Vocaluxe/VocaluxeSongServer/Library/CLibraryScanner.cs

using System.Security.Cryptography;
using System.Text;
using VocaluxeSongServer.Contracts;

namespace VocaluxeSongServer.Library
{
    /// <summary>A scanned library: the DTO sent to clients plus the lookup of downloadable files.</summary>
    public sealed class CLibrarySnapshot
    {
        public required LibraryDto Dto { get; init; }
        /// <summary>Relative path ('/') -> full path, case-sensitive like the file system on the server.</summary>
        public required IReadOnlyDictionary<string, string> FullPaths { get; init; }
    }

    /// <summary>
    ///     Finds UltraStar songs (text files with #TITLE, #ARTIST and #MP3/#AUDIO) below the library root.
    ///     Only the header of each text file is read; unchanged files are not read again on the next scan.
    /// </summary>
    public sealed class CLibraryScanner
    {
        private sealed record CTxtInfo(long Size, long Modified, CSongHeader? Header);

        public sealed record CSongHeader(string Title, string Artist);

        private readonly ILogger<CLibraryScanner> _Logger;
        private Dictionary<string, CTxtInfo> _TxtCache = new(StringComparer.Ordinal);

        public CLibraryScanner(ILogger<CLibraryScanner> logger)
        {
            _Logger = logger;
        }

        private static readonly EnumerationOptions _Recursive = new()
        {
            RecurseSubdirectories = true,
            IgnoreInaccessible = true,
            MatchCasing = MatchCasing.CaseInsensitive,
            AttributesToSkip = FileAttributes.Hidden | FileAttributes.System
        };

        private static readonly EnumerationOptions _TopLevel = new()
        {
            RecurseSubdirectories = false,
            IgnoreInaccessible = true,
            AttributesToSkip = FileAttributes.Hidden | FileAttributes.System | FileAttributes.Directory
        };

        public CLibrarySnapshot Scan(string root)
        {
            root = Path.GetFullPath(root);
            var songs = new List<SongDto>();
            var songFolders = new HashSet<string>(StringComparer.Ordinal);
            var newCache = new Dictionary<string, CTxtInfo>(StringComparer.Ordinal);

            foreach (var txt in Directory.EnumerateFiles(root, "*.txt", _Recursive))
            {
                FileInfo fi;
                try
                {
                    fi = new FileInfo(txt);
                }
                catch (Exception e)
                {
                    _Logger.LogWarning("Can't read {File}: {Message}", txt, e.Message);
                    continue;
                }

                var modified = new DateTimeOffset(fi.LastWriteTimeUtc).ToUnixTimeMilliseconds();
                if (!_TxtCache.TryGetValue(txt, out var info) || info.Size != fi.Length || info.Modified != modified)
                {
                    info = new CTxtInfo(fi.Length, modified, ReadHeader(txt));
                }

                newCache[txt] = info;
                if (info.Header == null)
                {
                    continue;
                }

                var rel = ToRelative(root, txt);
                songs.Add(new SongDto { Id = SongId(rel), Path = rel, Title = info.Header.Title, Artist = info.Header.Artist });
                songFolders.Add(fi.DirectoryName!);
            }

            _TxtCache = newCache;

            var files = new List<FileDto>();
            var fullPaths = new Dictionary<string, string>(StringComparer.Ordinal);
            foreach (var folder in songFolders)
            {
                foreach (var file in Directory.EnumerateFiles(folder, "*", _TopLevel))
                {
                    var fi = new FileInfo(file);
                    var rel = ToRelative(root, file);
                    files.Add(new FileDto { Path = rel, Size = fi.Length, Modified = new DateTimeOffset(fi.LastWriteTimeUtc).ToUnixTimeMilliseconds() });
                    fullPaths[rel] = file;
                }
            }

            songs.Sort((a, b) => string.CompareOrdinal(a.Path, b.Path));
            files.Sort((a, b) => string.CompareOrdinal(a.Path, b.Path));

