Update app.py

app.py

@@ -1,5 +1,6 @@
 import re
 import pandas as pd
+import gradio as gr
 
 
 def extract_po_data(text):
@@ -14,9 +15,9 @@ def extract_po_data(text):
     data = []
 
     for line in lines:
-        # Match table row patterns
+        # Match table row patterns with a flexible regex
         row_match = re.match(
-            r"^(?P<Item>\d+)\s+(?P<Description>.+?)\s+(?P<Qty>\d+)\s+(?P<Unit>(Nos\.|Set))\s+(?P<UnitPrice>[\d.]+)\s+(?P<TotalPrice>[\d.]+)$",
+            r"^\s*(?P<Item>\d+)\s+(?P<Description>.+?)\s+(?P<Qty>\d+)\s+(?P<Unit>(Nos\.|Set))\s+(?P<UnitPrice>[\d.]+)\s+(?P<TotalPrice>[\d.]+)\s*$",
             line,
         )
         if row_match:
@@ -78,18 +79,53 @@ def format_description(description):
     return "\n".join(lines)
 
 
-# Example Usage
-if __name__ == "__main__":
-    # Example raw text (replace this with actual extracted text from PDF)
-    raw_text = """
-    1 Stainless Steel RATING AND DIAGRAM PLATE As per Drg.No. G 000822 RI RDP 50KVA NT00l 51 SIZE : l50mm X 160mm X 1.00mm Thick With Serial No:NT00151 97 to 121 Mfd:-2022 24 Nos. 3.00 72.00
+def save_to_excel(df, output_path="extracted_po_data.xlsx"):
+    """
+    Saves the extracted data to an Excel file.
+    Args:
+        df (pd.DataFrame): DataFrame containing the structured data.
+        output_path (str): Path to save the Excel file.
+    Returns:
+        str: Path to the saved file.
     """
+    df.to_excel(output_path, index=False)
+    return output_path
 
-    # Extract data
-    df, status = extract_po_data(raw_text)
 
-    # Output results
+def process_text_input(text):
+    """
+    Processes the raw text input, extracts data, and saves it to an Excel file.
+    Args:
+        text (str): Raw text input.
+    Returns:
+        str: Path to the saved Excel file.
+    """
+    df, status = extract_po_data(text)
     if df is not None:
-        print(df)
-    else:
-        print(status)
+        output_path = save_to_excel(df)
+        return output_path, status
+    return None, status
+
+
+# Gradio Interface
+def create_interface():
+    """
+    Creates a Gradio interface for processing PO data.
+    """
+    interface = gr.Interface(
+        fn=process_text_input,
+        inputs=gr.Textbox(label="Paste Raw Text from PDF", lines=10, placeholder="Paste extracted text here..."),
+        outputs=[
+            gr.File(label="Download Extracted Excel"),
+            gr.Textbox(label="Status"),
+        ],
+        title="PO Data Extraction",
+        description="Paste the raw text from the PDF to extract purchase order data into an Excel file.",
+    )
+    return interface
+
+
+if __name__ == "__main__":
+    # Run the Gradio app
+    app = create_interface()
+    app.launch()