Add iso8601 parsing

This commit is contained in:
Nathan McRae 2024-03-10 22:43:11 -07:00
parent 71af3653c4
commit 0807f0ffc2
2 changed files with 137 additions and 732 deletions

View File

@ -1,5 +1,7 @@
using System.Reflection;
using System.Globalization;
using System.Reflection;
using System.Text;
using System.Text.RegularExpressions;
namespace NathanMcRae;
@ -32,6 +34,51 @@ public class SaneTsv
public class Int32Type : ColumnType { }
public class Int64Type : ColumnType { }
public class BinaryType : ColumnType { }
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+)");
protected enum FormatType
{
@ -71,6 +118,31 @@ public class SaneTsv
return (CommentedTsv<T>)Parse<T>(inputBuffer, FormatType.COMMENTED_TSV);
}
public static CommentedTsv<T> ParseExtraTsv<T>(byte[] inputBuffer) where T : CommentedTsvRecord, new()
{
CommentedTsv<T> parsed = (CommentedTsv<T>)Parse<T>(inputBuffer, FormatType.COMMENTED_TSV);
if (parsed.FileComment == null)
{
throw new Exception($"ExtraTSV expects the file to start with '# ExtraTSV Vx.y.z' where x.y.z is a version compatible with {MajorVersion}.{MinorVersion}.{PatchVersion}");
}
Match match = VersionRegex.Match(parsed.FileComment);
if (!match.Success)
{
throw new Exception($"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($"File has major version ({fileMajorVersion}) which is newer than this parser's version {MajorVersion}");
}
return parsed;
}
protected static Tsv<T> Parse<T>(byte[] inputBuffer, FormatType format) where T : TsvRecord, new()
{
Tsv<T> parsed;
@ -169,7 +241,7 @@ public class SaneTsv
throw new Exception($"Header field {fields.Count} is not valid UTF-8", e);
}
string columnTypeString;
string[] columnTypeStrings;
string columnName;
if (columnString.Contains(':'))
{
@ -177,8 +249,8 @@ public class SaneTsv
{
throw new Exception($"Header field {j} contains ':', which is not allowed for column names");
}
columnTypeString = columnString.Split(":").Last();
columnName = columnString.Substring(0, columnString.Length - columnTypeString.Length - 1);
columnTypeStrings = columnString.Split(":");
columnName = string.Join(":", columnTypeStrings.Take(columnTypeStrings.Length - 1));
}
else
{
@ -186,20 +258,30 @@ public class SaneTsv
{
throw new Exception($"Header field {fields.Count} has no type");
}
columnTypeString = "";
columnTypeStrings = new string[] { "" };
columnName = columnString;
}
ColumnType type;
switch (columnTypeString)
switch (columnTypeStrings.Last())
{
case "":
numTypesBlank++;
type = new StringType();
break;
case "string":
type = new StringType();
if (columnTypeStrings[columnTypeStrings.Length - 2] == "iso8601")
{
type = new Iso8601Type();
columnName = string.Join(":", columnTypeStrings.Take(columnTypeStrings.Length - 2));
}
// TODO: ISO8601 time spans
// TODO: ISO8601 time durations
else
{
type = new StringType();
}
break;
case "boolean":
type = new BooleanType();
@ -232,9 +314,11 @@ public class SaneTsv
type = new BinaryType();
break;
default:
throw new Exception($"Invalid type '{columnTypeString}' for column {j}");
throw new Exception($"Invalid type '{columnTypeStrings.Last()}' for column {j}");
}
// TODO: physical unit types
// TODO: Allow lax parsing (only worry about parsing columns that are given in the specifying type
if (columnNames[j] != columnName)
@ -692,6 +776,15 @@ public class SaneTsv
properties[j].SetValue(record, parsedInt64);
}