            var dto = new LibraryDto
            {
                Version = ComputeVersion(files),
                ScannedAt = DateTimeOffset.UtcNow.ToString("O"),
                Songs = songs,
                Files = files
            };
            return new CLibrarySnapshot { Dto = dto, FullPaths = fullPaths };
        }

        public static string ToRelative(string root, string path)
        {
            return Path.GetRelativePath(root, path).Replace('\\', '/');
        }

        public static string SongId(string relativePath)
        {
            var hash = SHA256.HashData(Encoding.UTF8.GetBytes(relativePath));
            return Convert.ToHexString(hash, 0, 8).ToLowerInvariant();
        }

        private static string ComputeVersion(List<FileDto> files)
        {
            using var sha = IncrementalHash.CreateHash(HashAlgorithmName.SHA256);
            foreach (var f in files)
            {
                sha.AppendData(Encoding.UTF8.GetBytes(f.Path + "\n" + f.Size + "\n" + f.Modified + "\n"));
            }

            return Convert.ToHexString(sha.GetHashAndReset(), 0, 12).ToLowerInvariant();
        }

        /// <summary>
        ///     Reads the header of an UltraStar text file. Returns null if it is not a song
        ///     (no #TITLE, #ARTIST or #MP3/#AUDIO tag before the first note line).
        /// </summary>
        public CSongHeader? ReadHeader(string path)
        {
            try
            {
                // The header is at the top; 64 KB is plenty even with long comments.
                byte[] bytes;
                using (var fs = File.OpenRead(path))
                {
                    var len = (int)Math.Min(fs.Length, 64 * 1024);
                    bytes = new byte[len];
                    fs.ReadExactly(bytes, 0, len);
                    if (fs.Length > len)
                    {
                        // Cut at the last line break so a UTF-8 character isn't split (which would look like Latin-1).
                        var lastBreak = Array.LastIndexOf(bytes, (byte)'\n');
                        if (lastBreak > 0)
                        {
                            Array.Resize(ref bytes, lastBreak);
                        }
                    }
                }

                return ParseHeader(DecodeText(bytes));
            }
            catch (Exception e)
            {
                _Logger.LogWarning("Can't read {File}: {Message}", path, e.Message);
                return null;
            }
        }

        public static CSongHeader? ParseHeader(string text)
        {
            string? title = null, artist = null;
            var hasAudio = false;
            foreach (var rawLine in text.Split('\n'))
            {
                var line = rawLine.Trim();
                if (line.Length == 0)
                {
                    continue;
                }

                if (line[0] != '#')
                {
                    break;
                }

                var colon = line.IndexOf(':');
                if (colon < 0)
                {
                    continue;
                }

                var tag = line.Substring(1, colon - 1).Trim().ToUpperInvariant();
                var value = line.Substring(colon + 1).Trim();
                switch (tag)
                {
                    case "TITLE":
                        title = value;
                        break;
                    case "ARTIST":
                        artist = value;
                        break;
                    case "MP3":
                    case "AUDIO":
                        hasAudio |= value.Length > 0;
                        break;
                }
            }

            return !string.IsNullOrEmpty(title) && !string.IsNullOrEmpty(artist) && hasAudio ? new CSongHeader(title, artist) : null;
        }

        /// <summary>UTF-8 (with or without BOM) if valid, otherwise Windows-1252/Latin-1 like most old song files.</summary>
        public static string DecodeText(byte[] bytes)
        {
            var span = bytes.AsSpan();
            if (span.StartsWith(new byte[] { 0xEF, 0xBB, 0xBF }))
            {
                span = span[3..];
            }

            try
            {
                return new UTF8Encoding(false, true).GetString(span);
            }
            catch (DecoderFallbackException)
            {
                return Encoding.Latin1.GetString(span);
            }
        }
    }
}