StrangeTools/WarcViewerBlazorWinForm/Backend/IO/Archiving/FileReader.cs

using SharpCompress.Archives;

namespace WarcViewerBlazorWinForm.Backend.IO.Archiving
{
    internal class FileReader : IDisposable
    {
        private const string EXT_WARC = ".warc";
        private const string EXT_XZ = ".xz";

        private string? _inputFilePath;

        private IArchive? _archive;
        private IArchiveEntry[]? _archiveEntries;

        public bool? IsDirectRead { get; private set; }
        public ReadOnlyMemory<string>? PackagedEntryNames { get; set; }

        public async Task LoadFileAsync(string inputFilePath)
        {
            if (File.Exists(inputFilePath) == false) throw new FileNotFoundException("file not found", inputFilePath);

            await Task.Run(() =>
            {
                _inputFilePath = inputFilePath;

                IsDirectRead = Path.GetExtension(_inputFilePath).Equals(EXT_WARC, StringComparison.CurrentCultureIgnoreCase);

                if (IsDirectRead == true) return;
                _archive = ArchiveFactory.Open(_inputFilePath);
                _archiveEntries = _archive.Entries.ToArray();
                PackagedEntryNames = _archiveEntries.Select(p => p.Key).ToArray();
            });
        }

        public Stream OpenReadDirectly()
        {
            if (_inputFilePath == null) throw new InvalidOperationException("not load file yet");
            if (IsDirectRead == false) throw new InvalidOperationException($"archived file, should use {nameof(OpenEntryRead)} method");
            return File.OpenRead(_inputFilePath);
        }

        public Stream OpenEntryRead(int archiveEntryIndex)
        {
            if (_archiveEntries == null) throw new InvalidOperationException($"direct read warc file, should use {nameof(OpenReadDirectly)} method");
            var entry = _archiveEntries[archiveEntryIndex];
            return ReadEntryInternal(entry);
        }

        public Stream OpenEntryRead(string archiveEntryKey)
        {
            if (_archiveEntries == null) throw new InvalidOperationException($"direct read warc file, should use {nameof(OpenReadDirectly)} method");
            var entry = _archiveEntries.FirstOrDefault(p => string.Equals(p.Key, archiveEntryKey, StringComparison.OrdinalIgnoreCase));
            return entry == null
                ? throw new FileNotFoundException("entry not found", archiveEntryKey)
                : ReadEntryInternal(entry);
        }

        private static Stream ReadEntryInternal(IArchiveEntry entry)
        {
            var ext = Path.GetExtension(entry.Key).ToLower();

            switch (ext)
            {
                case EXT_WARC: return entry.OpenEntryStream();
                case EXT_XZ: return XzRwOps.ReadXzStream(entry.OpenEntryStream());
                default: throw new NotSupportedException("not supported inner compressed extension");
            }
        }

        public void Dispose() => _archive?.Dispose();

        public long GetLengthDirectly()
        {
            if (_inputFilePath == null) throw new InvalidOperationException("not load file yet");
            if (IsDirectRead == false) throw new InvalidOperationException($"archived file, should use {nameof(GetEntryLength)} method");
            return new FileInfo(_inputFilePath).Length;
        }

        public long GetEntryLength(int archiveEntryIndex)
        {
            if (_archiveEntries == null) throw new InvalidOperationException($"direct read warc file, should use {nameof(GetLengthDirectly)} method");
            var entry = _archiveEntries[archiveEntryIndex];
            var stream = ReadEntryInternal(entry);
            return stream.GetLengthStupid();
        }

        public long GetEntryLength(string archiveEntryKey)
        {
            if (_archiveEntries == null) throw new InvalidOperationException($"direct read warc file, should use {nameof(GetLengthDirectly)} method");
            var entry = _archiveEntries.FirstOrDefault(p => string.Equals(p.Key, archiveEntryKey, StringComparison.OrdinalIgnoreCase));
            if (entry == null) throw new FileNotFoundException("entry not found", archiveEntryKey);
            var stream = ReadEntryInternal(entry);
            return stream.GetLengthStupid();
        }

        public static async Task<long> GetLengthAsync(FileDescriptor fd)
        {
            using var ar = new FileReader();
            await ar.LoadFileAsync(fd.FilePath);
            if (fd.IsReadDirectly) return ar.GetLengthDirectly();
            if (fd.ArchiveEntryKey == null) throw new ArgumentNullException(nameof(fd.ArchiveEntryKey));
            return ar.GetEntryLength(fd.ArchiveEntryKey);
        }

        public static async Task<Stream> OpenReadStreamAsync(FileDescriptor fd)
        {
            var ar = new FileReader();
            await ar.LoadFileAsync(fd.FilePath);
            if (fd.IsReadDirectly) return ar.OpenReadDirectly();
            if (fd.ArchiveEntryKey == null) throw new ArgumentNullException(nameof(fd.ArchiveEntryKey));
            return new DisposeStream(ar.OpenEntryRead(fd.ArchiveEntryKey), ar);
        }

        private class DisposeStream(Stream underlyingStream, IDisposable disposable) : Stream
        {
            protected override void Dispose(bool disposing)
            {
                underlyingStream?.Dispose();
                underlyingStream = null!;
                disposable?.Dispose();
                disposable = null!;
            }

            public override void Flush() => underlyingStream.Flush();

            public override int Read(byte[] buffer, int offset, int count) => underlyingStream.Read(buffer, offset, count);

            public override long Seek(long offset, SeekOrigin origin) => underlyingStream.Seek(offset, origin);

            public override void SetLength(long value) => underlyingStream.SetLength(value);

            public override void Write(byte[] buffer, int offset, int count) => underlyingStream.Write(buffer, offset, count);

            public override bool CanRead => underlyingStream.CanRead;
            public override bool CanSeek => underlyingStream.CanSeek;
            public override bool CanWrite => underlyingStream.CanWrite;
            public override long Length => underlyingStream.Length;

            public override long Position
            {
                get => underlyingStream.Position;
                set => underlyingStream.Position = value;
            }
        }
    }
}