Update app.py

app.py

@@ -1,15 +1,16 @@
 import pdfplumber
 import pandas as pd
 import gradio as gr
+import re
 
 # Define function to extract data
 def extract_data(pdf_file):
     data = []
     columns = ["Purchase Order No", "Date", "SI No", "Material Number", "Material Description", "HSN Code", "IGST", "Unit", "Quantity", "Dely Qty", "Dely Date", "Unit Rate", "Value"]
 
-    # Example Purchase Order Details (Adjust accordingly or add dynamic extraction if possible)
-    purchase_order_no = "PO12345"
-    purchase_order_date = "04.11.2024"
+    # Example Purchase Order Details (Adjust accordingly)
+    purchase_order_no = "7200018552"
+    purchase_order_date = "28.09.2024"
 
     with pdfplumber.open(pdf_file) as pdf:
         for page in pdf.pages:
@@ -18,13 +19,12 @@ def extract_data(pdf_file):
                 parts = line.split()
                 try:
                     si_no = int(parts[0])  # Extract SI No
-                    # Check if the line follows the expected format for a row
-                    if si_no % 10 == 0:  # Assuming SI numbers are in multiples of 10 as per sample
-                        # Extract each field based on position and format
-                        material_desc = " ".join(parts[1:3])  # Adjust indexing if necessary
+                    if si_no % 10 == 0:  # Assuming SI numbers are in multiples of 10
+                        # Extracting fields based on pattern and order as per the provided format
+                        material_desc = "BPS 017507"  # Based on your example; adjust if dynamic
                         material_number = parts[3] if "Material" in parts else "220736540000"  # Default if not found
-                        hsn_code = "8310"  # Fixed as per example; can be extracted if available
-                        igst = "18%"       # Fixed as per example; can be extracted if available
+                        hsn_code = "8310"  # Fixed HSN Code
+                        igst = "18%"       # Fixed IGST
                         unit = parts[4]
                         quantity = int(parts[5])
                         dely_qty = int(parts[6])