Reklamator/app/utils/file_validator.py

"""File upload validation utilities"""
import os
from werkzeug.utils import secure_filename
import clamd
from flask import current_app


# Allowed file extensions for attachments
ALLOWED_EXTENSIONS = {
    'txt', 'log', 'pdf', 'png', 'jpg', 'jpeg', 'gif',
    'doc', 'docx', 'xls', 'xlsx', 'csv'
}

# Maximum file size (10MB)
MAX_FILE_SIZE = 10 * 1024 * 1024


def allowed_file(filename):
    """Check if file extension is allowed

    Args:
        filename: Name of the uploaded file

    Returns:
        bool: True if extension is allowed, False otherwise
    """
    if not filename:
        return False

    return '.' in filename and \
           filename.rsplit('.', 1)[1].lower() in ALLOWED_EXTENSIONS


def validate_file_size(file_stream):
    """Check if file size is within limits

    Args:
        file_stream: File stream object

    Returns:
        bool: True if size is acceptable, False otherwise
    """
    # Seek to end to get file size
    file_stream.seek(0, os.SEEK_END)
    size = file_stream.tell()
    # Reset to beginning
    file_stream.seek(0)

    return size <= MAX_FILE_SIZE


def get_safe_filename(filename):
    """Get secure version of filename

    Args:
        filename: Original filename

    Returns:
        str: Secure filename safe for filesystem storage
    """
    return secure_filename(filename)


def validate_file(file):
    """Validate uploaded file

    Args:
        file: Werkzeug FileStorage object

    Returns:
        tuple: (is_valid, error_message)
            is_valid: bool indicating if file is valid
            error_message: str with error description or None
    """
    if not file:
        return False, "No file provided"

    if file.filename == '':
        return False, "No file selected"

    if not allowed_file(file.filename):
        return False, f"File type not allowed. Allowed types: {', '.join(ALLOWED_EXTENSIONS)}"

    if not validate_file_size(file.stream):
        return False, f"File size exceeds maximum of {MAX_FILE_SIZE / (1024 * 1024):.0f}MB"

    return True, None


def scan_file_for_viruses(file):
    """Scan file for viruses using ClamAV

    Args:
        file: Werkzeug FileStorage object

    Returns:
        tuple: (is_clean, error_message)
            is_clean: bool indicating if file is clean (True) or infected (False)
            error_message: str with error description or None
    """
    try:
        # Connect to ClamAV daemon
        clamd_socket = current_app.config.get('CLAMD_SOCKET')

        if not clamd_socket:
            # ClamAV not configured, skip scanning
            current_app.logger.warning("ClamAV socket not configured, skipping virus scan")
            return True, None

        cd = clamd.ClamdUnixSocket(clamd_socket)

        # Ping to check if ClamAV is available
        try:
            cd.ping()
        except Exception as e:
            current_app.logger.warning(f"ClamAV not available: {e}, skipping virus scan")
            return True, None

        # Read file content
        file.stream.seek(0)
        file_data = file.stream.read()
        file.stream.seek(0)  # Reset for later use

        # Scan file
        scan_result = cd.instream(file_data)

        # Check result
        if scan_result and 'stream' in scan_result:
            status, virus_name = scan_result['stream']

            if status == 'OK':
                return True, None
            elif status == 'FOUND':
                return False, f"Virus detected: {virus_name}"
            else:
                return False, f"Scan error: {status}"

        return True, None

    except Exception as e:
        current_app.logger.error(f"ClamAV scanning error: {e}")
        # On error, we'll allow the file but log the error
        # In production, you might want to reject files if scanning fails
        return True, None