@nocobase/ai 2.1.30 → 2.1.32

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

@@ -19,7 +19,15 @@ 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 });
@@ -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/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@nocobase/ai",
-  "version": "2.1.30",
+  "version": "2.1.32",
   "description": "",
   "license": "Apache-2.0",
   "main": "./lib/index.js",
@@ -17,10 +17,10 @@
     "@langchain/mcp-adapters": "1.1.3",
     "@langchain/ollama": "1.2.7",
     "@langchain/openai": "1.4.7",
-    "@nocobase/data-source-manager": "2.1.30",
-    "@nocobase/logger": "2.1.30",
-    "@nocobase/resourcer": "2.1.30",
-    "@nocobase/utils": "2.1.30",
+    "@nocobase/data-source-manager": "2.1.32",
+    "@nocobase/logger": "2.1.32",
+    "@nocobase/resourcer": "2.1.32",
+    "@nocobase/utils": "2.1.32",
     "d3-dsv": "2",
     "fast-glob": "^3.3.2",
     "flexsearch": "^0.8.2",
@@ -37,5 +37,5 @@
     "url": "git+https://github.com/nocobase/nocobase.git",
     "directory": "packages/ai"
   },
-  "gitHead": "c1a4bff0fd3c8b30e06fc6f50ad33e054b04061a"
+  "gitHead": "737a1324f650d7bc66594203b68eb0180553a551"
 }

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[]> => {
@@ -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);
   }