Update app.py

app.py

@@ -116,13 +116,20 @@ def extract_numeric_values(pdf_file, schedule1_file=None, client_name="Unknown C
                     raw = pytesseract.image_to_string(val_img, config=config_val).strip()
                     value_text = re.sub(r"[^\d,.\-+]", "", raw)
                     schedule1_values.append(value_text)
+                    if draw_bboxes:
+                        page.draw_rect(rect, color=(0, 1, 0), width=1.5)  # green boxes
+                schedule1_pdf_path = None
+                if draw_bboxes and schedule1_file:
+                    schedule1_pdf_path = "schedule1_bbox.pdf"
+                    doc.save(schedule1_pdf_path)
+        
                 output += [f"Schedule 1 Line {i*2+1 if i < 2 else 8}: {val}" for i, val in enumerate(schedule1_values)]
-            doc.close()
+                doc.close()
 
         output_dir = "./Clients_Output_Data_Form_1040.csv"
         save_to_csv_flat(all_extracted_values, schedule1_values, client_name=client_name, csv_path=output_dir)
 
-        return "\n".join(output), output_dir, output_pdf_path
+        return "\n".join(output), output_dir, output_pdf_path, schedule1_pdf_path
 
     except Exception as e:
         return f"Error occurred:\n{str(e)}", None, None
@@ -214,13 +221,15 @@ iface = gr.Interface(
         gr.File(label="Upload Main Form 1040 PDF (Required)", file_types=[".pdf"]),
         gr.File(label="Upload Schedule 1 PDF (Optional)", file_types=[".pdf"]),
         gr.Textbox(label="Client Name", placeholder="Enter client name"),
-        gr.Checkbox(label="Draw Bounding Boxes on Form 1040?")
+        gr.Checkbox(label="Draw Bounding Boxes on Form 1040 and Schedule 1?")
     ],
     outputs=[
-        gr.Textbox(label="Extracted Numeric Values", lines=20),
-        gr.File(label="Download Excel Output"),
-        gr.File(label="Download PDF with Bounding Boxes")
+    gr.Textbox(label="Extracted Numeric Values", lines=20),
+    gr.File(label="Download Excel Output"),
+    gr.File(label="1040 Bounding Boxes PDF"),
+    gr.File(label="Schedule 1 Bounding Boxes PDF")
     ],
+
     title="Form 1040 & Schedule 1 Extractor",
     description="Upload a 2-page 1040 Form and (optionally) Schedule 1 to extract numeric fields. Optionally draw bounding boxes."
 )