@nocobase/ai 2.3.0-alpha.1 → 3.0.0-alpha.2

package/lib/document-loader/index.d.ts

@@ -10,5 +10,7 @@ import { Document } from '@langchain/core/documents';
 export type DocumentLoaderWorkerOptions = {
     filePath: string;
     mimeType?: string;
+    /** Timeout in milliseconds for the worker to complete. Defaults to 5 minutes. */
+    timeout?: number;
 };
 export declare const loadByWorker: (extname: string, options: DocumentLoaderWorkerOptions) => Promise<Document[]>;

package/lib/document-loader/index.js

@@ -42,12 +42,14 @@ __export(document_loader_exports, {
 module.exports = __toCommonJS(document_loader_exports);
 var import_node_worker_threads = require("node:worker_threads");
 var import_node_path = __toESM(require("node:path"));
+const DEFAULT_WORKER_TIMEOUT = 5 * 60 * 1e3;
 const loadByWorker = /* @__PURE__ */ __name(async (extname, options) => {
   const isTsRuntime = __filename.endsWith(".ts");
   const workerPath = import_node_path.default.join(__dirname, `loader.worker.${isTsRuntime ? "ts" : "js"}`);
   const worker = new import_node_worker_threads.Worker(workerPath, {
     execArgv: isTsRuntime ? ["--require", "tsx/cjs"] : void 0
   });
+  const timeout = options.timeout ?? DEFAULT_WORKER_TIMEOUT;
   return new Promise((resolve, reject) => {
     let settled = false;
     const close = /* @__PURE__ */ __name((error, result) => {
@@ -55,12 +57,16 @@ const loadByWorker = /* @__PURE__ */ __name(async (extname, options) => {
         return;
       }
       settled = true;
+      clearTimeout(timer);
       if (error) {
         reject(error);
         return;
       }
       resolve(result || []);
     }, "close");
+    const timer = setTimeout(() => {
+      close(new Error(`Document loading timed out after ${Math.round(timeout / 1e3)}s`));
+    }, timeout);
     worker.once("message", (payload) => {
       if (payload == null ? void 0 : payload.error) {
         close(new Error(payload.error));

package/lib/document-loader/loader.worker.js

@@ -19,16 +19,24 @@ var import_xlsx = require("./xlsx");
 var _a;
 const loadPdf = /* @__PURE__ */ __name(async (filePath) => {
   const loader = new import_pdf.PDFLoader(filePath);
-  return loader.load();
+  try {
+    return await loader.load();
+  } catch (error) {
+    const err = error;
+    if ((err == null ? void 0 : err.name) === "PasswordException" || /password/i.test(err == null ? void 0 : err.message)) {
+      throw new Error("The PDF file is password-protected and cannot be parsed. Please upload an unlocked version.");
+    }
+    throw error;
+  }
 }, "loadPdf");
 const loadDoc = /* @__PURE__ */ __name(async (filePath, type) => {
   const loader = new import_docx.DocxLoader(filePath, { type });
   return loader.load();
 }, "loadDoc");
-const loadPpt = /* @__PURE__ */ __name(async (filePath) => {
+const loadPptx = /* @__PURE__ */ __name(async (filePath) => {
   const loader = new import_pptx.PPTXLoader(filePath);
   return loader.load();
-}, "loadPpt");
+}, "loadPptx");
 const loadTxt = /* @__PURE__ */ __name(async (filePath) => {
   const loader = new import_text.TextLoader(filePath);
   return loader.load();
@@ -41,9 +49,8 @@ const loadByExtname = /* @__PURE__ */ __name(async (payload) => {
   switch (payload.extname) {
     case ".pdf":
       return loadPdf(payload.filePath);
-    case ".ppt":
     case ".pptx":
-      return loadPpt(payload.filePath);
+      return loadPptx(payload.filePath);
     case ".doc":
       return loadDoc(payload.filePath, "doc");
     case ".docx":
@@ -52,6 +59,7 @@ const loadByExtname = /* @__PURE__ */ __name(async (payload) => {
       return loadCsv(payload.filePath);
     case ".xls":
     case ".xlsx":
+    case ".xlsm":
       return (0, import_xlsx.loadXlsx)(payload.filePath, payload.mimeType);
     case ".json":
     case ".md":
@@ -74,8 +82,9 @@ const loadByExtname = /* @__PURE__ */ __name(async (payload) => {
     };
     (_a2 = import_node_worker_threads.parentPort) == null ? void 0 : _a2.postMessage(response);
   } catch (error) {
+    const err = error;
     const response = {
-      error: String((error == null ? void 0 : error.stack) || error)
+      error: (err == null ? void 0 : err.message) || String(error)
     };
     (_b = import_node_worker_threads.parentPort) == null ? void 0 : _b.postMessage(response);
   }

