streamlit_app.py

import streamlit as st
import os
import configparser
import glob
import shutil
from datetime import datetime
from pathlib import Path
import re
from utils import parse_winapp2_ini, scan_files, delete_files, get_file_size, format_size

st.set_page_config(
    page_title="Windows Data Cleaner",
    page_icon="🧹",
    layout="wide",
    initial_sidebar_state="expanded"
)

# Custom CSS for better UI
st.markdown("""
<style>
    .main-header {
        font-size: 2.5rem;
        font-weight: bold;
        color: #1f77b4;
        margin-bottom: 1rem;
    }
    .stat-box {
        background-color: #f0f2f6;
        padding: 1rem;
        border-radius: 0.5rem;
        margin: 0.5rem 0;
    }
    .warning-box {
        background-color: #fff3cd;
        border: 1px solid #ffc107;
        padding: 1rem;
        border-radius: 0.5rem;
        margin: 1rem 0;
    }
    .success-box {
        background-color: #d4edda;
        border: 1px solid #28a745;
        padding: 1rem;
        border-radius: 0.5rem;
        margin: 1rem 0;
    }
</style>
""", unsafe_allow_html=True)

# Header with anycoder link
st.markdown('<div class="main-header">🧹 Windows Data Cleaner</div>', unsafe_allow_html=True)
st.markdown('[Built with anycoder](https://huggingface.co/spaces/akhaliq/anycoder)', unsafe_allow_html=True)
st.markdown("---")

# Initialize session state
if 'cleaning_results' not in st.session_state:
    st.session_state.cleaning_results = None
if 'scanned_files' not in st.session_state:
    st.session_state.scanned_files = []
if 'ini_config' not in st.session_state:
    st.session_state.ini_config = None
if 'scan_complete' not in st.session_state:
    st.session_state.scan_complete = False

# Sidebar configuration
with st.sidebar:
    st.header("⚙️ Configuration")
    
    # INI File Upload
    st.subheader("WinApp2.ini File")
    ini_file = st.file_uploader(
        "Upload winapp2.ini",
        type=["ini", "txt"],
        help="Upload the winapp2.ini configuration file"
    )
    
    # Base directory for cleaning
    st.subheader("Base Directory")
    base_dir = st.text_input(
        "Base Directory Path",
        value=os.getenv('USERPROFILE', 'C:/Users') or 'C:/Users',
        help="Root directory to scan for files"
    )
    
    # Cleaning options
    st.subheader("Cleaning Options")
    dry_run = st.checkbox("Dry Run Mode", value=True, help="Preview files without deleting")
    confirm_delete = st.checkbox("Require Confirmation", value=True, help="Ask before deleting files")
    
    # Advanced options
    with st.expander("Advanced Options"):
        include_hidden = st.checkbox("Include Hidden Files", value=False)
        include_system = st.checkbox("Include System Files", value=False, help="Warning: May affect system stability")
        max_file_age_days = st.number_input("Max File Age (days)", min_value=0, value=365, help="Only clean files older than this")

# Main content area
st.header("📋 Configuration File")

if ini_file:
    try:
        # Parse the INI file
        with st.spinner("Parsing winapp2.ini..."):
            config_content = ini_file.getvalue().decode('utf-8')
            st.session_state.ini_config = parse_winapp2_ini(config_content)
        
        if st.session_state.ini_config:
            st.success("✅ Configuration file loaded successfully!")
            
            # Display configuration sections
            st.subheader("Available Cleaning Sections")
            for section_name, section_data in st.session_state.ini_config.items():
                with st.expander(f"📁 {section_name}", expanded=False):
                    st.code(f"Description: {section_data.get('description', 'No description')}")
                    st.code(f"Warning: {section_data.get('warning', 'No warnings')}")
                    
                    if 'path' in section_data:
                        st.code(f"Paths: {section_data['path']}")
                    
                    if 'file' in section_data:
                        st.code(f"File Patterns: {section_data['file']}")
                    
                    if 'regex' in section_data:
                        st.code(f"Regex Patterns: {section_data['regex']}")
    except Exception as e:
        st.error(f"❌ Error parsing INI file: {str(e)}")
else:
    st.info("📄 Please upload a winapp2.ini configuration file to continue.")
    st.code("""
# Example winapp2.ini structure
[SectionName]
Description=Description of what this cleans
Warning=Warning message
path=C:\\Users\\*\\AppData\\Local\\Temp
file=*.tmp; *.log
regex=pattern_to_match
""")

# Scan and Preview Section
st.header("🔍 Scan & Preview")

if st.session_state.ini_config and base_dir:
    if st.button("🔎 Scan for Files", type="primary", disabled=not st.session_state.ini_config):
        with st.spinner("Scanning directories..."):
            try:
                scanned_files = scan_files(
                    st.session_state.ini_config,
                    base_dir,
                    include_hidden=include_hidden,
                    include_system=include_system,
                    max_age_days=max_file_age_days
                )
                st.session_state.scanned_files = scanned_files
                st.session_state.scan_complete = True
                
                # Calculate totals
                total_files = len(scanned_files)
                total_size = sum(f['size'] for f in scanned_files)
                
                st.session_state.scan_summary = {
                    'total_files': total_files,
                    'total_size': total_size,
                    'sections': len(st.session_state.ini_config)
                }
                
                st.success(f"✅ Scan complete! Found {total_files} files ({format_size(total_size)})")
            except Exception as e:
                st.error(f"❌ Error during scan: {str(e)}")
                st.session_state.scan_complete = False

# Display scan results
if st.session_state.scan_complete and st.session_state.scanned_files:
    st.subheader("📊 Scan Results")
    
