tests
+ */
+
+import { describe, it, expect, beforeEach, afterEach, vi } from "vitest";
+import { Logger } from "@lighthouse-tooling/shared";
+import { MockLighthouseService } from "../services/MockLighthouseService.js";
+import {
+  LighthouseCreateDatasetTool,
+  LighthouseListDatasetsTool,
+  LighthouseGetDatasetTool,
+  LighthouseUpdateDatasetTool,
+} from "../tools/index.js";
+
+describe("Dataset Tools", () => {
+  let mockService: MockLighthouseService;
+  let logger: Logger;
+
+  beforeEach(() => {
+    mockService = new MockLighthouseService();
+    logger = Logger.getInstance({ level: "error", component: "test" });
+  });
+
+  afterEach(() => {
+    mockService.clear();
+  });
+
+  describe("LighthouseCreateDatasetTool", () => {
+    let tool: LighthouseCreateDatasetTool;
+
+    beforeEach(() => {
+      tool = new LighthouseCreateDatasetTool(mockService, logger);
+    });
+
+    it("should have correct tool definition", () => {
+      const definition = LighthouseCreateDatasetTool.getDefinition();
+      expect(definition.name).toBe("lighthouse_create_dataset");
+      expect(definition.description).toContain("Create a new dataset");
+      expect(definition.requiresAuth).toBe(true);
+    });
+
+    it("should validate required parameters", async () => {
+      const result = await tool.execute({});
+      expect(result.success).toBe(false);
+      expect(result.error).toContain("name is required");
+    });
+
+    it("should validate filePaths parameter", async () => {
+      const result = await tool.execute({
+        name: "Test Dataset",
+        filePaths: [],
+      });
+      expect(result.success).toBe(false);
+      expect(result.error).toContain("filePaths is required and must be a non-empty array");
+    });
+
+    it("should create dataset successfully with valid parameters", async () => {
+      // Create test files first
+      const testFiles = ["/tmp/test1.txt", "/tmp/test2.txt"];
+
+      // Mock FileUtils to avoid actual file operations
+      const { FileUtils } = await import("@lighthouse-tooling/shared");
+      vi.spyOn(FileUtils, "fileExists").mockResolvedValue(true);
+      vi.spyOn(FileUtils, "getFileInfo").mockResolvedValue({
+        path: "/tmp/test.txt",
+        name: "test.txt",
+        extension: ".txt",
+        size: 1024,
+        lastModified: new Date(),
+        hash: "mock-hash",
+      });
+
+      const result = await tool.execute({
+        name: "Test Dataset",
+        description: "A test dataset",
+        filePaths: testFiles,
+        encrypt: false,
+        tags: ["test", "dataset"],
+      });
+
+      expect(result.success).toBe(true);
+      expect(result.data).toBeDefined();
+      expect(result.data.dataset.name).toBe("Test Dataset");
+      expect(result.data.dataset.fileCount).toBe(2);
+    });
+
+    it("should handle file validation errors", async () => {
+      // Mock FileUtils to simulate file not found
+      const { FileUtils } = await import("@lighthouse-tooling/shared");
+      vi.spyOn(FileUtils, "fileExists").mockResolvedValue(false);
+
+      const result = await tool.execute({
+        name: "Test Dataset",
+        filePaths: ["/nonexistent/file.txt"],
+      });
+
+      expect(result.success).toBe(false);
+      expect(result.error).toContain("File not found");
+    });
+  });
+
+  describe("LighthouseListDatasetsTool", () => {
+    let tool: LighthouseListDatasetsTool;
+
+    beforeEach(() => {
+      tool = new LighthouseListDatasetsTool(mockService, logger);
+    });
+
+    it("should have correct tool definition", () => {
+      const definition = LighthouseListDatasetsTool.getDefinition();
+      expect(definition.name).toBe("lighthouse_list_datasets");
+      expect(definition.description).toContain("List all datasets");
+      expect(definition.requiresAuth).toBe(true);
+    });
+
+    it("should list datasets with default parameters", async () => {
+      const result = await tool.execute({});
+      expect(result.success).toBe(true);
+      expect(result.data).toBeDefined();
+      expect(result.data.datasets).toBeInstanceOf(Array);
+      expect(result.data.pagination).toBeDefined();
+    });
+
+    it("should validate limit parameter", async () => {
+      const result = await tool.execute({
+        limit: 150, // exceeds max of 100
+      });
+      expect(result.success).toBe(false);
