Update app.py

app.py

@@ -46,7 +46,7 @@ def format_description(description):
     """
     # Extract parts of the description based on the expected structure
     line1_match = re.search(r"Stainless Steel RATING AND DIAGRAM PLATE", description)
-    line2_match = re.search(r"As per Drg\.No\..*?\d+", description)
+    line2_match = re.search(r"As per Drg\.No\..*?[A-Z0-9]+\s", description)
     line3_match = re.search(r"SIZE\s*:\s*\d+mm\s*X\s*\d+mm\s*X\s*[\d.]+mm\s*Thick", description)
     line4_match = re.search(r"With Serial No:.*", description)
 
@@ -55,11 +55,11 @@ def format_description(description):
     if line1_match:
         formatted_description.append(line1_match.group())
     if line2_match:
-        formatted_description.append(line2_match.group())
+        formatted_description.append(line2_match.group().strip())
     if line3_match:
-        formatted_description.append(line3_match.group())
+        formatted_description.append(line3_match.group().strip())
     if line4_match:
-        formatted_description.append(line4_match.group())
+        formatted_description.append(line4_match.group().strip())
 
     # Join the lines with a newline character
     return "\n".join(formatted_description)
@@ -151,6 +151,7 @@ def parse_po_items_with_filters(text):
     return df, "Data extracted successfully."
 
 
+
 # Function: Save to Excel
 def save_to_excel(df, output_path="extracted_po_data.xlsx"):
     df.to_excel(output_path, index=False)