Compare commits

..

2 Commits

Author SHA1 Message Date
Nathan McRae
0807f0ffc2 Add iso8601 parsing 2024-03-10 22:43:21 -07:00
Nathan McRae
71af3653c4 Change column type management
The column types were tracked just as a Type. This changes them to be an instance
so they can track additional information (such as the specific units of a physical
units type). Becuase of this, the column type attributes need the type to be passed as strings.
(see CS0181).
2024-03-10 22:35:30 -07:00
2 changed files with 297 additions and 838 deletions

View File

@ -1,5 +1,7 @@
using System.Reflection; using System.Globalization;
using System.Reflection;
using System.Text; using System.Text;
using System.Text.RegularExpressions;
namespace NathanMcRae; namespace NathanMcRae;
@ -32,6 +34,51 @@ public class SaneTsv
public class Int32Type : ColumnType { } public class Int32Type : ColumnType { }
public class Int64Type : ColumnType { } public class Int64Type : ColumnType { }
public class BinaryType : 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 protected enum FormatType
{ {
@ -71,6 +118,31 @@ public class SaneTsv
return (CommentedTsv<T>)Parse<T>(inputBuffer, FormatType.COMMENTED_TSV); 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($"File has major version ({fileMajorVersion}) which is newer than this parser's version {MajorVersion}");
}
return parsed;
}
protected static Tsv<T> Parse<T>(byte[] inputBuffer, FormatType format) where T : TsvRecord, new() protected static Tsv<T> Parse<T>(byte[] inputBuffer, FormatType format) where T : TsvRecord, new()
{ {
Tsv<T> parsed; Tsv<T> parsed;
@ -84,7 +156,7 @@ public class SaneTsv
} }
parsed.Records = new List<T>(); parsed.Records = new List<T>();
var columnTypes = new List<Type>(); var columnTypes = new List<ColumnType>();
var columnNames = new List<string>(); var columnNames = new List<string>();
var columnPropertyInfos = new List<PropertyInfo>(); var columnPropertyInfos = new List<PropertyInfo>();
int columnCount = 0; int columnCount = 0;
@ -169,7 +241,7 @@ public class SaneTsv
throw new Exception($"Header field {fields.Count} is not valid UTF-8", e); throw new Exception($"Header field {fields.Count} is not valid UTF-8", e);
} }
string columnTypeString; string[] columnTypeStrings;
string columnName; string columnName;
if (columnString.Contains(':')) if (columnString.Contains(':'))
{ {
@ -177,8 +249,8 @@ public class SaneTsv
{ {
throw new Exception($"Header field {j} contains ':', which is not allowed for column names"); throw new Exception($"Header field {j} contains ':', which is not allowed for column names");
} }
columnTypeString = columnString.Split(":").Last(); columnTypeStrings = columnString.Split(":");
columnName = columnString.Substring(0, columnString.Length - columnTypeString.Length - 1); columnName = string.Join(":", columnTypeStrings.Take(columnTypeStrings.Length - 1));
} }
else else
{ {
@ -186,55 +258,67 @@ public class SaneTsv
{ {
throw new Exception($"Header field {fields.Count} has no type"); throw new Exception($"Header field {fields.Count} has no type");
} }
columnTypeString = ""; columnTypeStrings = new string[] { "" };
columnName = columnString; columnName = columnString;
} }
Type type; ColumnType type;
switch (columnTypeString) switch (columnTypeStrings.Last())
{ {
case "": case "":
numTypesBlank++; numTypesBlank++;
type = typeof(StringType); type = new StringType();
break; break;
case "string": case "string":
type = typeof(StringType); if (columnTypeStrings[columnTypeStrings.Length - 2] == "iso8601")
{
type = new Iso8601Type();
columnName = string.Join(":", columnTypeStrings.Take(columnTypeStrings.Length - 2));
}
// TODO: ISO8601 time spans
// TODO: ISO8601 time durations
else
{
type = new StringType();
}
break; break;
case "boolean": case "boolean":
type = typeof(BooleanType); type = new BooleanType();
break; break;
case "float32": case "float32":
type = typeof(Float32Type); type = new Float32Type();
break; break;
case "float32-le": case "float32-le":