package/lib/tools-manager/index.d.ts

@@ -22,6 +22,7 @@ export declare class DefaultToolsManager implements ToolsManager {
 }
 export declare function defineTools(options: ToolsOptions): ToolsOptions;
 export declare const SYSTEM_TOOLS: {
+    GET_SKILL: string;
     WEB_SEARCH: string;
     KNOWLEDGE_BASE: string;
     WORK_FLOW_TASK_OUTPUT: string;

package/lib/tools-manager/index.js

@@ -132,6 +132,7 @@ function defineTools(options) {
 }
 __name(defineTools, "defineTools");
 const SYSTEM_TOOLS = {
+  GET_SKILL: "getSkill",
   WEB_SEARCH: "subAgentWebSearch",
   KNOWLEDGE_BASE: "knowledge-base-retrieve",
   WORK_FLOW_TASK_OUTPUT: "aiEmployeeWorkflowTaskOutput"

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@nocobase/ai",
-  "version": "2.3.0-alpha.1",
+  "version": "3.0.0-alpha.2",
   "description": "",
   "license": "Apache-2.0",
   "main": "./lib/index.js",
@@ -17,17 +17,17 @@
     "@langchain/mcp-adapters": "1.1.3",
     "@langchain/ollama": "1.2.7",
     "@langchain/openai": "1.4.7",
-    "@nocobase/data-source-manager": "2.3.0-alpha.1",
-    "@nocobase/logger": "2.3.0-alpha.1",
-    "@nocobase/resourcer": "2.3.0-alpha.1",
-    "@nocobase/utils": "2.3.0-alpha.1",
+    "@nocobase/data-source-manager": "3.0.0-alpha.2",
+    "@nocobase/logger": "3.0.0-alpha.2",
+    "@nocobase/resourcer": "3.0.0-alpha.2",
+    "@nocobase/utils": "3.0.0-alpha.2",
     "d3-dsv": "2",
     "fast-glob": "^3.3.2",
     "flexsearch": "^0.8.2",
     "gray-matter": "^4.0.3",
     "langchain": "1.2.39",
     "mammoth": "^1.10.0",
-    "officeparser": "^5.2.0",
+    "officeparser": "^6.0.4",
     "pdf-parse": "^1.1.1",
     "word-extractor": "^1.0.4",
     "xlsx": "https://cdn.sheetjs.com/xlsx-0.20.3/xlsx-0.20.3.tgz"
@@ -37,5 +37,5 @@
     "url": "git+https://github.com/nocobase/nocobase.git",
     "directory": "packages/ai"
   },
-  "gitHead": "2377df8ceb12549149017f7f14a61207bf6e49a2"
+  "gitHead": "a3faea21082b30989301de04591ee1b91fd4e6a5"
 }

package/src/__tests__/document-loader.test.ts

@@ -60,4 +60,15 @@ describe('Document loader worker', () => {
     expect(documents[0].pageContent).toBe('Sheet: Sheet1\nname\tvalue\nalpha\t1');
     expect(documents[0].metadata.source).toBe(filePath);
   });
+
+  it('does not parse legacy ppt files', async () => {
+    const filePath = await createTempFile('source.ppt', 'legacy powerpoint content');
+
+    const documents = await loadByWorker('.ppt', {
+      filePath,
+      mimeType: 'application/vnd.ms-powerpoint',
+    });
+
+    expect(documents).toEqual([]);
+  });
 });

package/src/document-loader/index.ts

@@ -14,14 +14,19 @@ import path from 'node:path';
 export type DocumentLoaderWorkerOptions = {
   filePath: string;
   mimeType?: string;
+  /** Timeout in milliseconds for the worker to complete. Defaults to 5 minutes. */
+  timeout?: number;
 };
 
