KaraokeVideoDownloader/fix_artist_name_format_simple.py

#!/usr/bin/env python3
"""
Fix artist name formatting for Let's Sing Karaoke channel.

This script specifically targets the "Last Name, First Name" format and converts it to
"First Name Last Name" format in ID3 tags. It only processes entries where there is exactly one comma
followed by exactly 2 words, to avoid affecting multi-artist entries.

Usage:
    python fix_artist_name_format_simple.py --preview  # Show what would be changed
    python fix_artist_name_format_simple.py --apply    # Actually make the changes
    python fix_artist_name_format_simple.py --external "D:\Karaoke\Karaoke\MP4\Let's Sing Karaoke"  # Use external directory
"""

import json
import os
import re
import shutil
import argparse
from pathlib import Path
from typing import Dict, List, Tuple, Optional

# Try to import mutagen for ID3 tag manipulation
try:
    from mutagen.mp4 import MP4
    MUTAGEN_AVAILABLE = True
except ImportError:
    MUTAGEN_AVAILABLE = False
    print("WARNING: mutagen not available - install with: pip install mutagen")


def is_lastname_firstname_format(artist_name: str) -> bool:
    """
    Check if artist name is in "Last Name, First Name" format.
    
    Args:
        artist_name: The artist name to check
        
    Returns:
        True if the name matches "Last Name, First Name" format with exactly 1 or 2 words after comma
    """
    if ',' not in artist_name:
        return False
    
    # Split by comma
    parts = artist_name.split(',', 1)
    if len(parts) != 2:
        return False
    
    last_name = parts[0].strip()
    first_name_part = parts[1].strip()
    
    # Check if there are exactly 1 or 2 words after the comma
    words_after_comma = first_name_part.split()
    if len(words_after_comma) not in [1, 2]:
        return False
    
    # Additional check: make sure it's not a multi-artist entry
    # If there are more than 4 words total in the artist name, it might be multi-artist
    total_words = len(artist_name.split())
    if total_words > 4:  # Last, First Name (4 words max for single artist)
        return False
    
    return True


def convert_lastname_firstname(artist_name: str) -> str:
    """
    Convert "Last Name, First Name" to "First Name Last Name".
    
    Args:
        artist_name: The artist name to convert
        
    Returns:
        The converted artist name
    """
    if ',' not in artist_name:
        return artist_name
    
    parts = artist_name.split(',', 1)
    if len(parts) != 2:
        return artist_name
    
    last_name = parts[0].strip()
    first_name = parts[1].strip()
    
    return f"{first_name} {last_name}"


def process_artist_name(artist_name: str) -> str:
    """
    Process an artist name, handling both single artists and multiple artists separated by "&".
    
    Args:
        artist_name: The artist name to process
        
    Returns:
        The processed artist name
    """
    if '&' in artist_name:
        # Split by "&" and process each artist individually
        artists = [artist.strip() for artist in artist_name.split('&')]
        processed_artists = []
        
        for artist in artists:
            if is_lastname_firstname_format(artist):
                processed_artist = convert_lastname_firstname(artist)
                processed_artists.append(processed_artist)
            else:
                processed_artists.append(artist)
        
        # Rejoin with "&"
        return ' & '.join(processed_artists)
    else:
        # Single artist
        if is_lastname_firstname_format(artist_name):
            return convert_lastname_firstname(artist_name)
        else:
            return artist_name


def update_id3_tags(file_path: str, new_artist: str, apply_changes: bool = False) -> bool:
    """
    Update the ID3 tags in an MP4 file.
    
    Args:
        file_path: Path to the MP4 file
        new_artist: New artist name to set
        apply_changes: Whether to actually apply changes or just preview
        
    Returns:
        True if successful, False otherwise
    """
    if not MUTAGEN_AVAILABLE:
        print(f"WARNING: mutagen not available - cannot update ID3 tags for {file_path}")
        return False
    
    try:
        mp4 = MP4(file_path)
        
        if apply_changes:
            # Update the artist tag
            mp4["\xa9ART"] = new_artist
            mp4.save()
            print(f"UPDATED ID3 tag: {os.path.basename(file_path)} -> Artist: '{new_artist}'")
        else:
            # Just preview what would be changed
            current_artist = mp4.get("\xa9ART", ["Unknown"])[0] if "\xa9ART" in mp4 else "Unknown"
            print(f"WOULD UPDATE ID3 tag: {os.path.basename(file_path)} -> Artist: '{current_artist}' -> '{new_artist}'")
        
        return True
        
    except Exception as e:
        print(f"ERROR: Failed to update ID3 tags for {file_path}: {e}")
        return False


def scan_external_directory(directory_path: str, debug: bool = False) -> List[Dict]:
    """
    Scan external directory for MP4 files with "Last Name, First Name" format in ID3 tags.
    