else if (columnTypes[j].GetType() == typeof(Iso8601Type))
{
if (!DateTime.TryParseExact(fieldString, "yyyy-MM-ddTHH:mm:ss.ffff", CultureInfo.InvariantCulture, DateTimeStyles.None, out DateTime parsed))
{
throw new Exception($"ISO 8601 timestamp format error on line {line}, field {j}");
}
properties[j].SetValue(record, parsed);
}
else
{
throw new Exception($"Unexpected type {columnTypes[j]}");
@ -1156,6 +1249,10 @@ public class SaneTsv
{
return new BinaryType();
}
else if (type == typeof(DateTime))
{
return new Iso8601Type();
}
else
{
throw new Exception($"Invalid type: {type.GetType()}");
@ -1208,6 +1305,10 @@ public class SaneTsv
{
return "binary";
}
else if (type.GetType() == typeof(Iso8601Type))
{
return "iso8601:string";
}
else
{
throw new Exception($"Invalid type: {type.GetType()}");
@ -1229,6 +1330,11 @@ public class SaneTsv
return SerializeTsv<T>(data, FormatType.COMMENTED_TSV, fileComment);
}
public static byte[] SerializeExtraTsv<T>(IList<T> data) where T : TsvRecord
{
return SerializeTsv<T>(data, FormatType.COMMENTED_TSV, $" ExtraTSV V{MajorVersion}.{MinorVersion}.{PatchVersion}");
}
protected static byte[] SerializeTsv<T>(IList<T> data, FormatType tsvFormat, string fileComment = null)
{
var bytes = new List<byte>();
@ -1468,6 +1574,10 @@ public class SaneTsv
{
fieldEncoded = (byte[])datum;
}
else if (columnTypes[j].GetType() == typeof(Iso8601Type))
{
fieldEncoded = Encoding.UTF8.GetBytes(((DateTime)datum).ToString("yyyy-MM-ddTHH:mm:ss.ffff"));
}
else
{
throw new Exception($"Unexpected column type {columnTypes[j]} for column {j}");

View File

@ -1,63 +1,9 @@
using NathanMcRae;
using System.Reflection;
using System.Text;
internal class Program : SaneTsv
{
public class TestRecord : SaneTsv.TsvRecord
{
[SaneTsv.TypedTsvColumn("string-test")]
public string StringTest { get; set; }
[SaneTsv.TypedTsvColumn("bool-test")]
public bool BoolTest { get; set; }
[SaneTsv.TypedTsvColumn("float32-test")]
public float Float32Test { get; set; }
[SaneTsv.TypedTsvColumn("float32-le-test", typeof(SaneTsv.Float32LEType))]
public float Float32LETest { get; set; }
[SaneTsv.TypedTsvColumn("float64-test")]
public double Float64Test { get; set; }
[SaneTsv.TypedTsvColumn("float64-le-test", typeof(SaneTsv.Float64LEType))]
public double Float64LETest { get; set; }
[SaneTsv.TypedTsvColumn("uint32-test")]
public UInt32 UInt32Test { get; set; }
[SaneTsv.TypedTsvColumn("uint64-test")]
public UInt64 UInt64Test { get; set; }
[SaneTsv.TypedTsvColumn("int32-test")]
public Int32 Int32Test { get; set; }
[SaneTsv.TypedTsvColumn("int64-test")]
public Int64 Int64Test { get; set; }
[SaneTsv.TypedTsvColumn("binary-test")]
public byte[] BinaryTest { get; set; }
public TestRecord(string stringTest, bool boolTest, float float32Test, float float32LETest, double float64Test, double float64LETest, UInt32 uInt32Test, UInt64 uInt64Test, Int32 int32Test, Int64 int64Test, byte[] binaryTest)
{
StringTest = stringTest;
BoolTest = boolTest;
Float32Test = float32Test;
Float32LETest = float32LETest;
Float64Test = float64Test;
Float64LETest = float64LETest;
UInt32Test = uInt32Test;
UInt64Test = uInt64Test;
Int32Test = int32Test;
Int64Test = int64Test;
BinaryTest = binaryTest;
}
public TestRecord() { }
}
public class BoolTestRecord : SaneTsv.CommentedTsvRecord
public class DateTest : SaneTsv.CommentedTsvRecord
{
[SaneTsv.TypedTsvColumn("column1:ty#pe")]
public bool Column1 { get; set; }
