Code Editor for app.py

import streamlit as st
from PIL import Image
import pytesseract
import pandas as pd
import io
import re
from datetime import datetime

st.title("Receipt Scanner & Processor")

# Initialize session state for storing receipts
if "receipts" not in st.session_state:
    st.session_state.receipts = []

def extract_receipt_info(text):
    """Extract key information from receipt text"""
    # Initialize dictionary to store receipt information
    receipt_info = {
        'date': None,
        'total': None,
        'store': None,
        'items': []
    }
    
    lines = text.split('\n')
    
    # Try to find date (assuming common date formats)
    date_pattern = r'\d{1,2}[-/]\d{1,2}[-/]\d{2,4}'
    for line in lines:
        date_match = re.search(date_pattern, line)
        if date_match:
            try:
                receipt_info['date'] = datetime.strptime(date_match.group(), '%m/%d/%Y').strftime('%Y-%m-%d')
                break
            except:
                continue
    
    # Try to find total (looking for lines with "TOTAL" or similar)
    total_pattern = r'(?i)total[:\s]*[$]?\s*(\d+\.?\d*)'
    for line in lines:
        total_match = re.search(total_pattern, line)
        if total_match:
            receipt_info['total'] = float(total_match.group(1))
            break
    
    # Try to find store name (usually in first few lines)
    if len(lines) > 0:
        receipt_info['store'] = lines[0].strip()
    
    return receipt_info

# File uploader
uploaded_file = st.file_uploader("Upload receipt image", type=['png', 'jpg', 'jpeg'])

if uploaded_file is not None:
    # Display the uploaded image
    image = Image.open(uploaded_file)
    st.image(image, caption='Uploaded Receipt', use_column_width=True)
    
    # Add a process button
    if st.button('Process Receipt'):
        with st.spinner('Processing receipt...'):
            # Convert image to text using OCR
            text = pytesseract.image_to_string(image)
            
            # Extract information
            receipt_info = extract_receipt_info(text)
            
            # Add timestamp
            receipt_info['processed_at'] = datetime.now().strftime('%Y-%m-%d %H:%M:%S')
            
            # Add to session state
            st.session_state.receipts.append(receipt_info)
            
            # Show extracted information
            st.subheader("Extracted Information")
            st.write("Store:", receipt_info['store'])
            st.write("Date:", receipt_info['date'])
            st.write("Total:", f"${receipt_info['total']:.2f}" if receipt_info['total'] else "Not found")
            
            # Show raw text
            with st.expander("Show Raw Text"):
                st.text(text)

# Display history of processed receipts
if st.session_state.receipts:
    st.subheader("Processing History")
    df = pd.DataFrame(st.session_state.receipts)
    st.table(df)
    
    # Export option
    if st.button('Export to CSV'):
        csv = df.to_csv(index=False)
        st.download_button(
            label="Download CSV",
            data=csv,
            file_name="receipts_export.csv",
            mime="text/csv"
        )