@nocobase/ai 2.2.0-beta.9 → 2.2.0

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

@@ -7,4 +7,10 @@
  * For more information, please refer to: https://www.nocobase.com/agreement.
  */
 import { Document } from '@langchain/core/documents';
-export declare const loadByWorker: (extname: string, blob: Blob) => Promise<Document[]>;
+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

@@ -42,13 +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 loadByWorker = /* @__PURE__ */ __name(async (extname, blob) => {
-  const buffer = Buffer.from(await blob.arrayBuffer());
+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) => {
@@ -56,12 +57,16 @@ const loadByWorker = /* @__PURE__ */ __name(async (extname, blob) => {
         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));
@@ -77,8 +82,8 @@ const loadByWorker = /* @__PURE__ */ __name(async (extname, blob) => {
     });
     worker.postMessage({
       extname,
-      mimeType: blob.type,
-      buffer: Uint8Array.from(buffer)
+      filePath: options.filePath,
+      mimeType: options.mimeType
     });
   }).finally(() => {
     worker.terminate().catch(() => void 0);

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

@@ -17,47 +17,54 @@ var import_text = require("./vendor/langchain/document_loaders/fs/text");
 var import_csv = require("@langchain/community/document_loaders/fs/csv");
 var import_xlsx = require("./xlsx");
 var _a;
-const loadPdf = /* @__PURE__ */ __name(async (blob) => {
-  const loader = new import_pdf.PDFLoader(blob);
-  return loader.load();
+const loadPdf = /* @__PURE__ */ __name(async (filePath) => {
+  const loader = new import_pdf.PDFLoader(filePath);
+  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 (blob, type) => {
-  const loader = new import_docx.DocxLoader(blob, { type });
+const loadDoc = /* @__PURE__ */ __name(async (filePath, type) => {
+  const loader = new import_docx.DocxLoader(filePath, { type });
   return loader.load();
 }, "loadDoc");
-const loadPpt = /* @__PURE__ */ __name(async (blob) => {
-  const loader = new import_pptx.PPTXLoader(blob);
+const loadPptx = /* @__PURE__ */ __name(async (filePath) => {
+  const loader = new import_pptx.PPTXLoader(filePath);
   return loader.load();
-}, "loadPpt");
-const loadTxt = /* @__PURE__ */ __name(async (blob) => {
-  const loader = new import_text.TextLoader(blob);
+}, "loadPptx");
+const loadTxt = /* @__PURE__ */ __name(async (filePath) => {
+  const loader = new import_text.TextLoader(filePath);
   return loader.load();
 }, "loadTxt");
-const loadCsv = /* @__PURE__ */ __name(async (blob) => {
-  const loader = new import_csv.CSVLoader(blob);
+const loadCsv = /* @__PURE__ */ __name(async (filePath) => {
+  const loader = new import_csv.CSVLoader(filePath);
   return loader.load();
 }, "loadCsv");
 const loadByExtname = /* @__PURE__ */ __name(async (payload) => {
-  const blob = new Blob([Buffer.from(payload.buffer)], { type: payload.mimeType ?? "application/octet-stream" });
   switch (payload.extname) {
     case ".pdf":
-      return loadPdf(blob);
-    case ".ppt":
+      return loadPdf(payload.filePath);
     case ".pptx":
-      return loadPpt(blob);
+      return loadPptx(payload.filePath);
     case ".doc":
-      return loadDoc(blob, "doc");
+      return loadDoc(payload.filePath, "doc");
     case ".docx":
-      return loadDoc(blob, "docx");
+      return loadDoc(payload.filePath, "docx");
     case ".csv":
-      return loadCsv(blob);
+      return loadCsv(payload.filePath);
     case ".xls":
     case ".xlsx":
-      return (0, import_xlsx.loadXlsx)(blob);
+    case ".xlsm":
+      return (0, import_xlsx.loadXlsx)(payload.filePath, payload.mimeType);
     case ".json":
     case ".md":
     case ".txt":
-      return loadTxt(blob);
+      return loadTxt(payload.filePath);
     default:
       return [];
   }
@@ -75,8 +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/document-loader/xlsx.d.ts

@@ -7,4 +7,4 @@
  * For more information, please refer to: https://www.nocobase.com/agreement.
  */
 import { Document } from '@langchain/core/documents';
-export declare const loadXlsx: (blob: Blob) => Promise<Document[]>;
+export declare const loadXlsx: (filePath: string, mimeType?: string) => Promise<Document[]>;