@ -66,75 +12,19 @@ internal class Program : SaneTsv
public byte[] column2 { get; set; }
[SaneTsv.TypedTsvColumn("columnthree\nyep")]
public string Column3 { get; set; }
}
public class BoolTestRecord2 : SaneTsv.CommentedTsvRecord
{
[SaneTsv.TypedTsvColumn("column1:type")]
public bool Column1 { get; set; }
[SaneTsv.TypedTsvColumn]
public byte[] column2 { get; set; }
[SaneTsv.TypedTsvColumn("columnthree\nyep")]
public string Column3 { get; set; }
}
public class BoolTestRecord3 : SaneTsv.CommentedTsvRecord
{
[SaneTsv.TsvColumn("column1")]
public string Column1 { get; set; }
[SaneTsv.TsvColumn]
public string column2 { get; set; }
[SaneTsv.TsvColumn("columnthree\nyep")]
public string Column3 { get; set; }
}
public class SerdeTestRecord : SaneTsv.CommentedTsvRecord
{
[SaneTsv.TypedTsvColumn("column1")]
public bool Column1 { get; set; }
[SaneTsv.TypedTsvColumn]
public byte[] column2 { get; set; }
[SaneTsv.TypedTsvColumn("columnthree\nyep")]
public string Column3 { get; set; }
}
public class FloatTestRecord : SaneTsv.CommentedTsvRecord
{
[SaneTsv.TypedTsvColumn("somefloat")]
public double SomeFloat { get; set; }
[SaneTsv.TypedTsvColumn("binfloat", typeof(SaneTsv.Float64LEType))]
public double BinFloat { get; set; }
}
public class StringTestRecord : SaneTsv.TsvRecord
{
[SaneTsv.TypedTsvColumn("column1")]
public string Column1 { get; set; }
[SaneTsv.TypedTsvColumn]
public string column2 { get; set; }
[SaneTsv.TypedTsvColumn("columnthree\nyep")]
public string Column3 { get; set; }
public DateTime Column3 { get; set; }
}
private static void Main(string[] args)
{
{
string testName = "Bool test";
string testString1 = "column1:ty\\#pe:boolean\tcolumn2:binary\tcolumnthree\\nyep:string" +
"\nTRUE\tvalue\\\\t\0woo\tvaluetrhee" +
"\nFALSE\tnother\tno\\ther";
string testName = "Parse date";
string testString1 = "# ExtraTSV V0.0.1\n" +
"column1:ty\\#pe:boolean\tcolumn2:binary\tcolumnthree\\nyep:iso8601:string" +
"\nTRUE\tvalue\\\\t\0woo\t2024-02-15T18:03:30.0000" +
"\nFALSE\tnother\t2024-02-15T18:03:39.0001";
Tsv<BoolTestRecord> parsed = SaneTsv.ParseTypedTsv<BoolTestRecord>(Encoding.UTF8.GetBytes(testString1));
CommentedTsv<DateTest> parsed = SaneTsv.ParseExtraTsv<DateTest>(Encoding.UTF8.GetBytes(testString1));
if (parsed.Records[0].Column1)
{
Console.WriteLine($"Passed {testName}");
@ -146,64 +36,15 @@ internal class Program : SaneTsv
}
{
string testName = "Bad bool test";
try
{
string testString1 = "column1:type:boolean\tcolumn2:binary\tcolumnthree\\nyep:string" +
"\nTUE\tvalue\\\\t\0woo\tvaluetrhee" +
"\nFALSE\tnother\tno\\ther";
string testName = "Serde date";
string testString1 = "# ExtraTSV V0.0.1\n" +
"column1:ty\\#pe:boolean\tcolumn2:binary\tcolumnthree\\nyep:iso8601:string" +
"\nTRUE\tvalue\\\\t\0woo\t2024-02-15T18:03:30.0000" +
"\nFALSE\tnother\t2024-02-15T18:03:39.0001";
Tsv<BoolTestRecord> parsed = SaneTsv.ParseTypedTsv<BoolTestRecord>(Encoding.UTF8.GetBytes(testString1));
Console.WriteLine($"Failed {testName}");
}
catch (Exception)
{
Console.WriteLine($"Passed {testName}");
}
}
{
string testName = "Comment test";
string testString1 = "#This is a file comment\n" +
"#One more file comment line\n" +
"column1:type:boolean\tcolumn2:binary\tcolumnthree\\nyep:string" +
"\n#This is a comment" +
"\n#Another comment line" +
"\nTRUE\tvalue\\\\t\0woo\tvaluetrhee" +
"\nFALSE\tnother\tno\\ther";
