Upload 67 files

utils/documentParser.ts

@@ -2,10 +2,17 @@
 // @ts-ignore
 import mammoth from 'mammoth';
 // @ts-ignore
-import * as pdfjsLib from 'pdfjs-dist';
+import * as pdfjsProxy from 'pdfjs-dist';
 
-// Set worker for PDF.js. Using the same version from ESM CDN.
-pdfjsLib.GlobalWorkerOptions.workerSrc = 'https://esm.sh/pdfjs-dist@3.11.174/build/pdf.worker.min.js';
+// Handle potential default export (ESM vs CommonJS interop issues)
+const pdfjsLib = (pdfjsProxy as any).default || pdfjsProxy;
+
+// Set worker for PDF.js safely. Using the same version from ESM CDN.
+if (pdfjsLib && pdfjsLib.GlobalWorkerOptions) {
+    pdfjsLib.GlobalWorkerOptions.workerSrc = 'https://esm.sh/pdfjs-dist@3.11.174/build/pdf.worker.min.js';
+} else {
+    console.warn("PDF.js GlobalWorkerOptions not found. PDF parsing might fail.");
+}
 
 export interface ParsedDocument {
     fileName: string;
@@ -56,6 +63,10 @@ const parsePdf = async (file: File): Promise<string> => {
         reader.onload = async (e) => {
             try {
                 const typedarray = new Uint8Array(e.target?.result as ArrayBuffer);
+                // Ensure we call getDocument on the resolved library object
+                if (!pdfjsLib || !pdfjsLib.getDocument) {
+                    throw new Error('PDF.js library not loaded correctly');
+                }
                 const pdf = await pdfjsLib.getDocument(typedarray).promise;
                 let fullText = '';