package/lib/document-loader/xlsx.js

@@ -41,6 +41,7 @@ __export(xlsx_exports, {
 });
 module.exports = __toCommonJS(xlsx_exports);
 var import_documents = require("@langchain/core/documents");
+var import_promises = require("node:fs/promises");
 var XLSX = __toESM(require("xlsx"));
 const normalizeCellValue = /* @__PURE__ */ __name((value) => {
   if (value === void 0 || value === null) {
@@ -64,10 +65,10 @@ const sheetToLines = /* @__PURE__ */ __name((sheet) => {
   });
   return rows.map((row) => trimTrailingEmptyCells(Array.isArray(row) ? row : [])).filter((row) => row.length > 0).map((row) => row.map((cell) => normalizeCellValue(cell)).join("	")).filter((line) => line.trim().length > 0);
 }, "sheetToLines");
-const loadXlsx = /* @__PURE__ */ __name(async (blob) => {
-  const buffer = await blob.arrayBuffer();
+const loadXlsx = /* @__PURE__ */ __name(async (filePath, mimeType) => {
+  const buffer = await (0, import_promises.readFile)(filePath);
   const workbook = XLSX.read(buffer, {
-    type: "array",
+    type: "buffer",
     cellText: true
   });
   const documents = [];
@@ -84,8 +85,8 @@ const loadXlsx = /* @__PURE__ */ __name(async (blob) => {
       new import_documents.Document({
         pageContent: [`Sheet: ${sheetName}`, ...lines].join("\n"),
         metadata: {
-          source: "blob",
-          blobType: blob.type,
+          source: filePath,
+          blobType: mimeType,
           sheetName,
           sheetIndex: index
         }

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": "2.2.0-beta.9",
+  "version": "2.2.0",
   "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": "2.2.0-beta.9",
-    "@nocobase/logger": "2.2.0-beta.9",
-    "@nocobase/resourcer": "2.2.0-beta.9",
-    "@nocobase/utils": "2.2.0-beta.9",
+    "@nocobase/data-source-manager": "2.2.0",
+    "@nocobase/logger": "2.2.0",
+    "@nocobase/resourcer": "2.2.0",
+    "@nocobase/utils": "2.2.0",
     "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": "60e3d7abbaa0c7cead76f71a4f3d5eedb6b8acdb"
+  "gitHead": "d5052cae37177054c236f1bf7bba3aed6ef029d0"
 }

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

@@ -0,0 +1,74 @@
+/**
+ * This file is part of the NocoBase (R) project.
+ * Copyright (c) 2020-2024 NocoBase Co., Ltd.
+ * Authors: NocoBase Team.
+ *
+ * This project is dual-licensed under AGPL-3.0 and NocoBase Commercial License.
+ * For more information, please refer to: https://www.nocobase.com/agreement.
+ */
+
+import { loadByWorker } from '../document-loader';
+import { mkdtemp, rm, writeFile } from 'node:fs/promises';
+import os from 'node:os';
+import path from 'node:path';
+import * as XLSX from 'xlsx';
+
+describe('Document loader worker', () => {
+  const tempDirs: string[] = [];
+
+  const createTempFile = async (filename: string, content: string | Buffer) => {
+    const tempDir = await mkdtemp(path.join(os.tmpdir(), 'nocobase-document-loader-test-'));
+    tempDirs.push(tempDir);
+    const filePath = path.join(tempDir, filename);
+    await writeFile(filePath, content);
+    return filePath;
+  };
+
+  afterEach(async () => {
+    await Promise.all(tempDirs.splice(0).map((tempDir) => rm(tempDir, { recursive: true, force: true })));
+  });
+
+  it('loads text files by path', async () => {
+    const filePath = await createTempFile('source.txt', 'hello knowledge base\nsecond line\n');
+
+    const documents = await loadByWorker('.txt', {
+      filePath,
+      mimeType: 'text/plain',
+    });
+
+    expect(documents).toHaveLength(1);
+    expect(documents[0].pageContent).toBe('hello knowledge base\nsecond line\n');
+    expect(documents[0].metadata.source).toBe(filePath);
+  });
+
+  it('loads xlsx files by path', async () => {
+    const worksheet = XLSX.utils.aoa_to_sheet([
+      ['name', 'value'],
+      ['alpha', 1],
+    ]);
+    const workbook = XLSX.utils.book_new();
+    XLSX.utils.book_append_sheet(workbook, worksheet, 'Sheet1');
+    const buffer = XLSX.write(workbook, { type: 'buffer', bookType: 'xlsx' });
+    const filePath = await createTempFile('source.xlsx', buffer);
+
+    const documents = await loadByWorker('.xlsx', {
+      filePath,
+      mimeType: 'application/vnd.openxmlformats-officedocument.spreadsheetml.sheet',
+    });
+
+    expect(documents).toHaveLength(1);
+    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

@@ -11,13 +11,22 @@ import { Document } from '@langchain/core/documents';
 import { Worker } from 'node:worker_threads';
 import path from 'node:path';
 