CommentedTsv<BoolTestRecord2> parsed = SaneTsv.ParseCommentedTsv<BoolTestRecord2>(Encoding.UTF8.GetBytes(testString1));
}
//{
// string testName = "Serde test";
// string testString1 = "column1\tcolumn2\tcolumnthree\\nyep" +
// "\nTRUE\tvalue\\\\twoo\tvaluetrhee" +
// "\nFALSE\tnother\tno\\ther";
// Tsv<SerdeTestRecord> parsed = SaneTsv.ParseSimpleTsv<SerdeTestRecord>(Encoding.UTF8.GetBytes(testString1));
// string serialized = Encoding.UTF8.GetString(SaneTsv.SerializeSimpleTsv(parsed.ColumnNames, parsed.Records.Select(r => r.Fields.Select(f => f.ToString()).ToArray()).ToArray()));
// if (testString1 == serialized)
// {
// Console.WriteLine($"Passed {testName}");
// }
// else
// {
// Console.WriteLine($"Failed {testName}");
// }
//}
{
string testName = "Float binary test";
var bytes = new List<byte>();
bytes.AddRange(Encoding.UTF8.GetBytes("somefloat:float64\tbinfloat:float64-le" +
"\n1.5\t")); bytes.AddRange(BitConverter.GetBytes(1.5));
bytes.AddRange(Encoding.UTF8.GetBytes("\n-8.0000005E-14\t")); bytes.AddRange(BitConverter.GetBytes(-8.0000005E-14));
Tsv<FloatTestRecord> parsed = SaneTsv.ParseTypedTsv<FloatTestRecord>(bytes.ToArray());
if (parsed.Records[0].BinFloat == parsed.Records[0].SomeFloat)
CommentedTsv<DateTest> parsed = SaneTsv.ParseExtraTsv<DateTest>(Encoding.UTF8.GetBytes(testString1));
string serialized = Encoding.UTF8.GetString(SaneTsv.SerializeExtraTsv<DateTest>(parsed.Records));
if (serialized == testString1)
{
Console.WriteLine($"Passed {testName}");
}
@ -213,552 +54,6 @@ internal class Program : SaneTsv
}
}
{
string testName = "Serde test";
TestRecord[] data =
{
new TestRecord("test", true, 44.5f, 44.5f, -88e-3, -88e-3, 7773, 88888888, -7773, -88888888, new byte[] { 0, 1, 2, 3 }),
new TestRecord("test2", false, 44.5000005f, 44.5000005f, -88e-30, -88e-30, 7773, 88888888, -7773, -88888888, new byte[] { 0, 1, 2, 3, 4 }),
new TestRecord("test2", false, float.NaN, float.NaN, double.NaN, double.NaN, 7773, 88888888, -7773, -88888888, new byte[] { 0, 1, 2, 3, 4 }),
new TestRecord("test2", false, float.NegativeInfinity, float.NegativeInfinity, double.NegativeInfinity, double.NegativeInfinity, 7773, 88888888, -7773, -88888888, new byte[] { 0, 1, 2, 3, 4 }),
new TestRecord("test2", false, float.PositiveInfinity, float.PositiveInfinity, double.PositiveInfinity, double.PositiveInfinity, 7773, 88888888, -7773, -88888888, new byte[] { 0, 1, 2, 3, 4 }),
};
byte[] serialized = SaneTsv.SerializeTypedTsv(data);
Tsv<TestRecord> parsed = SaneTsv.ParseTypedTsv<TestRecord>(serialized);
if ((float)parsed.Records[1].Float32Test == 44.5000005f)
{
Console.WriteLine($"Passed {testName}");
}
else
{
Console.WriteLine($"Failed {testName}");
}
}
{
string testName = "Trying to parse a not commented record as a Commented TSV test";
// These should not compile:
//byte[] serialized = SaneTsv.SerializeCommentedTsv(data);
// Gives this error: error CS7036: There is no argument given that corresponds to the required parameter 'fileComment' of 'SaneTsv.SerializeCommentedTsv<T>(IList<T>, string)'
//Tsv<TestRecord> parsed = SaneTsv.ParseCommentedTsv<TestRecord>(serialized);
// Gives this error: error CS0311: The type 'Program.TestRecord' cannot be used as type parameter 'T' in the generic type or method 'SaneTsv.ParseCommentedTsv<T>(byte[])'. There is no implicit reference conversion from 'Program.TestRecord' to 'NathanMcRae.SaneTsv.CommentedTsvRecord'.
}
{
string testName = "Try to parsed a Commented TSV as a Simple TSV";
string testString1 = "#This is a file comment\n" +
"#One more file comment line\n" +
"column1:type:boolean\tcolumn2:binary\tcolumnthree\\nyep:string" +
"\n#This is a comment" +
"\n#Another comment line" +
"\nTRUE\tvalue\\\\t\0woo\tvaluetrhee" +
"\nFALSE\tnother\tno\\ther";
try
{
Tsv<BoolTestRecord2> parsed = SaneTsv.ParseSimpleTsv<BoolTestRecord2>(Encoding.UTF8.GetBytes(testString1));
Console.WriteLine($"Failed {testName}");
}
catch (Exception e)
{
Console.WriteLine($"Passed {testName}");
}
}
{
string testName = "Try to parsed a Commented TSV as a Typed TSV";
string testString1 = "#This is a file comment\n" +
"#One more file comment line\n" +
"column1:type:boolean\tcolumn2:binary\tcolumnthree\\nyep:string" +
"\n#This is a comment" +
"\n#Another comment line" +
"\nTRUE\tvalue\\\\t\0woo\tvaluetrhee" +
"\nFALSE\tnother\tno\\ther";
try
{
Tsv<BoolTestRecord2> parsed = SaneTsv.ParseTypedTsv<BoolTestRecord2>(Encoding.UTF8.GetBytes(testString1));
Console.WriteLine($"Failed {testName}");
}
catch (Exception e)
{
Console.WriteLine($"Passed {testName}");
}
}
{
string testName = "Try to parsed a Typed TSV as a Simple TSV";
string testString1 =
"column1:type:boolean\tcolumn2:binary\tcolumnthree\\nyep:string" +
"\nTRUE\tvalue\\\\t\0woo\tvaluetrhee" +
"\nFALSE\tnother\tno\\ther";
try
{
Tsv<BoolTestRecord2> parsed = SaneTsv.ParseSimpleTsv<BoolTestRecord2>(Encoding.UTF8.GetBytes(testString1));
Console.WriteLine($"Failed {testName}");
}
catch (Exception e)
{
Console.WriteLine($"Passed {testName}");
}
}
{
string testName = "Timing comparison of simple parse methods and comparison of simple serialization methods";
int N = 1000000;
var records = new StringTestRecord[N];
var rand = new Random(1);
for (int i = 0; i < N; i++)
{
records[i] = new StringTestRecord()
{
Column1 = rand.Next().ToString(),
column2 = rand.Next().ToString(),
Column3 = rand.Next().ToString(),
};
}
string[][] recordStrings = records.Select(record => new string[] { record.Column1, record.column2, record.Column3 }).ToArray();
DateTime lastTime = DateTime.Now;
byte[] serialized1 = SaneTsv.SerializeSimpleTsv<StringTestRecord>(records);
TimeSpan speccedSerializationTime = DateTime.Now - lastTime;
Console.WriteLine($"Specced serialization time: {speccedSerializationTime}");
lastTime = DateTime.Now;
byte[] serialized2 = SaneTsv.SerializeSimpleTsv(new string[] { "column1", "column2", "columnthree\nyep" }, recordStrings);
TimeSpan unspeccedSerializationTime = DateTime.Now - lastTime;
Console.WriteLine($"Unspecced serialization time: {unspeccedSerializationTime}");
lastTime = DateTime.Now;
Tsv<StringTestRecord> parsed = SaneTsv.ParseSimpleTsv<StringTestRecord>(serialized1);
TimeSpan speccedParseTime = DateTime.Now - lastTime;
Console.WriteLine($"Specced parse time: {speccedParseTime}");
lastTime = DateTime.Now;
(string[] columns, string[][] data) = SaneTsv.ParseSimpleTsv(serialized2);
TimeSpan unspeccedParseTime = DateTime.Now - lastTime;
Console.WriteLine($"Unspecced parse time: {unspeccedParseTime}");
}
{
string testName = "With and without file comment";
string testString1 = "#This is a file comment\n" +
"#One more file comment line\n" +
"column1:type:boolean\tcolumn2:binary\tcolumnthree\\nyep:string" +
"\n#This is a comment" +
"\n#Another comment line" +
"\nTRUE\tvalue\\\\t\0woo\tvaluetrhee" +
"\nFALSE\tnother\tno\\ther";
string testString2 = "column1:type:boolean\tcolumn2:binary\tcolumnthree\\nyep:string" +
"\n#This is a comment" +
"\n#Another comment line" +
"\nTRUE\tvalue\\\\t\0woo\tvaluetrhee" +
"\nFALSE\tnother\tno\\ther";
CommentedTsv<BoolTestRecord2> parsed = SaneTsv.ParseCommentedTsv<BoolTestRecord2>(Encoding.UTF8.GetBytes(testString1));
CommentedTsv<BoolTestRecord2> parsed2 = SaneTsv.ParseCommentedTsv<BoolTestRecord2>(Encoding.UTF8.GetBytes(testString2));
if (parsed.FileComment == "This is a file comment\nOne more file comment line" && parsed2.FileComment == null)
{
Console.WriteLine($"Passed {testName}");
}
else
{
Console.WriteLine($"Failed {testName}");
}
}
{
string testName = "With and without types";
string testString1 = "column1:type:boolean\tcolumn2:binary\tcolumnthree\\nyep:string" +
"\nTRUE\tvalue\\\\twoo\tvaluetrhee" +
"\nFALSE\tnother\tno\\ther";
try
{
Tsv<BoolTestRecord2> parsed = SaneTsv.ParseTypedTsv<BoolTestRecord2>(Encoding.UTF8.GetBytes(testString1));
Console.WriteLine($"Passed {testName} 1A");
}
catch (Exception e)
{
Console.WriteLine($"Failed {testName} 1A");
}
try
{
Tsv<BoolTestRecord2> parsed2 = SaneTsv.ParseSimpleTsv<BoolTestRecord2>(Encoding.UTF8.GetBytes(testString1));
Console.WriteLine($"Failed {testName} 1B");
}
catch (Exception e)
{
Console.WriteLine($"Passed {testName} 1B");
}
try
{
(string[] columns, string[][] data) = SaneTsv.ParseSimpleTsv(Encoding.UTF8.GetBytes(testString1));
Console.WriteLine($"Failed {testName} 1C");
}
catch (Exception e)
{
Console.WriteLine($"Passed {testName} 1C");
}
string testString2 = "column1\tcolumn2\tcolumnthree\\nyep" +
"\nTRUE\tvalue\\\\twoo\tvaluetrhee" +
"\nFALSE\tnother\tno\\ther";
try
{
Tsv<BoolTestRecord2> parsed = SaneTsv.ParseTypedTsv<BoolTestRecord2>(Encoding.UTF8.GetBytes(testString2));
Console.WriteLine($"Failed {testName} 2A");
}
catch (Exception e)
{
Console.WriteLine($"Passed {testName} 2A");
}
try
{
Tsv<BoolTestRecord2> parsed2 = SaneTsv.ParseSimpleTsv<BoolTestRecord2>(Encoding.UTF8.GetBytes(testString1));
Console.WriteLine($"Failed {testName} 2B");
}
catch (Exception e)
{
Console.WriteLine($"Passed {testName} 2B");
}
try
{
(string[] columns, string[][] data) = SaneTsv.ParseSimpleTsv(Encoding.UTF8.GetBytes(testString1));
Console.WriteLine($"Failed {testName} 2C");
}
catch (Exception e)
{
Console.WriteLine($"Passed {testName} 2C");
}
}
{
string testName = "With and without line comment";
string testString1 = "column1:type:boolean\tcolumn2:binary\tcolumnthree\\nyep:string" +
"\n#This is a comment" +
"\n#Another comment line" +
"\nTRUE\tvalue\\\\t\0woo\tvaluetrhee" +
"\nFALSE\tnother\tno\\ther";
try
{
CommentedTsv<BoolTestRecord2> parsed = SaneTsv.ParseCommentedTsv<BoolTestRecord2>(Encoding.UTF8.GetBytes(testString1));
Console.WriteLine($"Passed {testName} 1A");
}
catch (Exception e)
{
Console.WriteLine($"Failed {testName} 1A");
}
try
{
Tsv<BoolTestRecord2> parsed = SaneTsv.ParseTypedTsv<BoolTestRecord2>(Encoding.UTF8.GetBytes(testString1));
Console.WriteLine($"Failed {testName} 1B");
}
catch (Exception e)
{
Console.WriteLine($"Passed {testName} 1B");