    # Summary statistics
    col1, col2, col3 = st.columns(3)
    with col1:
        st.metric(
            label="Total Files",
            value=st.session_state.scan_summary['total_files']
        )
    with col2:
        st.metric(
            label="Total Size",
            value=format_size(st.session_state.scan_summary['total_size'])
        )
    with col3:
        st.metric(
            label="Sections",
            value=st.session_state.scan_summary['sections']
        )
    
    # File list with selection
    st.subheader("📁 Files to Clean")
    
    # Create a table of files
    if st.session_state.scanned_files:
        # Display first 100 files in table
        display_files = st.session_state.scanned_files[:100]
        
        file_data = {
            'Section': [f['section'] for f in display_files],
            'Path': [f['path'] for f in display_files],
            'Size': [format_size(f['size']) for f in display_files],
            'Age (days)': [f['age_days'] for f in display_files]
        }
        
        st.table(file_data)
        
        if len(st.session_state.scanned_files) > 100:
            st.info(f"Showing first 100 of {len(st.session_state.scanned_files)} files")
    
    # Download full list
    if len(st.session_state.scanned_files) > 0:
        import json
        file_list_json = json.dumps(st.session_state.scanned_files, indent=2)
        st.download_button(
            label="📥 Download Full File List",
            data=file_list_json,
            file_name="scan_results.json",
            mime="application/json"
        )

# Cleaning Execution Section
st.header("🧹 Execute Cleaning")

if st.session_state.scan_complete and st.session_state.scanned_files:
    warning_msg = """
    <div class="warning-box">
    <strong>⚠️ WARNING:</strong> This operation will delete files from your system. 
    Please review the file list carefully before proceeding. 
    Make sure you have backups of important data.
    </div>
    """
    st.markdown(warning_msg, unsafe_allow_html=True)
    
    # Execution controls
    col1, col2 = st.columns([3, 1])
    
    with col1:
        execute_clean = st.button(
            "🗑️ Execute Clean Operation",
            type="primary",
            disabled=dry_run and not st.session_state.scanned_files
        )
    
    with col2:
        reset_scan = st.button("🔄 Reset Scan")
    
    if reset_scan:
        st.session_state.scanned_files = []
        st.session_state.scan_complete = False
        st.session_state.cleaning_results = None
        st.rerun()
    
    if execute_clean:
        if confirm_delete and not dry_run:
            if not st.checkbox("✅ I confirm I want to delete these files"):
                st.warning("Please confirm before proceeding")
                st.stop()
        
        with st.spinner("Executing clean operation..."):
            try:
                results = delete_files(
                    st.session_state.scanned_files,
                    dry_run=dry_run
                )
                
                st.session_state.cleaning_results = results
                
                if dry_run:
                    success_msg = """
                    <div class="success-box">
                    <strong>✅ Dry Run Complete:</strong> No files were deleted. 
                    This was a preview of what would be cleaned.
                    </div>
                    """
                else:
                    success_msg = """
                    <div class="success-box">
                    <strong>✅ Cleaning Complete:</strong> Files have been deleted successfully.
                    </div>
                    """
                
                st.markdown(success_msg, unsafe_allow_html=True)
                
                # Display results
                if results:
                    st.subheader("📊 Cleaning Results")
                    
                    result_col1, result_col2, result_col3 = st.columns(3)
                    with result_col1:
                        st.metric("Files Processed", results.get('processed', 0))
                    with result_col2:
                        st.metric("Files Deleted", results.get('deleted', 0))
                    with result_col3:
                        st.metric("Space Freed", format_size(results.get('size_freed', 0)))
                    
                    if results.get('errors', []):
                        st.subheader("⚠️ Errors")
                        for error in results['errors']:
                            st.error(error)
                
            except Exception as e:
                st.error(f"❌ Error during cleaning: {str(e)}")
                import traceback
                st.code(traceback.format_exc())

# Help and Documentation
st.header("❓ Help & Documentation")

with st.expander("How to Use winapp2.ini", expanded=False):
    st.markdown("""
    **winapp2.ini** is a configuration file format used by cleaning tools like BleachBit.
    
    **Structure:**
    ```ini
    [SectionName]
    Description=What this section cleans
    Warning=Warning message for users
    path=C:\\path\\to\\directory
    file=*.ext; *.log
    regex=pattern_to_match
    
    **Common Sections:**
    - `Temp`: Temporary files
    - `Cache`: Browser and application caches
    - `Logs`: Application log files
    - `Recycle`: Recycle bin contents
    """)

with st.expander("Safety Tips", expanded=False):
    st.markdown("""
    **⚠️ Important Safety Guidelines:**
    
    1. **Always use Dry Run mode first** to preview what will be deleted
    2. **Review the file list** before executing deletion
    3. **Backup important data** before cleaning
    4. **Don't clean system files** unless you know what they do
    5. **Start with user data** (Temp, Cache, Logs) before system files
    6. **Check application behavior** after cleaning to ensure nothing broke
    """)

with st.expander("Troubleshooting", expanded=False):
    st.markdown("""
    **Common Issues:**
    
    - **Permission Errors**: Run as administrator or check file permissions
    - **File Not Found**: Paths may be incorrect or files already deleted
    - **INI Parse Errors**: Check INI file format and encoding
    - **Slow Scanning**: Large directories or network drives can be slow
    
    ** Solutions:**
    - Use specific paths instead of wildcards
    - Exclude system directories if not needed
    - Run in dry run mode first
    - Check antivirus software isn't blocking access
    """)

# Footer
st.markdown("---")
st.markdown("Made with ❤️ using Streamlit | [Built with anycoder](https://huggingface.co/spaces/akhaliq/anycoder)")