Update agents/document_agent.py

agents/document_agent.py

@@ -253,9 +253,12 @@ class DocumentAgent(BaseAgent):
         text = ""
         try:
             self.logger.logger.info("[DocumentAgent] Extracting text using PyMuPDF...")
-            with fitz.open(file_name) as doc:
-                for page in doc:
-                    text += page.get_text()
+            # with fitz.open(file_name) as doc:
+            #     for page in doc:
+            #         text += page.get_text()
+            with pdfplumber.open(file_name) as pdf:
+                for page in pdf.pages:
+                    text += page.extract_text()
             if len(text.strip()) < 5:
                 raise ValueError("PyMuPDF extraction too short, switching to PDFPlumber")
         except Exception as e: