"""
Utility functions for MIME type handling and file format determination.
"""

def isTextMimeType(mimeType: str) -> bool:
    """
    Determines if a MIME type represents a text format that should not be base64 encoded.
    
    Args:
        mimeType: The MIME type to check
        
    Returns:
        True if the content is a text format, False otherwise
    """
    return (
        mimeType.startswith("text/") or 
        mimeType in [
            "application/json", 
            "application/xml", 
            "application/javascript", 
            "application/x-python",
            "image/svg+xml"
        ]
    )

def determineContentEncoding(fileName: str, content: any, mimeType: str = None) -> bool:
    """
    Determines if content should be base64 encoded based on file type and MIME type.
    
    Args:
        fileName: Name of the file including extension
        content: The content of the file
        mimeType: Optional MIME type of the content
        
    Returns:
        True if content should be base64 encoded, False otherwise
    """
    # If MIME type is provided, use it for determination
    if mimeType:
        if isTextMimeType(mimeType):
            return False if isinstance(content, str) else True
    
    # Import here to avoid circular imports
    import os
    
    # Extract file extension
    _, extension = os.path.splitext(fileName)
    extension = extension.lower().lstrip('.')
    
    # Determine if we should base64 encode based on file type
    text_extensions = {'txt', 'csv', 'json', 'xml', 'html', 'md', 'svg', 'js', 'css', 'py'}
    
    # If it's a text format and content is a string, don't base64 encode
    if extension in text_extensions and isinstance(content, str):
        return False
    
    # For binary formats, always base64 encode
    binary_extensions = {'jpg', 'jpeg', 'png', 'gif', 'pdf', 'doc', 'docx', 'xls', 'xlsx', 'zip', 'rar'}
    if extension in binary_extensions:
        return True
        
    # If content is bytes, base64 encode regardless of extension
    if isinstance(content, bytes):
        return True
        
    # Default for unknown types
    return not isinstance(content, str)