}
try
{
Tsv<BoolTestRecord2> parsed2 = SaneTsv.ParseSimpleTsv<BoolTestRecord2>(Encoding.UTF8.GetBytes(testString1));
Console.WriteLine($"Failed {testName} 1C");
}
catch (Exception e)
{
Console.WriteLine($"Passed {testName} 1C");
}
try
{
(string[] columns, string[][] data) = SaneTsv.ParseSimpleTsv(Encoding.UTF8.GetBytes(testString1));
Console.WriteLine($"Failed {testName} 1D");
}
catch (Exception e)
{
Console.WriteLine($"Passed {testName} 1D");
}
}
{
string testName = "End of file comment";
string testString1 = "column1:type:boolean\tcolumn2:binary\tcolumnthree\\nyep:string" +
"\nTRUE\tvalue\\\\t\0woo\tvaluetrhee" +
"\nFALSE\tnother\tno\\ther" +
"\n# Hey, you're not supposed to have comments at the end of the tsv!";
try
{
CommentedTsv<BoolTestRecord2> parsed = SaneTsv.ParseCommentedTsv<BoolTestRecord2>(Encoding.UTF8.GetBytes(testString1));
Console.WriteLine($"Failed {testName} 1A");
}
catch (Exception e)
{
Console.WriteLine($"Passed {testName} 1A");
}
try
{
Tsv<BoolTestRecord2> parsed = SaneTsv.ParseTypedTsv<BoolTestRecord2>(Encoding.UTF8.GetBytes(testString1));
Console.WriteLine($"Failed {testName} 1B");
}
catch (Exception e)
{
Console.WriteLine($"Passed {testName} 1B");
}
string testString2 = "column1\tcolumn2\tcolumnthree\\nyep" +
"\nTRUE\tvalue\\\\t\0woo\tvaluetrhee" +
"\nFALSE\tnother\tno\\ther" +
"\n# Hey, you're not supposed to have comments at the end of the tsv!";
try
{
Tsv<BoolTestRecord3> parsed3 = SaneTsv.ParseSimpleTsv<BoolTestRecord3>(Encoding.UTF8.GetBytes(testString2));
Console.WriteLine($"Failed {testName} 1C");
}
catch (Exception e)
{
Console.WriteLine($"Passed {testName} 1C");
}
try
{
(string[] columns, string[][] data) = SaneTsv.ParseSimpleTsv(Encoding.UTF8.GetBytes(testString2));
Console.WriteLine($"Failed {testName} 1D");
}
catch (Exception e)
{
Console.WriteLine($"Passed {testName} 1D");
}
}
{
string testName = "Partial parsing";
string line1 = "column1\tcolumn2\tcolumnthree\\nyep";
string line2 = "\nTRUE\tvalue\\\\t\0woo\tvaluetrhee";
string line3 = "\nFALSE\tnother\tno\\ther";
byte[] inputBuffer = Encoding.UTF8.GetBytes(line1 + line2 + line3);
var headerTypes = new List<Type>();
var headerNames = new List<string>();
var headerPropertyInfos = new List<PropertyInfo>();
int columnCount = 0;
foreach (PropertyInfo property in typeof(BoolTestRecord3).GetProperties())
{
TsvColumnAttribute attribute = (TsvColumnAttribute)Attribute.GetCustomAttribute(property, typeof(TsvColumnAttribute));
if (attribute == null)
{
continue;
}
headerNames.Add(attribute.ColumnName ?? property.Name);
headerTypes.Add(attribute.ColumnType ?? GetColumnFromType(property.PropertyType));
headerPropertyInfos.Add(property);
// TODO: Check that the property type and given column type are compatible
columnCount++;
}
BoolTestRecord3[] records = SaneTsv.Parse<BoolTestRecord3>(inputBuffer,
FormatType.SIMPLE_TSV,
headerPropertyInfos.ToArray(),
headerTypes.ToArray(),
line1.Length + line2.Length + 1,
inputBuffer.Length);
if (records.Length == 0 )
{
Console.WriteLine($"Passed {testName} 1");
}
else
{
Console.WriteLine($"Failed {testName} 1");
}
BoolTestRecord3[] records2 = SaneTsv.Parse<BoolTestRecord3>(inputBuffer,
FormatType.SIMPLE_TSV,
headerPropertyInfos.ToArray(),
