Update app.py

app.py

@@ -12,49 +12,43 @@ def extract_data_from_pdf(pdf_file):
         for page in pdf.pages:
             text = page.extract_text()
 
-            # Extract PO number
+            # Extract PO number (only once at the start)
             if po_number is None:
                 po_match = re.search(r"Purchase Order : (\w+)", text)
                 po_number = po_match.group(1) if po_match else "N/A"
 
-            # Regex pattern for row data
+            # Regex pattern for extracting rows
             row_pattern = re.compile(
-                r"(\d+)\s+(\d+)\s+(\w+)\s+(\d{4}-\d{2}-\d{2})\s+([\d.]+)\s+([\d.]+)\s+([\d.]+)"
+                r"(\d+)\s+(\d{9})\s+(\w+)\s+(\d{4}-\d{2}-\d{2})\s+([\d.]+)\s+([\d.]+)\s+([\d.]+)\s+INR\s+([\d.]+)"
             )
 
-            # Extract matching rows
+            # Extract each row using the pattern
             for match in row_pattern.finditer(text):
-                (
-                    pos,
-                    item_code,
-                    unit,
-                    delivery_date,
-                    quantity,
-                    basic_price,
-                    amount,
-                ) = match.groups()
+                pos, item_code, unit, delivery_date, quantity, basic_price, discount, amount = match.groups()
 
+                # Extract subtotal if present
                 sub_total_match = re.search(r"SUB TOTAL : ([\d.]+)", text)
                 sub_total = sub_total_match.group(1) if sub_total_match else "0.0"
 
-                data.append(
-                    {
-                        "Purchase Order": po_number,
-                        "Pos.": pos,
-                        "Item Code": item_code,
-                        "Unit": unit,
-                        "Delivery Date": delivery_date,
-                        "Quantity": quantity,
-                        "Basic Price": basic_price,
-                        "Amount": amount,
-                        "SUB TOTAL": sub_total,
-                    }
-                )
+                # Append data for each matched row
+                data.append({
+                    "Purchase Order": po_number,
+                    "Pos.": pos,
+                    "Item Code": item_code,
+                    "Unit": unit,
+                    "Delivery Date": delivery_date,
+                    "Quantity": quantity,
+                    "Basic Price": basic_price,
+                    "Discount": discount,
+                    "Amount": amount,
+                    "SUB TOTAL": sub_total,
+                })
 
     # Convert data to DataFrame and save to Excel
     df = pd.DataFrame(data)
-    output = df.to_excel("output.xlsx", index=False)
-    return "output.xlsx"
+    output_file = "output.xlsx"
+    df.to_excel(output_file, index=False)
+    return output_file
 
 # Gradio Interface
 iface = gr.Interface(