type = typeof(Float32LEType); type = new Float32LEType();
break; break;
case "float64": case "float64":
type = typeof(Float64Type); type = new Float64Type();
break; break;
case "float64-le": case "float64-le":
type = typeof(Float64LEType); type = new Float64LEType();
break; break;
case "uint32": case "uint32":
type = typeof(UInt32Type); type = new UInt32Type();
break; break;
case "uint64": case "uint64":
type = typeof(UInt64Type); type = new UInt64Type();
break; break;
case "int32": case "int32":
type = typeof(Int32Type); type = new Int32Type();
break; break;
case "int64": case "int64":
type = typeof(Int64Type); type = new Int64Type();
break; break;
case "binary": case "binary":
type = typeof(BinaryType); type = new BinaryType();
break; break;
default: default:
throw new Exception($"Invalid type '{columnTypeString}' for column {j}"); throw new Exception($"Invalid type '{columnTypeStrings.Last()}' for column {j}");
} }
// TODO: physical unit types
// TODO: Allow lax parsing (only worry about parsing columns that are given in the specifying type // TODO: Allow lax parsing (only worry about parsing columns that are given in the specifying type
if (columnNames[j] != columnName) if (columnNames[j] != columnName)
@ -242,7 +326,7 @@ public class SaneTsv
throw new Exception($"Column {j} has name {columnName}, but expected {columnNames[j]}"); throw new Exception($"Column {j} has name {columnName}, but expected {columnNames[j]}");
} }
if (columnTypes[j] != type) if (columnTypes[j].GetType() != type.GetType())
{ {
throw new Exception($"Column {j} has type {type}, but expected {columnTypes[j]}"); throw new Exception($"Column {j} has type {type}, but expected {columnTypes[j]}");
} }
@ -348,7 +432,7 @@ public class SaneTsv
// startIndex is in we'd have to go back to the start of the record's comment, and to know // startIndex is in we'd have to go back to the start of the record's comment, and to know
// exactly where that comment started we'd have to go back to the start of the record before that // exactly where that comment started we'd have to go back to the start of the record before that
// (not including that other record's comment). // (not including that other record's comment).
protected static T[] Parse<T>(byte[] inputBuffer, FormatType format, PropertyInfo[] columnPropertyInfos, Type[] columnTypes, int startIndex, int endIndex) where T : TsvRecord, new() protected static T[] Parse<T>(byte[] inputBuffer, FormatType format, PropertyInfo[] columnPropertyInfos, ColumnType[] columnTypes, int startIndex, int endIndex) where T : TsvRecord, new()
{ {
var fieldBytes = new List<byte>(); var fieldBytes = new List<byte>();
var fields = new List<byte[]>(); var fields = new List<byte[]>();
@ -509,12 +593,12 @@ public class SaneTsv
return parsed.ToArray(); return parsed.ToArray();
} }
protected static T ParseCurrentCommentedRecord<T>(Type[] columnTypes, PropertyInfo[] properties, List<byte[]> fields, string comment, int line) where T : CommentedTsvRecord, new() protected static T ParseCurrentCommentedRecord<T>(ColumnType[] columnTypes, PropertyInfo[] properties, List<byte[]> fields, string comment, int line) where T : CommentedTsvRecord, new()
{ {
return (T)ParseCurrentRecord<T>(columnTypes, properties, fields, comment, line); return (T)ParseCurrentRecord<T>(columnTypes, properties, fields, comment, line);
} }
protected static T ParseCurrentRecord<T>(Type[] columnTypes, PropertyInfo[] properties, List<byte[]> fields, string comment, int line) where T : TsvRecord, new() protected static T ParseCurrentRecord<T>(ColumnType[] columnTypes, PropertyInfo[] properties, List<byte[]> fields, string comment, int line) where T : TsvRecord, new()
{ {
T record = new T(); T record = new T();
@ -532,7 +616,7 @@ public class SaneTsv
for (int j = 0; j < fields.Count; j++) for (int j = 0; j < fields.Count; j++)
{ {
// All other types require the content to be UTF-8. Binary fields can ignore that. // All other types require the content to be UTF-8. Binary fields can ignore that.
