プチ業務改善 ファイルのエンコーディングを気にせず読み込む便利クラス

ちょっとしたツールを作る際にエンコーディングを気にしなくて済むと便利。
いつもWindows環境で仕事をしているんだけど、お目にかかるファイルのエンコーディングはShiftJISとUTF8、UTF16くらい。
ユニコードエンコーディングの場合は基本的にBOMがついているので、エンコーディングを自動判定するクラスを作ってみた。
備忘録として残しておく(^^)

FileHelper.rb

using System;
using System.Collections.Generic;
using System.Diagnostics;
using System.IO;
using System.Linq;
using System.Text;

namespace GeoLibrary.IO
{
    public class FileHelper
    {
        public static StreamReader OpenReader(FileInfo target)
        {
            return new StreamReaderOpener(target).Open();
        }

        public static IEnumerable<string> OpenAndReadAllLines(FileInfo target)
        {
            using (var sr = OpenReader(target))
            {
                return sr.ReadToEnd().Split('\n').Select(_ => _.TrimEnd('\r'));
            }
        }

        private class StreamReaderOpener
        {
            private readonly FileInfo _target;

            internal StreamReaderOpener(FileInfo target)
            {
                if (target == null)
                {
                    throw new ArgumentNullException(nameof(target));
                }
                _target = target;
            }

            internal StreamReader Open()
            {
                if (!File.Exists(_target.FullName))
                {
                    var message = $"File does not exists => {_target.FullName}";
                    throw new ArgumentException(message);
                }
                var stream = _target.OpenRead();
                var encodingDetector = new FileEncodingDetector(() => stream);
                return new StreamReader(stream, encodingDetector.Detect());
            }
        }

        private class FileEncodingDetector
        {
            private readonly IDictionary<ByteOrderMarker.BomType, Encoding> EncodingDictionary
                = new Dictionary<ByteOrderMarker.BomType, Encoding>()
            {
				{ ByteOrderMarker.BomType.Utf8, Encoding.UTF8 },
				{ ByteOrderMarker.BomType.Unicode, Encoding.Unicode },
				{ ByteOrderMarker.BomType.BigEndianUnicode, Encoding.BigEndianUnicode },
                { ByteOrderMarker.BomType.Utf32, Encoding.UTF32 },
            };
            private readonly Func<FileStream> _streamGetter;

            internal FileEncodingDetector(Func<FileStream> streamGetter)
            {
                if (streamGetter == null)
                {
                    throw new ArgumentNullException(nameof(streamGetter));
                }
                _streamGetter = streamGetter;
            }

            internal Encoding Detect()
            {
                var stream = _streamGetter();
                Debug.Assert(stream != null);
                var position = stream.Position;
                var bomType = new ByteOrderMarker(TryReadByteFromTop(4)).ToType();
                stream.Seek(position, SeekOrigin.Begin);
                return EncodingDictionary.ContainsKey(bomType) ?
                    EncodingDictionary[bomType] : Encoding.Default;
            }

            private void MoveToTopPosition()
            {
                var stream = _streamGetter();
                Debug.Assert(stream != null);
                stream.Seek(0, SeekOrigin.Begin);
            }

            private IEnumerable<byte> TryReadByteFromTop(int count)
            {
                var stream = _streamGetter();
                Debug.Assert(stream != null);
                MoveToTopPosition();
                var bytes = new List<byte>();
                for (var i = 0; i < count; ++i)
                {
                    if (stream.Length <= i)
                    {
                        break;
                    }
                    bytes.Add((byte)stream.ReadByte());
                }
                return bytes;
            }
        }

        private class ByteOrderMarker
        {
            internal enum BomType
            {
                Utf8,
                Unicode,
                BigEndianUnicode,
                Utf32,
                Unknown,
            }

            private readonly IEnumerable<byte> _bom;

            internal ByteOrderMarker(IEnumerable<byte> bom)
            {
                _bom = bom;
            }

            internal BomType ToType()
            {
                var dic = new[] {
					new { Bom = new byte[] { 0xEF, 0xBB, 0xBF }, Type = BomType.Utf8 },
					new { Bom = new byte[] { 0xFF, 0xFE }, Type = BomType.Unicode },
					new { Bom = new byte[] { 0xFE, 0xFF }, Type = BomType.BigEndianUnicode },
					new { Bom = new byte[] { 0xFF, 0xFE, 0x00, 0x00 }, Type = BomType.Utf32 },
                };

                var candidate = dic.Where(_ => Enumerable.SequenceEqual(_.Bom, _bom));
                Debug.Assert(candidate.Count() <= 1);
                return candidate.Count() != 0 ?
                    candidate.First().Type : BomType.Unknown;
            }
        }
    }
}