+      expect(result.error).toContain("limit must be between 1 and 100");
+    });
+
+    it("should validate offset parameter", async () => {
+      const result = await tool.execute({
+        offset: -1,
+      });
+      expect(result.success).toBe(false);
+      expect(result.error).toContain("offset must be 0 or greater");
+    });
+  });
+
+  describe("LighthouseGetDatasetTool", () => {
+    let tool: LighthouseGetDatasetTool;
+
+    beforeEach(() => {
+      tool = new LighthouseGetDatasetTool(mockService, logger);
+    });
+
+    it("should have correct tool definition", () => {
+      const definition = LighthouseGetDatasetTool.getDefinition();
+      expect(definition.name).toBe("lighthouse_get_dataset");
+      expect(definition.description).toContain("Retrieve detailed information");
+      expect(definition.requiresAuth).toBe(true);
+    });
+
+    it("should validate datasetId parameter", async () => {
+      const result = await tool.execute({});
+      expect(result.success).toBe(false);
+      expect(result.error).toContain("datasetId is required");
+    });
+
+    it("should retrieve dataset successfully", async () => {
+      // First create a dataset
+      await mockService.createDataset({
+        name: "Test Dataset",
+        description: "Test description",
+        filePaths: [],
+        encrypt: false,
+      });
+
+      // Get the created dataset ID (it will be the first one)
+      const datasets = await mockService.listDatasets();
+      const datasetId = datasets.datasets[0].id;
+
+      const result = await tool.execute({
+        datasetId,
+      });
+      expect(result.success).toBe(true);
+      expect(result.data).toBeDefined();
+      expect(result.data.dataset.id).toBe(datasetId);
+    });
+
+    it("should handle non-existent dataset", async () => {
+      const result = await tool.execute({
+        datasetId: "nonexistent",
+      });
+      expect(result.success).toBe(false);
+      expect(result.error).toContain("Dataset not found");
+    });
+  });
+
+  describe("LighthouseUpdateDatasetTool", () => {
+    let tool: LighthouseUpdateDatasetTool;
+
+    beforeEach(() => {
+      tool = new LighthouseUpdateDatasetTool(mockService, logger);
+    });
+
+    it("should have correct tool definition", () => {
+      const definition = LighthouseUpdateDatasetTool.getDefinition();
+      expect(definition.name).toBe("lighthouse_update_dataset");
+      expect(definition.description).toContain("Update an existing dataset");
+      expect(definition.requiresAuth).toBe(true);
+    });
+
+    it("should validate datasetId parameter", async () => {
+      const result = await tool.execute({});
+      expect(result.success).toBe(false);
+      expect(result.error).toContain("datasetId is required");
+    });
+
+    it("should require at least one update operation", async () => {
+      const result = await tool.execute({
+        datasetId: "dataset_1",
+      });
+      expect(result.success).toBe(false);
+      expect(result.error).toContain("At least one update operation must be specified");
+    });
+
+    it("should update dataset with new description", async () => {
+      // First create a dataset
+      await mockService.createDataset({
+        name: "Test Dataset",
+        description: "Original description",
+        filePaths: [],
+        encrypt: false,
+      });
+
+      // Get the created dataset ID
+      const datasets = await mockService.listDatasets();
+      const datasetId = datasets.datasets[0].id;
+
+      const result = await tool.execute({
+        datasetId,
+        description: "Updated description",
+      });
+      expect(result.success).toBe(true);
+      expect(result.data).toBeDefined();
+      expect(result.data.dataset.description).toBe("Updated description");
+    });
+
+    it("should validate addFiles parameter", async () => {
+      // Mock FileUtils to simulate file not found
+      const { FileUtils } = await import("@lighthouse-tooling/shared");
+      vi.spyOn(FileUtils, "fileExists").mockResolvedValue(false);
+
+      const result = await tool.execute({
+        datasetId: "dataset_1",
+        addFiles: ["/nonexistent/file.txt"],
+      });
+      expect(result.success).toBe(false);
+      expect(result.error).toContain("File not found");
+    });
+  });
+});