if (columnTypes[j] == typeof(BinaryType)) if (columnTypes[j].GetType() == typeof(BinaryType))
{ {
// TODO: Use faster method for property setting // TODO: Use faster method for property setting
// e.g. https://blog.marcgravell.com/2012/01/playing-with-your-member.html // e.g. https://blog.marcgravell.com/2012/01/playing-with-your-member.html
@ -541,7 +625,7 @@ public class SaneTsv
properties[j].SetValue(record, fields[j]); properties[j].SetValue(record, fields[j]);
continue; continue;
} }
else if (columnTypes[j] == typeof(Float32LEType)) else if (columnTypes[j].GetType() == typeof(Float32LEType))
{ {
byte[] floatBytes; byte[] floatBytes;
if (!LittleEndian) if (!LittleEndian)
@ -560,7 +644,7 @@ public class SaneTsv
continue; continue;
} }
else if (columnTypes[j] == typeof(Float64LEType)) else if (columnTypes[j].GetType() == typeof(Float64LEType))
{ {
byte[] floatBytes; byte[] floatBytes;
if (!LittleEndian) if (!LittleEndian)
@ -592,11 +676,11 @@ public class SaneTsv
// TODO: Add checking for numeric types format // TODO: Add checking for numeric types format
if (columnTypes[j] == typeof(StringType)) if (columnTypes[j].GetType() == typeof(StringType))
{ {
properties[j].SetValue(record, fieldString); properties[j].SetValue(record, fieldString);
} }
else if (columnTypes[j] == typeof(BooleanType)) else if (columnTypes[j].GetType() == typeof(BooleanType))
{ {
bool parsedBool; bool parsedBool;
if (fieldString == "TRUE") if (fieldString == "TRUE")
@ -614,7 +698,7 @@ public class SaneTsv
properties[j].SetValue(record, parsedBool); properties[j].SetValue(record, parsedBool);
} }
else if (columnTypes[j] == typeof(Float32Type)) else if (columnTypes[j].GetType() == typeof(Float32Type))
{ {
float parsedFloat; float parsedFloat;
if (!float.TryParse(fieldString, out parsedFloat)) if (!float.TryParse(fieldString, out parsedFloat))
@ -635,7 +719,7 @@ public class SaneTsv
properties[j].SetValue(record, parsedFloat); properties[j].SetValue(record, parsedFloat);
} }
else if (columnTypes[j] == typeof(Float64Type)) else if (columnTypes[j].GetType() == typeof(Float64Type))
{ {
double parsedDouble; double parsedDouble;
if (!double.TryParse(fieldString, out parsedDouble)) if (!double.TryParse(fieldString, out parsedDouble))
@ -656,7 +740,7 @@ public class SaneTsv
properties[j].SetValue(record, parsedDouble); properties[j].SetValue(record, parsedDouble);
} }
else if (columnTypes[j] == typeof(UInt32Type)) else if (columnTypes[j].GetType() == typeof(UInt32Type))
{ {
if (!UInt32.TryParse(fieldString, out UInt32 parsedUInt32)) if (!UInt32.TryParse(fieldString, out UInt32 parsedUInt32))
{ {
@ -665,7 +749,7 @@ public class SaneTsv
properties[j].SetValue(record, parsedUInt32); properties[j].SetValue(record, parsedUInt32);
} }
else if (columnTypes[j] == typeof(UInt64Type)) else if (columnTypes[j].GetType() == typeof(UInt64Type))
{ {
if (!UInt64.TryParse(fieldString, out UInt64 parsedUInt64)) if (!UInt64.TryParse(fieldString, out UInt64 parsedUInt64))
{ {
@ -674,7 +758,7 @@ public class SaneTsv
properties[j].SetValue(record, parsedUInt64); properties[j].SetValue(record, parsedUInt64);
} }
else if (columnTypes[j] == typeof(Int32Type)) else if (columnTypes[j].GetType() == typeof(Int32Type))
{ {
if (!Int32.TryParse(fieldString, out Int32 parsedInt32)) if (!Int32.TryParse(fieldString, out Int32 parsedInt32))
{ {
@ -683,7 +767,7 @@ public class SaneTsv
properties[j].SetValue(record, parsedInt32); properties[j].SetValue(record, parsedInt32);
} }
else if (columnTypes[j] == typeof(Int64Type)) else if (columnTypes[j].GetType() == typeof(Int64Type))
{ {
if (!Int64.TryParse(fieldString, out Int64 parsedInt64)) if (!Int64.TryParse(fieldString, out Int64 parsedInt64))
{ {
