Merge branch 'main' into next

packages/core/ai/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));

packages/core/ai/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[]> => {
@@ -90,8 +98,9 @@ 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);
   }