    Args:
        directory_path: Path to the external directory
        debug: Whether to show debug information
        
    Returns:
        List of files that need ID3 tag updates
    """
    if not os.path.exists(directory_path):
        print(f"ERROR: Directory not found: {directory_path}")
        return []
    
    if not MUTAGEN_AVAILABLE:
        print("ERROR: mutagen not available - cannot scan ID3 tags")
        return []
    
    files_to_update = []
    total_files = 0
    files_with_artist_tags = 0
    
    # Scan for MP4 files
    for file_path in Path(directory_path).glob("*.mp4"):
        total_files += 1
        try:
            mp4 = MP4(str(file_path))
            current_artist = mp4.get("\xa9ART", ["Unknown"])[0] if "\xa9ART" in mp4 else "Unknown"
            
            if current_artist != "Unknown":
                files_with_artist_tags += 1
                
                if debug:
                    print(f"DEBUG: {file_path.name} -> Artist: '{current_artist}'")
                
                # Process the artist name to handle multiple artists
                processed_artist = process_artist_name(current_artist)
                
                if processed_artist != current_artist:
                    files_to_update.append({
                        'file_path': str(file_path),
                        'filename': file_path.name,
                        'old_artist': current_artist,
                        'new_artist': processed_artist
                    })
                    
                    if debug:
                        print(f"DEBUG: MATCH FOUND - {file_path.name}: '{current_artist}' -> '{processed_artist}'")
                
        except Exception as e:
            if debug:
                print(f"WARNING: Could not read ID3 tags from {file_path.name}: {e}")
    
    print(f"INFO: Scanned {total_files} MP4 files, {files_with_artist_tags} had artist tags, {len(files_to_update)} need updates")
    return files_to_update


def update_id3_tags_for_files(files_to_update: List[Dict], apply_changes: bool = False) -> int:
    """
    Update ID3 tags for a list of files.
    
    Args:
        files_to_update: List of files to update
        apply_changes: Whether to actually apply changes or just preview
        
    Returns:
        Number of files successfully updated
    """
    updated_count = 0
    
    for file_info in files_to_update:
        file_path = file_info['file_path']
        new_artist = file_info['new_artist']
        
        if update_id3_tags(file_path, new_artist, apply_changes):
            updated_count += 1
    
    return updated_count


def main():
    """Main function to run the artist name fix script."""
    parser = argparse.ArgumentParser(description="Fix artist name formatting in ID3 tags for Let's Sing Karaoke")
    parser.add_argument('--preview', action='store_true', help='Show what would be changed without making changes')
    parser.add_argument('--apply', action='store_true', help='Actually apply the changes')
    parser.add_argument('--external', type=str, help='Path to external karaoke directory')
    parser.add_argument('--debug', action='store_true', help='Show debug information')
    
    args = parser.parse_args()
    
    # Default to preview mode if no action specified
    if not args.preview and not args.apply:
        args.preview = True
    
    print("Artist Name Format Fix Script (ID3 Tags Only)")
    print("=" * 60)
    print("This script will fix 'Last Name, First Name' format to 'First Name Last Name'")
    print("Only targeting Let's Sing Karaoke channel to avoid affecting other channels.")
    print("Focusing on ID3 tags only - filenames will not be changed.")
    print()
    
    if not MUTAGEN_AVAILABLE:
        print("ERROR: mutagen library not available!")
        print("Please install it with: pip install mutagen")
        return
    
    if args.preview:
        print("PREVIEW MODE - No changes will be made")
    else:
        print("APPLY MODE - Changes will be made")
    print()
    
    # Process external directory if specified
    if args.external:
        print(f"Scanning external directory: {args.external}")
        external_files = scan_external_directory(args.external, debug=args.debug)
        
        if external_files:
            print(f"\nFound {len(external_files)} files with 'Last Name, First Name' format in ID3 tags:")
            for file_info in external_files:
                print(f"  * {file_info['filename']}: '{file_info['old_artist']}' -> '{file_info['new_artist']}'")
            
            if args.apply:
                print(f"\nUpdating ID3 tags in external files...")
                updated_count = update_id3_tags_for_files(external_files, apply_changes=True)
                print(f"SUCCESS: Updated ID3 tags in {updated_count} external files")
            else:
                print(f"\nWould update ID3 tags in {len(external_files)} external files")
        else:
            print("SUCCESS: No files with 'Last Name, First Name' format found in ID3 tags")
    
    print("\n" + "=" * 60)
    print("Summary complete.")


if __name__ == "__main__":
    main()