@ -692,6 +776,15 @@ public class SaneTsv
properties[j].SetValue(record, parsedInt64); 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 else
{ {
throw new Exception($"Unexpected type {columnTypes[j]}"); throw new Exception($"Unexpected type {columnTypes[j]}");
@ -1066,99 +1159,159 @@ public class SaneTsv
return records.ToArray(); return records.ToArray();
} }
public static Type GetColumnFromType(Type type) public static ColumnType GetColumnFromString(string type)
{ {
if (type == typeof(string)) if (type == "string")
{ {
return typeof(StringType); return new StringType();
} }
else if (type == typeof(bool)) else if (type == "boolean")
{ {
return typeof(BooleanType); return new BooleanType();
} }
else if (type == typeof(float)) else if (type == "float32")
{ {
return typeof(Float32Type); return new Float32Type();
} }
else if (type == typeof(double)) else if (type == "float32-le")
{ {
return typeof(Float64Type); return new Float32LEType();
} }
else if (type == typeof(UInt32)) else if (type == "float64")
{ {
return typeof(UInt32Type); return new Float64Type();
} }
else if (type == typeof(UInt64)) else if (type == "float64-le")
{ {
return typeof(UInt64Type); return new Float64LEType();
} }
else if (type == typeof(Int32)) else if (type == "uint32")
{ {
return typeof(Int32Type); return new UInt32Type();
} }
else if (type == typeof(Int64)) else if (type == "uint64")
{ {
return typeof(Int64Type); return new UInt64Type();
} }
else if (type == typeof(byte[])) else if (type == "int32")
{ {
return typeof(BinaryType); return new Int32Type();
}
else if (type == "int64")
{
return new Int64Type();
}
else if (type == "binary")
{
return new BinaryType();
} }
else else
{ {
throw new Exception($"Invalid type: {type}"); throw new Exception($"Invalid type: {type.GetType()}");
} }
} }
public static string GetNameFromColumn(Type type) public static ColumnType GetColumnFromType(Type type)
{ {
if (type == typeof(StringType)) if (type == typeof(string))
{ {
return "string"; return new StringType();
} }
else if (type == typeof(BooleanType)) else if (type == typeof(bool))
{ {
return "boolean"; return new BooleanType();
} }
else if (type == typeof(Float32Type)) else if (type == typeof(float))
{ {
return "float32"; return new Float32Type();
} }
else if (type == typeof(Float32LEType)) else if (type == typeof(double))
{ {
return "float32-le"; return new Float64Type();
} }
else if (type == typeof(Float64Type)) else if (type == typeof(UInt32))
{ {
return "float64"; return new UInt32Type();
} }
else if (type == typeof(Float64LEType)) else if (type == typeof(UInt64))
{ {
return "float64-le"; return new UInt64Type();
} }
else if (type == typeof(UInt32Type)) else if (type == typeof(Int32))
{ {
return "uint32"; return new Int32Type();
} }
else if (type == typeof(UInt64Type)) else if (type == typeof(Int64))
{ {
return "uint64"; return new Int64Type();
} }
else if (type == typeof(Int32Type)) else if (type == typeof(byte[]))
{ {
return "int32"; return new BinaryType();
} }
else if (type == typeof(Int64Type)) else if (type == typeof(DateTime))
{ {
return "int64"; return new Iso8601Type();
}
else if (type == typeof(BinaryType))
{
return "binary";
} }
else else
{ {
throw new Exception($"Invalid type: {type}"); throw new Exception($"Invalid type: {type.GetType()}");
}
}
public static string GetNameFromColumn(ColumnType type)
{
if (type.GetType() == typeof(StringType))
{
return "string";
}
else if (type.GetType() == typeof(BooleanType))
{
return "boolean";
}
else if (type.GetType() == typeof(Float32Type))
{
return "float32";
}
else if (type.GetType() == typeof(Float32LEType))
{
return "float32-le";
}
else if (type.GetType() == typeof(Float64Type))
{
return "float64";
}
