HTML5文字実体参照をタブ区切りファイルに変換

private static void OutputAsTsv(string inputFilename, string outputFilename)
{
    using var inputStream = File.OpenRead(inputFilename);
    using var outputWriter = new StreamWriter(outputFilename, false);

    int no = 1;
    var document = JsonDocument.Parse(inputStream);
    foreach (var jsonProperty in document.RootElement.EnumerateObject())
    {
        // 文字実体参照名とコードポイント配列の取得
        var cer = jsonProperty.Name;
        var value = jsonProperty.Value;
        var codepoints = value.GetProperty("codepoints")
            .EnumerateArray().Select(e => e.GetInt32()).ToArray();

        // 出力データの編集
        WriteRecord(outputWriter, no++, codepoints, cer);
    }
}

private static void WriteRecord(StreamWriter writer, int no, int[] codepoints, string cer)
{
    int cp1 = codepoints[0];
    int? cp2 = codepoints.Length > 1 ? codepoints[1] : (int?)null;

    var cp1str = cp1.ToString();
    var cp2str = cp2?.ToString() ?? string.Empty;
    (var ncrDec1, var ncrHex1) = ToNcr(cp1);
    (var ncrDec2, var ncrHex2) = ToNcr(cp2);

    writer.WriteLine($"{no}\t{cer}\t{cp1str}\t{cp2str}" +
        $"\t{ncrDec1}\t{ncrDec2}\t{ncrHex1}\t{ncrHex2}");
}

private static (string dec, string hex) ToNcr(int? codepoint)
{
    if (!codepoint.HasValue) return (string.Empty, string.Empty);
    return ($"&#{codepoint.Value};" , $"&#x{codepoint.Value:X5};");
}

private static void OutputUniqueCodepointAsTsv(string inputFilename, string outputFilename)
{
    // コードポイント配列・文字実体参照ディクショナリ
    var dic = new Dictionary<int[], string>(new IntArrayEqualityComparer());

    // 重複するコードポイントを除外
    using var inputStream = File.OpenRead(inputFilename);
    var document = JsonDocument.Parse(inputStream);
    foreach (var jsonProperty in document.RootElement.EnumerateObject())
    {
        // 文字実体参照名とコードポイント配列の取得
        var newCer = jsonProperty.Name;
        var value = jsonProperty.Value;
        var codepoints = value.GetProperty("codepoints")
            .EnumerateArray().Select(e => e.GetInt32()).ToArray();

        if (!dic.ContainsKey(codepoints))
        {
            dic[codepoints] = newCer;
            continue;
        }

        // 同一のコードポイントが存在する場合、ルールに基づいて更新
        var oldCer = dic[codepoints];
        var oldCerLower = oldCer.ToLower();
        var newCerLower = newCer.ToLower();

        bool priorNew;
        // "&Xyz;" vs "&xyz" → 左辺を優先
        if (oldCerLower == newCerLower + ";") priorNew = false;
        // "&Xyz" vs "&xyz;" → 右辺を優先
        else if (oldCerLower + ";" == newCerLower) priorNew = true;
        // 大小文字無視で一致の場合、小文字側を優先
        else if (oldCerLower == newCerLower) priorNew = (newCer == newCerLower);
        // 大小文字無視で不一致の場合、桁数が少ない側を優先
        else priorNew = oldCer.Length > newCer.Length;

        // debug
        //Console.WriteLine($"{string.Join(",", codepoints)}: " +
        //    (priorNew ? newCer : oldCer ) + $" <- [ {oldCer} / {newCer} ]");

        if (priorNew) dic[codepoints] = newCer;
    }

    // データ出力
    using var outputWriter = new StreamWriter(outputFilename, false);
    var kvlist = dic.ToList();
    for (var i = 0; i < kvlist.Count; i++)
    {
        var codepoints = kvlist[i].Key;
        var cer = kvlist[i].Value;
        WriteRecord(outputWriter, i + 1, codepoints, cer);
    }
}

class IntArrayEqualityComparer : IEqualityComparer<int[]>
{
    public bool Equals([AllowNull] int[] x, [AllowNull] int[] y)
        => Enumerable.SequenceEqual(x, y);
    public int GetHashCode([DisallowNull] int[] obj)
    {
        var hash = new HashCode();
        foreach (var v in obj) hash.Add(obj[0]);
        return hash.ToHashCode();
    }
}

private static void WriteRecord(StreamWriter writer, int no, int[] codepoints, string cer)
{
    int cp1 = codepoints[0];
    int? cp2 = codepoints.Length > 1 ? codepoints[1] : (int?)null;

    var cp1str = cp1.ToString();
    var cp2str = cp2?.ToString() ?? string.Empty;
    (var ncrDec1, var ncrHex1) = ToNcr(cp1);
    (var ncrDec2, var ncrHex2) = ToNcr(cp2);

    writer.WriteLine($"{no}\t{cer}\t{cp1str}\t{cp2str}" +
        $"\t{ncrDec1}\t{ncrDec2}\t{ncrHex1}\t{ncrHex2}");
}

private static (string dec, string hex) ToNcr(int? codepoint)
{
    if (!codepoint.HasValue) return (string.Empty, string.Empty);
    return ($"&#{codepoint.Value};" , $"&#x{codepoint.Value:X5};");
}