+const DEFAULT_WORKER_TIMEOUT = 5 * 60 * 1000;
+
 export const loadByWorker = async (extname: string, options: DocumentLoaderWorkerOptions): Promise<Document[]> => {
   const isTsRuntime = __filename.endsWith('.ts');
   const workerPath = path.join(__dirname, `loader.worker.${isTsRuntime ? 'ts' : 'js'}`);
   const worker = new Worker(workerPath, {
     execArgv: isTsRuntime ? ['--require', 'tsx/cjs'] : undefined,
   });
+  const timeout = options.timeout ?? DEFAULT_WORKER_TIMEOUT;
   return new Promise<Document[]>((resolve, reject) => {
     let settled = false;
     const close = (error?: Error, result?: Document[]) => {
@@ -29,6 +34,7 @@ export const loadByWorker = async (extname: string, options: DocumentLoaderWorke
         return;
       }
       settled = true;
+      clearTimeout(timer);
       if (error) {
         reject(error);
         return;
@@ -36,6 +42,10 @@ export const loadByWorker = async (extname: string, options: DocumentLoaderWorke
       resolve(result || []);
     };
 
+    const timer = setTimeout(() => {
+      close(new Error(`Document loading timed out after ${Math.round(timeout / 1000)}s`));
+    }, timeout);
+
     worker.once('message', (payload: { documents?: Document[]; error?: string }) => {
       if (payload?.error) {
         close(new Error(payload.error));

package/src/document-loader/loader.worker.ts

@@ -29,7 +29,15 @@ type WorkerResponse = {
 
 const loadPdf = async (filePath: string): Promise<Document[]> => {
   const loader = new PDFLoader(filePath);
-  return loader.load();
+  try {
+    return await loader.load();
+  } catch (error) {
+    const err = error as Error;
+    if (err?.name === 'PasswordException' || /password/i.test(err?.message)) {
+      throw new Error('The PDF file is password-protected and cannot be parsed. Please upload an unlocked version.');
+    }
+    throw error;
+  }
 };
 
 const loadDoc = async (filePath: string, type: 'docx' | 'doc'): Promise<Document[]> => {
@@ -37,7 +45,7 @@ const loadDoc = async (filePath: string, type: 'docx' | 'doc'): Promise<Document
   return loader.load();
 };
 
-const loadPpt = async (filePath: string): Promise<Document[]> => {
+const loadPptx = async (filePath: string): Promise<Document[]> => {
   const loader = new PPTXLoader(filePath);
   return loader.load();
 };
@@ -56,9 +64,8 @@ const loadByExtname = async (payload: ParsePayload): Promise<Document[]> => {
   switch (payload.extname) {
     case '.pdf':
       return loadPdf(payload.filePath);
-    case '.ppt':
     case '.pptx':
-      return loadPpt(payload.filePath);
+      return loadPptx(payload.filePath);
     case '.doc':
       return loadDoc(payload.filePath, 'doc');
     case '.docx':
@@ -67,6 +74,7 @@ const loadByExtname = async (payload: ParsePayload): Promise<Document[]> => {
       return loadCsv(payload.filePath);
     case '.xls':
     case '.xlsx':
+    case '.xlsm':
       return loadXlsx(payload.filePath, payload.mimeType);
     case '.json':
     case '.md':
@@ -89,8 +97,9 @@ parentPort?.on('message', async (payload: ParsePayload) => {
     };
     parentPort?.postMessage(response);
   } catch (error) {
+    const err = error as Error;
     const response: WorkerResponse = {
-      error: String(error?.stack || error),
+      error: err?.message || String(error),
     };
     parentPort?.postMessage(response);
   }

package/src/tools-manager/index.ts

@@ -106,6 +106,7 @@ export function defineTools(options: ToolsOptions) {
 }
 
 export const SYSTEM_TOOLS = {
+  GET_SKILL: 'getSkill',
   WEB_SEARCH: 'subAgentWebSearch',
   KNOWLEDGE_BASE: 'knowledge-base-retrieve',
   WORK_FLOW_TASK_OUTPUT: 'aiEmployeeWorkflowTaskOutput',