else if (type.GetType() == typeof(Float64LEType))
{
return "float64-le";
}
else if (type.GetType() == typeof(UInt32Type))
{
return "uint32";
}
else if (type.GetType() == typeof(UInt64Type))
{
return "uint64";
}
else if (type.GetType() == typeof(Int32Type))
{
return "int32";
}
else if (type.GetType() == typeof(Int64Type))
{
return "int64";
}
else if (type.GetType() == typeof(BinaryType))
{
return "binary";
}
else if (type.GetType() == typeof(Iso8601Type))
{
return "iso8601:string";
}
else
{
throw new Exception($"Invalid type: {type.GetType()}");
} }
} }
@ -1174,14 +1327,29 @@ public class SaneTsv
public static byte[] SerializeCommentedTsv<T>(IList<T> data, string fileComment) where T : CommentedTsvRecord public static byte[] SerializeCommentedTsv<T>(IList<T> data, string fileComment) where T : CommentedTsvRecord
{ {
return SerializeTsv<T>(data, FormatType.COMMENTED_TSV); return SerializeTsv<T>(data, FormatType.COMMENTED_TSV, fileComment);
} }
protected static byte[] SerializeTsv<T>(IList<T> data, FormatType tsvFormat) public static byte[] SerializeExtraTsv<T>(IList<T> data) where T : TsvRecord
{
return SerializeTsv<T>(data, FormatType.COMMENTED_TSV, $" ExtraTSV V{MajorVersion}.{MinorVersion}.{PatchVersion}");
}
protected static byte[] SerializeTsv<T>(IList<T> data, FormatType tsvFormat, string fileComment = null)
{ {
var bytes = new List<byte>(); var bytes = new List<byte>();
var columnTypes = new List<Type>(); if (fileComment != null)
{
if (tsvFormat != FormatType.COMMENTED_TSV)
{
throw new Exception($"File comments are not valid for {tsvFormat}");
}
bytes.AddRange(Encoding.UTF8.GetBytes("#" + fileComment.Replace("\n", "\n#") + "\n"));
}
var columnTypes = new List<ColumnType>();
var columnNames = new List<string>(); var columnNames = new List<string>();
var columnPropertyInfos = new List<PropertyInfo>(); var columnPropertyInfos = new List<PropertyInfo>();
int columnCount = 0; int columnCount = 0;
@ -1197,8 +1365,8 @@ public class SaneTsv
string headerName = attribute.ColumnName ?? property.Name; string headerName = attribute.ColumnName ?? property.Name;
columnNames.Add(headerName); columnNames.Add(headerName);
Type headerType = attribute.ColumnType ?? GetColumnFromType(property.PropertyType); ColumnType headerType = attribute.ColumnType ?? GetColumnFromType(property.PropertyType);
if (tsvFormat == FormatType.SIMPLE_TSV && headerType != typeof(StringType)) if (tsvFormat == FormatType.SIMPLE_TSV && headerType.GetType() != typeof(StringType))
{ {
throw new Exception($"Serializing Simple TSV requires all columns be of type string, but column '{headerName}' has type '{headerType}'"); throw new Exception($"Serializing Simple TSV requires all columns be of type string, but column '{headerName}' has type '{headerType}'");
} }
@ -1278,7 +1446,7 @@ public class SaneTsv
return bytes.ToArray(); return bytes.ToArray();
} }
protected static void SerializeTsv<T>(IList<T> data, List<byte> bytes, PropertyInfo[] columnPropertyInfos, Type[] columnTypes, FormatType tsvFormat, int startIndex, int endIndex) protected static void SerializeTsv<T>(IList<T> data, List<byte> bytes, PropertyInfo[] columnPropertyInfos, ColumnType[] columnTypes, FormatType tsvFormat, int startIndex, int endIndex)
{ {
// Serialize data // Serialize data
for (int i = 0; i < data.Count; i++) for (int i = 0; i < data.Count; i++)
@ -1293,16 +1461,16 @@ public class SaneTsv
// Some fields definitely don't need escaping, so we add them directly to bytes // Some fields definitely don't need escaping, so we add them directly to bytes
bool skipEscaping = false; bool skipEscaping = false;
if (columnTypes[j] == typeof(StringType)) if (columnTypes[j].GetType() == typeof(StringType))
{ {