apps/mcp-server/src/server.ts

@@ -19,6 +19,10 @@ import { MockDatasetService } from "./services/MockDatasetService.js";
 import {
   LighthouseUploadFileTool,
   LighthouseFetchFileTool,
+  LighthouseCreateDatasetTool,
+  LighthouseListDatasetsTool,
+  LighthouseGetDatasetTool,
+  LighthouseUpdateDatasetTool,
   LighthouseGenerateKeyTool,
   LighthouseSetupAccessControlTool,
 } from "./tools/index.js";
@@ -125,56 +129,59 @@ export class LighthouseMCPServer {
     // Create tool instances with service dependencies
     const uploadFileTool = new LighthouseUploadFileTool(this.lighthouseService, this.logger);
     const fetchFileTool = new LighthouseFetchFileTool(this.lighthouseService, this.logger);
+    const createDatasetTool = new LighthouseCreateDatasetTool(this.lighthouseService, this.logger);
+    const listDatasetsTool = new LighthouseListDatasetsTool(this.lighthouseService, this.logger);
+    const getDatasetTool = new LighthouseGetDatasetTool(this.lighthouseService, this.logger);
+    const updateDatasetTool = new LighthouseUpdateDatasetTool(this.lighthouseService, this.logger);
     const generateKeyTool = new LighthouseGenerateKeyTool(this.lighthouseService, this.logger);
     const setupAccessControlTool = new LighthouseSetupAccessControlTool(
       this.lighthouseService,
       this.logger,
     );
 
-    // Register lighthouse_upload_file tool
+    // Register file operation tools
     this.registry.register(
       LighthouseUploadFileTool.getDefinition(),
       async (args) => await uploadFileTool.execute(args),
     );
 
-    // Register lighthouse_fetch_file tool
     this.registry.register(
       LighthouseFetchFileTool.getDefinition(),
       async (args) => await fetchFileTool.execute(args),
     );
 
-    // Register lighthouse_generate_key tool
+    // Register dataset management tools
+    this.registry.register(
+      LighthouseCreateDatasetTool.getDefinition(),
+      async (args) => await createDatasetTool.execute(args),
+    );
+
+    this.registry.register(
+      LighthouseListDatasetsTool.getDefinition(),
+      async (args) => await listDatasetsTool.execute(args),
+    );
+
+    this.registry.register(
+      LighthouseGetDatasetTool.getDefinition(),
+      async (args) => await getDatasetTool.execute(args),
+    );
+
+    this.registry.register(
+      LighthouseUpdateDatasetTool.getDefinition(),
+      async (args) => await updateDatasetTool.execute(args),
+    );
+
+    // Register encryption tools
     this.registry.register(
       LighthouseGenerateKeyTool.getDefinition(),
       async (args) => await generateKeyTool.execute(args),
     );
 
-    // Register lighthouse_setup_access_control tool
     this.registry.register(
       LighthouseSetupAccessControlTool.getDefinition(),
       async (args) => await setupAccessControlTool.execute(args),
     );
 
-    // Register lighthouse_create_dataset tool (keeping existing implementation)
-    const datasetTool = LIGHTHOUSE_MCP_TOOLS.find((t) => t.name === "lighthouse_create_dataset");
-    if (datasetTool) {
-      this.registry.register(datasetTool, async (args) => {
-        const result = await this.datasetService.createDataset({
-          name: args.name as string,
-          description: args.description as string | undefined,
-          files: args.files as string[],
-          metadata: args.metadata as Record<string, unknown> | undefined,
-          encrypt: args.encrypt as boolean | undefined,
-        });
-
-        return {
-          success: true,
-          data: result,
-          executionTime: 0,
-        };
-      });
-    }
-
     const registeredTools = this.registry.listTools();
     const registrationTime = Date.now() - startTime;