sane-tsv/SaneTsv/ExtraTsv/ExtraTsv.cs


using System.Globalization;
using System.Text.RegularExpressions;

namespace NathanMcRae;

public class ExtraTsv : SaneTsv
{
  public class Iso8601Type : ColumnType { }
  public class PhysicalUnitsType : ColumnType 
  {
    public string Units { get; }
    public PhysicalUnitsType(string Units) { }
  }

  public static readonly string[] ValidUnits =
  {
    "m",
    "s",
    "A",
    "K",
    "cd",
    "mol",
    "kg",
    "Hz",
    "rad",
    "sr",
    "N",
    "Pa",
    "J",
    "W",
    "C",
    "V",
    "F",
    "Ω",
    "S",
    "Wb",
    "T",
    "H",
    "°C",
    "lm",
    "lx",
    "Bq",
    "Gy",
    "Sv",
    "kat"
  };

  public static readonly int MajorVersion = 0;
  public static readonly int MinorVersion = 0;
  public static readonly int PatchVersion = 1;

  public static Regex VersionRegex = new Regex(@"^ ExtraTSV V(\d+)\.(\d+)\.(\d+)");

  public static ExtraTsv ParseExtraTsv(byte[] inputBuffer)
  {
    SaneTsv tsv = ParseCommentedTsv(inputBuffer);

    if (tsv.FileComment == null) {
      throw new Exception($"ExtraTSV expects the file to start with '# ExtraTSV Vx.y.z' where x.y.z is a version compatible with {MajorVersion}.{MinorVersion}.{PatchVersion}");
    }

    Match match = VersionRegex.Match(tsv.FileComment);
    if (!match.Success)
    {
      throw new Exception($"ExtraTSV expects the file to start with '# ExtraTSV Vx.y.z' where x.y.z is a version compatible with {MajorVersion}.{MinorVersion}.{PatchVersion}");
    }

    int fileMajorVersion = int.Parse(match.Groups[1].Value);

    if (fileMajorVersion != MajorVersion)
    {
      throw new Exception($"File has major version ({fileMajorVersion}) which is newer than this parser's version {MajorVersion}");
    }

    for (int i = 0; i < tsv.ColumnNames.Count(); i++)
    {
      string[] typeParts = tsv.ColumnNames[i].Split(":");
      if (typeParts[typeParts.Length - 1] == "iso8601" && tsv.ColumnTypes[i] == typeof(StringType))
      {
        string columnName = tsv.ColumnNames[i].Substring(0, tsv.ColumnNames[i].Length - ":iso8601".Length);
        tsv.ColumnNames[i] = columnName;
        tsv.ColumnTypes[i] = typeof(Iso8601Type);
      }
      // TODO: ISO8601 time spans
      // TODO: ISO8601 time durations
      else if (typeParts[typeParts.Length - 1] == "units" && (tsv.ColumnTypes[i] == typeof(Float64Type) || tsv.ColumnTypes[i] == typeof(Float32Type)))
      {
        if (typeParts.Count() > 1 && ValidUnits.Contains(typeParts[typeParts.Length - 2]))
        {
          // TODO: How to store type information since the ColumnTypes is of type Type?
        }
        else
        {
          throw new Exception($"Invalid units type '{typeParts[typeParts.Length - 2]}' for column {i}");
        }
      }
    }

    CultureInfo provider = CultureInfo.InvariantCulture;

    for (int i = 0; i < tsv.Records.Count; i++)
    {
      if (tsv.Records[i].Comment != null)
      {
        throw new Exception($"Line {tsv.Records[i].Line} has comment above it which is not allowed");
      }

      for (int j = 0; j < tsv.ColumnNames.Count(); j++)
      {
        if (tsv.ColumnTypes[j] == typeof(Iso8601Type))
        {
          if (!DateTime.TryParseExact((string)tsv.Records[i][j], "yyyy-MM-ddTHH:mm:ss.ffff", provider, DateTimeStyles.None, out DateTime parsed))
          {
            throw new Exception($"ISO 8601 timestamp format error on line {tsv.Records[i].Line}, field {j}");
          }

          tsv.Records[i].Fields[j] = parsed;
        }
      }
    }

    return (ExtraTsv)tsv;
  }
}