fieldEncoded = Encoding.UTF8.GetBytes((string)datum); fieldEncoded = Encoding.UTF8.GetBytes((string)datum);
} }
else if (columnTypes[j] == typeof(BooleanType)) else if (columnTypes[j].GetType() == typeof(BooleanType))
{ {
bytes.AddRange((bool)datum ? TrueEncoded : FalseEncoded); bytes.AddRange((bool)datum ? TrueEncoded : FalseEncoded);
skipEscaping = true; skipEscaping = true;
} }
else if (columnTypes[j] == typeof(Float32Type)) else if (columnTypes[j].GetType() == typeof(Float32Type))
{ {
if (datum is float f) if (datum is float f)
{ {
@ -1326,7 +1494,7 @@ public class SaneTsv
} }
skipEscaping = true; skipEscaping = true;
} }
else if (columnTypes[j] == typeof(Float32LEType)) else if (columnTypes[j].GetType() == typeof(Float32LEType))
{ {
if (LittleEndian) if (LittleEndian)
{ {
@ -1342,7 +1510,7 @@ public class SaneTsv
} }
} }
} }
else if (columnTypes[j] == typeof(Float64Type)) else if (columnTypes[j].GetType() == typeof(Float64Type))
{ {
if (datum is double d) if (datum is double d)
{ {
@ -1366,7 +1534,7 @@ public class SaneTsv
} }
skipEscaping = true; skipEscaping = true;
} }
else if (columnTypes[j] == typeof(Float64LEType)) else if (columnTypes[j].GetType() == typeof(Float64LEType))
{ {
if (LittleEndian) if (LittleEndian)
{ {
@ -1382,30 +1550,34 @@ public class SaneTsv
} }
} }
} }
else if (columnTypes[j] == typeof(UInt32Type)) else if (columnTypes[j].GetType() == typeof(UInt32Type))
{ {
bytes.AddRange(Encoding.UTF8.GetBytes(((UInt32)datum).ToString())); bytes.AddRange(Encoding.UTF8.GetBytes(((UInt32)datum).ToString()));
skipEscaping = true; skipEscaping = true;
} }
else if (columnTypes[j] == typeof(UInt64Type)) else if (columnTypes[j].GetType() == typeof(UInt64Type))
{ {
bytes.AddRange(Encoding.UTF8.GetBytes(((UInt64)datum).ToString())); bytes.AddRange(Encoding.UTF8.GetBytes(((UInt64)datum).ToString()));
skipEscaping = true; skipEscaping = true;
} }
else if (columnTypes[j] == typeof(Int32Type)) else if (columnTypes[j].GetType() == typeof(Int32Type))
{ {
bytes.AddRange(Encoding.UTF8.GetBytes(((Int32)datum).ToString())); bytes.AddRange(Encoding.UTF8.GetBytes(((Int32)datum).ToString()));
skipEscaping = true; skipEscaping = true;
} }
else if (columnTypes[j] == typeof(Int64Type)) else if (columnTypes[j].GetType() == typeof(Int64Type))
{ {
bytes.AddRange(Encoding.UTF8.GetBytes(((Int64)datum).ToString())); bytes.AddRange(Encoding.UTF8.GetBytes(((Int64)datum).ToString()));
skipEscaping = true; skipEscaping = true;
} }
else if (columnTypes[j] == typeof(BinaryType)) else if (columnTypes[j].GetType() == typeof(BinaryType))
{ {
fieldEncoded = (byte[])datum; fieldEncoded = (byte[])datum;
} }
else if (columnTypes[j].GetType() == typeof(Iso8601Type))
{
fieldEncoded = Encoding.UTF8.GetBytes(((DateTime)datum).ToString("yyyy-MM-ddTHH:mm:ss.ffff"));
}
else else
{ {
throw new Exception($"Unexpected column type {columnTypes[j]} for column {j}"); throw new Exception($"Unexpected column type {columnTypes[j]} for column {j}");
@ -1512,15 +1684,15 @@ public class SaneTsv
public class TsvColumnAttribute : Attribute public class TsvColumnAttribute : Attribute
{ {
public string ColumnName { get; } public string ColumnName { get; }
public virtual Type ColumnType { get; } public virtual ColumnType ColumnType { get; }
public TsvColumnAttribute() public TsvColumnAttribute()
{ {
ColumnType = typeof(StringType); ColumnType = new StringType();
} }
public TsvColumnAttribute(string columnName) public TsvColumnAttribute(string columnName)
{ {
ColumnType = typeof(StringType); ColumnType = new StringType();
ColumnName = columnName; ColumnName = columnName;
} }
} }
@ -1528,27 +1700,19 @@ public class SaneTsv
// TODO: Add column ordering // TODO: Add column ordering
