Update app.py

app.py

@@ -35,6 +35,13 @@ def clean_description(description, item_number=None):
 
     return description.strip()
 
+# Function to clean item description
+def clean_description(description):
+    """
+    Cleans up the description for an item to ensure it's correctly formatted.
+    """
+    return description.strip()
+
 # Function to extract PO items with better error handling and checks
 def parse_po_items_with_filters(text):
     """
@@ -60,7 +67,7 @@ def parse_po_items_with_filters(text):
             # Save the previous item if current_item is not None
             if current_item is not None:
                 # Clean and add the description to the current item
-                current_item["Description"] = clean_description(" ".join(description_accumulator).strip(), current_item["Item"])
+                current_item["Description"] = clean_description(" ".join(description_accumulator).strip())
                 data.append(current_item)
                 description_accumulator = []  # Reset description accumulator
                 print(f"Item {current_item['Item']} added to data.")  # Debugging
@@ -100,7 +107,7 @@ def parse_po_items_with_filters(text):
     # Finalize the last item
     if current_item is not None:
         # Clean and add the description to the current item
-        current_item["Description"] = clean_description(" ".join(description_accumulator).strip(), current_item["Item"])
+        current_item["Description"] = clean_description(" ".join(description_accumulator).strip())
         data.append(current_item)
         print(f"Finalized Item {current_item['Item']}")  # Debugging
 
@@ -112,6 +119,7 @@ def parse_po_items_with_filters(text):
                 row["Description"]
             )
             if item_3_match:
+                # Split item 2 description and assign item 3
                 data.insert(
                     i + 1,
                     {
@@ -137,6 +145,20 @@ def parse_po_items_with_filters(text):
     df = pd.DataFrame(data)
     return df, "Data extracted successfully."
 
+# Example text (as provided)
+text = """ 
+ITEM 1 Stainless Steel RATING AND DIAGRAM PLATE 24 Nos. 3.00 72.00
+As per Drg.No. G 000822 RI RDP 50KVA NT001 51 SIZE : 150mm X 160mm X 1.00mm Thick With Serial No:NT00151 97 to 121 Mfd:-2022
+ITEM 2 Stainless Steel RATING AND DIAGRAM PLATE 12 Nos. 3.80 45.60
+As per Drg.to.G 000816 R2 RDP 600KVA NT00152 SIZE : 150mm X 260mm X 1.00mm Thick With Serial No:NT00I53 38 to 50 Mfd:-2022
+"""
+
+# Running the function
+df, status = parse_po_items_with_filters(text)
+print(status)
+if df is not None:
+    print(df)
+
 # Function: Save to Excel
 def save_to_excel(df, output_path="extracted_po_data.xlsx"):
     """