Update toshiba.py

toshiba.py

@@ -21,23 +21,24 @@ def extract_toshiba_data(pdf_file):
                     if date_match:
                         order_date = date_match.group(1)
 
-            # Extract item details
+            # Extract item details using patterns
             for line in text:
-                item_match = re.match(r'(\d+)\s+(\d+)\s+(NOS|PCS)\s+([\d-]+)\s+([\d.]+)\s+([\d.]+)\s+([\d.]+)', line)
+                # Match each line with expected pattern for item rows
+                item_match = re.match(r'(\d+)\s+(\d+)\s+(.*?)\s+([\d-]+)\s+([\d.]+)\s+([\d.]+)\s+([\d.]+)\s+([\d.]+)', line)
                 if item_match:
-                    pos = int(item_match.group(1))
-                    item_code = item_match.group(2)
-                    unit = item_match.group(3)
-                    delivery_date = item_match.group(4)
-                    quantity = float(item_match.group(5))
-                    basic_price = float(item_match.group(6))
-                    amount = quantity * basic_price
-                    sub_total = float(item_match.group(7))
+                    pos = int(item_match.group(1))                   # Position number
+                    item_code = item_match.group(2)                 # Item Code
+                    item_name = item_match.group(3).strip()         # Item Name/Description (if available)
+                    delivery_date = item_match.group(4)             # Delivery Date
+                    quantity = float(item_match.group(5))           # Quantity
+                    basic_price = float(item_match.group(6))        # Basic Price
+                    amount = float(item_match.group(7))             # Calculated Amount
+                    sub_total = float(item_match.group(8))          # Subtotal or final price
 
-                    data.append([purchase_order, order_date, pos, item_code, unit, delivery_date, quantity, basic_price, amount, sub_total])
+                    data.append([purchase_order, order_date, pos, item_code, item_name, delivery_date, quantity, basic_price, amount, sub_total])
 
     # Define DataFrame with the corrected structure
-    df = pd.DataFrame(data, columns=["Purchase Order", "Order Date", "Pos", "Item Code", "Unit", "Delivery Date", "Quantity", "Basic Price", "Amount", "SUB TOTAL"])
+    df = pd.DataFrame(data, columns=["Purchase Order", "Order Date", "Pos", "Item Code", "Item Name", "Delivery Date", "Quantity", "Basic Price", "Amount", "SUB TOTAL"])
     
     # Save to Excel file
     temp_file = tempfile.NamedTemporaryFile(delete=False, suffix=".xlsx")