public class TypedTsvColumnAttribute : TsvColumnAttribute public class TypedTsvColumnAttribute : TsvColumnAttribute
{ {
public override Type ColumnType { get; } public override ColumnType ColumnType { get; }
public TypedTsvColumnAttribute() { } public TypedTsvColumnAttribute() { }
public TypedTsvColumnAttribute(string columnName) : base(columnName) { } public TypedTsvColumnAttribute(string columnName) : base(columnName) { }
public TypedTsvColumnAttribute(string columnName, Type columnType) : base(columnName) public TypedTsvColumnAttribute(string columnName, string columnType) : base(columnName)
{ {
if (columnType.BaseType != typeof(ColumnType)) ColumnType = GetColumnFromString(columnType);
{
throw new Exception("Column type must inherit from SaneTsv.ColumnType");
}
ColumnType = columnType;
} }
public TypedTsvColumnAttribute(Type columnType) public TypedTsvColumnAttribute(ColumnType columnType)
{ {
if (columnType.BaseType != typeof(ColumnType))
{
throw new Exception("Column type must inherit from SaneTsv.ColumnType");
}
ColumnType = columnType; ColumnType = columnType;
} }
} }

View File

@ -1,63 +1,9 @@
using NathanMcRae; using NathanMcRae;
using System.Reflection; using System.Reflection;
using System.Text; using System.Text;
internal class Program : SaneTsv internal class Program : SaneTsv
{ {
public class TestRecord : SaneTsv.TsvRecord public class DateTest : SaneTsv.CommentedTsvRecord
{
[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
{ {
[SaneTsv.TypedTsvColumn("column1:ty#pe")] [SaneTsv.TypedTsvColumn("column1:ty#pe")]
public bool Column1 { get; set; } public bool Column1 { get; set; }
@ -66,75 +12,19 @@ internal class Program : SaneTsv
public byte[] column2 { get; set; } public byte[] column2 { get; set; }
[SaneTsv.TypedTsvColumn("columnthree\nyep")] [SaneTsv.TypedTsvColumn("columnthree\nyep")]
public string Column3 { get; set; } public DateTime 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; }
} }
private static void Main(string[] args) private static void Main(string[] args)
{ {
{ {
string testName = "Bool test"; string testName = "Parse date";
string testString1 = "column1:ty\\#pe:boolean\tcolumn2:binary\tcolumnthree\\nyep:string" + string testString1 = "# ExtraTSV V0.0.1\n" +
"\nTRUE\tvalue\\\\t\0woo\tvaluetrhee" + "column1:ty\\#pe:boolean\tcolumn2:binary\tcolumnthree\\nyep:iso8601:string" +
"\nFALSE\tnother\tno\\ther"; "\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) if (parsed.Records[0].Column1)
{ {
Console.WriteLine($"Passed {testName}"); Console.WriteLine($"Passed {testName}");
@ -146,64 +36,15 @@ internal class Program : SaneTsv
} }
{ {
string testName = "Bad bool test"; string testName = "Serde date";
try string testString1 = "# ExtraTSV V0.0.1\n" +
{ "column1:ty\\#pe:boolean\tcolumn2:binary\tcolumnthree\\nyep:iso8601:string" +
string testString1 = "column1:type:boolean\tcolumn2:binary\tcolumnthree\\nyep:string" + "\nTRUE\tvalue\\\\t\0woo\t2024-02-15T18:03:30.0000" +
"\nTUE\tvalue\\\\t\0woo\tvaluetrhee" + "\nFALSE\tnother\t2024-02-15T18:03:39.0001";
"\nFALSE\tnother\tno\\ther";
Tsv<BoolTestRecord> parsed = SaneTsv.ParseTypedTsv<BoolTestRecord>(Encoding.UTF8.GetBytes(testString1)); CommentedTsv<DateTest> parsed = SaneTsv.ParseExtraTsv<DateTest>(Encoding.UTF8.GetBytes(testString1));
Console.WriteLine($"Failed {testName}"); string serialized = Encoding.UTF8.GetString(SaneTsv.SerializeExtraTsv<DateTest>(parsed.Records));
} if (serialized == testString1)
catch (Exception)
{
Console.WriteLine($"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)
{ {
Console.WriteLine($"Passed {testName}"); Console.WriteLine($"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"); Console.WriteLine("Done with tests");
} }
} }