-export const loadByWorker = async (extname: string, blob: Blob): Promise<Document[]> => {
-  const buffer = Buffer.from(await blob.arrayBuffer());
+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[]) => {
@@ -25,6 +34,7 @@ export const loadByWorker = async (extname: string, blob: Blob): Promise<Documen
         return;
       }
       settled = true;
+      clearTimeout(timer);
       if (error) {
         reject(error);
         return;
@@ -32,6 +42,10 @@ export const loadByWorker = async (extname: string, blob: Blob): Promise<Documen
       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));
@@ -48,8 +62,8 @@ export const loadByWorker = async (extname: string, blob: Blob): Promise<Documen
 
     worker.postMessage({
       extname,
-      mimeType: blob.type,
-      buffer: Uint8Array.from(buffer),
+      filePath: options.filePath,
+      mimeType: options.mimeType,
     });
   }).finally(() => {
     worker.terminate().catch(() => undefined);

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

@@ -19,7 +19,7 @@ import { loadXlsx } from './xlsx';
 type ParsePayload = {
   extname: string;
   mimeType?: string;
-  buffer: Uint8Array;
+  filePath: string;
 };
 
 type WorkerResponse = {
@@ -27,54 +27,59 @@ type WorkerResponse = {
   error?: string;
 };
 
-const loadPdf = async (blob: Blob): Promise<Document[]> => {
-  const loader = new PDFLoader(blob);
-  return loader.load();
+const loadPdf = async (filePath: string): Promise<Document[]> => {
+  const loader = new PDFLoader(filePath);
+  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 (blob: Blob, type: 'docx' | 'doc'): Promise<Document[]> => {
-  const loader = new DocxLoader(blob, { type });
+const loadDoc = async (filePath: string, type: 'docx' | 'doc'): Promise<Document[]> => {
+  const loader = new DocxLoader(filePath, { type });
   return loader.load();
 };
 
-const loadPpt = async (blob: Blob): Promise<Document[]> => {
-  const loader = new PPTXLoader(blob);
+const loadPptx = async (filePath: string): Promise<Document[]> => {
+  const loader = new PPTXLoader(filePath);
   return loader.load();
 };
 
-const loadTxt = async (blob: Blob): Promise<Document[]> => {
-  const loader = new TextLoader(blob);
+const loadTxt = async (filePath: string): Promise<Document[]> => {
+  const loader = new TextLoader(filePath);
   return loader.load();
 };
 
-const loadCsv = async (blob: Blob): Promise<Document[]> => {
-  const loader = new CSVLoader(blob);
+const loadCsv = async (filePath: string): Promise<Document[]> => {
+  const loader = new CSVLoader(filePath);
   return loader.load();
 };
 
 const loadByExtname = async (payload: ParsePayload): Promise<Document[]> => {
-  // @ts-ignore
-  const blob = new Blob([Buffer.from(payload.buffer)], { type: payload.mimeType ?? 'application/octet-stream' });
-
   switch (payload.extname) {
     case '.pdf':
-      return loadPdf(blob);
-    case '.ppt':
+      return loadPdf(payload.filePath);
     case '.pptx':
-      return loadPpt(blob);
+      return loadPptx(payload.filePath);
     case '.doc':
-      return loadDoc(blob, 'doc');
+      return loadDoc(payload.filePath, 'doc');
     case '.docx':
-      return loadDoc(blob, 'docx');
+      return loadDoc(payload.filePath, 'docx');
     case '.csv':
-      return loadCsv(blob);
+      return loadCsv(payload.filePath);
     case '.xls':
     case '.xlsx':
-      return loadXlsx(blob);
+    case '.xlsm':
+      return loadXlsx(payload.filePath, payload.mimeType);
     case '.json':
     case '.md':
     case '.txt':
-      return loadTxt(blob);
+      return loadTxt(payload.filePath);
     default:
       return [];
   }
@@ -92,8 +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/document-loader/xlsx.ts

@@ -8,6 +8,7 @@
  */
 
 import { Document } from '@langchain/core/documents';
+import { readFile } from 'node:fs/promises';
 import * as XLSX from 'xlsx';
 
 const normalizeCellValue = (value: unknown): string => {
@@ -43,10 +44,10 @@ const sheetToLines = (sheet: XLSX.WorkSheet): string[] => {
     .filter((line) => line.trim().length > 0);
 };
 
-export const loadXlsx = async (blob: Blob): Promise<Document[]> => {
-  const buffer = await blob.arrayBuffer();
+export const loadXlsx = async (filePath: string, mimeType?: string): Promise<Document[]> => {
+  const buffer = await readFile(filePath);
   const workbook = XLSX.read(buffer, {
-    type: 'array',
+    type: 'buffer',
     cellText: true,
   });
 
@@ -69,8 +70,8 @@ export const loadXlsx = async (blob: Blob): Promise<Document[]> => {
       new Document({
         pageContent: [`Sheet: ${sheetName}`, ...lines].join('\n'),
         metadata: {
-          source: 'blob',
-          blobType: blob.type,
+          source: filePath,
+          blobType: mimeType,
           sheetName,
           sheetIndex: index,
         },

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',