headerTypes.ToArray(),
line1.Length,
line1.Length + 3);
if (records2[0].Column3 == "valuetrhee")
{
Console.WriteLine($"Passed {testName} 2");
}
else
{
Console.WriteLine($"Failed {testName} 2");
}
string[][] data = SaneTsv.ParseSimpleTsv(inputBuffer, 3, line1.Length + line2.Length + 1, inputBuffer.Length);
if (data[0][1] == "nother")
{
Console.WriteLine($"Passed {testName} 3");
}
else
{
Console.WriteLine($"Failed {testName} 3");
}
string[][] data2 = SaneTsv.ParseSimpleTsv(inputBuffer, 3, line1.Length, line1.Length + 3);
if (data2.Length == 0)
{
Console.WriteLine($"Passed {testName} 4");
}
else
{
Console.WriteLine($"Failed {testName} 4");
}
}
{
string testName = "End of file \\n";
string testString1 = "column1:type:boolean\tcolumn2:binary\tcolumnthree\\nyep:string" +
"\nTRUE\tvalue\\\\t\0woo\tvaluetrhee" +
"\nFALSE\tnother\tno\\ther" +
"\n";
try
{
CommentedTsv<BoolTestRecord2> parsed = SaneTsv.ParseCommentedTsv<BoolTestRecord2>(Encoding.UTF8.GetBytes(testString1));
Console.WriteLine($"Failed {testName} 1A");
}
catch (Exception e)
{
Console.WriteLine($"Passed {testName} 1A");
}
try
{
Tsv<BoolTestRecord2> parsed = SaneTsv.ParseTypedTsv<BoolTestRecord2>(Encoding.UTF8.GetBytes(testString1));
Console.WriteLine($"Failed {testName} 1B");
}
catch (Exception e)
{
Console.WriteLine($"Passed {testName} 1B");
}
string testString2 = "column1\tcolumn2\tcolumnthree\\nyep" +
"\nTRUE\tvalue\\\\t\0woo\tvaluetrhee" +
"\nFALSE\tnother\tno\\ther" +
"\n";
try
{
Tsv<BoolTestRecord3> parsed3 = SaneTsv.ParseSimpleTsv<BoolTestRecord3>(Encoding.UTF8.GetBytes(testString2));
Console.WriteLine($"Failed {testName} 1C");
}
catch (Exception e)
{
Console.WriteLine($"Passed {testName} 1C");
}
try
{
(string[] columns, string[][] data) = SaneTsv.ParseSimpleTsv(Encoding.UTF8.GetBytes(testString2));
Console.WriteLine($"Failed {testName} 1D");
}
catch (Exception e)
{
Console.WriteLine($"Passed {testName} 1D");
}
}
{
string testName = "End of file partial record";
string testString1 = "column1:type:boolean\tcolumn2:binary\tcolumnthree\\nyep:string" +
"\nTRUE\tvalue\\\\t\0woo\tvaluetrhee" +
"\nFALSE\tnother\tno\\ther" +
"\nTRUE\t";
try
{
CommentedTsv<BoolTestRecord2> parsed = SaneTsv.ParseCommentedTsv<BoolTestRecord2>(Encoding.UTF8.GetBytes(testString1));
Console.WriteLine($"Failed {testName} 1A");
}
catch (Exception e)
{
Console.WriteLine($"Passed {testName} 1A");
}
try
{
Tsv<BoolTestRecord2> parsed = SaneTsv.ParseTypedTsv<BoolTestRecord2>(Encoding.UTF8.GetBytes(testString1));
Console.WriteLine($"Failed {testName} 1B");
}
catch (Exception e)
{
Console.WriteLine($"Passed {testName} 1B");
}
string testString2 = "column1\tcolumn2\tcolumnthree\\nyep" +
"\nTRUE\tvalue\\\\t\0woo\tvaluetrhee" +
"\nFALSE\tnother\tno\\ther" +
"\nTRUE\t";
try
{
Tsv<BoolTestRecord3> parsed3 = SaneTsv.ParseSimpleTsv<BoolTestRecord3>(Encoding.UTF8.GetBytes(testString2));
Console.WriteLine($"Failed {testName} 1C");
}
catch (Exception e)
{
Console.WriteLine($"Passed {testName} 1C");
}
try
{
(string[] columns, string[][] data) = SaneTsv.ParseSimpleTsv(Encoding.UTF8.GetBytes(testString2));
Console.WriteLine($"Failed {testName} 1D");
}
catch (Exception e)
{
Console.WriteLine($"Passed {testName} 1D");
}
}
Console.WriteLine("Done with tests");
}
}