Compare commits

..

7 Commits

Author SHA1 Message Date
Nathan McRae
0fd092685d Remove unused SaneTsv fields 2024-03-10 12:55:21 -07:00
Nathan McRae
55fa00a6e7 Clarify terminology of 'header'/'column'
The header is just the region containing the column name / types.
2024-03-09 13:13:41 -08:00
Nathan McRae
d428af51bb Flesh out tests 2024-03-09 09:58:25 -08:00
Nathan McRae
aef92e87d4 Update some comments 2024-03-09 09:58:09 -08:00
Nathan McRae
b56236cbb7 Fix some end-of-file simple TSV parsing 2024-03-09 09:57:56 -08:00
Nathan McRae
7230f982ac Fix some minor line/column numbering issues 2024-03-09 09:57:27 -08:00
Nathan McRae
f4145bacd2 Fix column attribute bug 2024-03-09 09:57:00 -08:00
2 changed files with 484 additions and 69 deletions

View File

@ -43,9 +43,6 @@ public class SaneTsv
public static readonly byte[] TrueEncoded = Encoding.UTF8.GetBytes("TRUE"); public static readonly byte[] TrueEncoded = Encoding.UTF8.GetBytes("TRUE");
public static readonly byte[] FalseEncoded = Encoding.UTF8.GetBytes("FALSE"); public static readonly byte[] FalseEncoded = Encoding.UTF8.GetBytes("FALSE");
// TODO: We need to be able to update all these in tandem somehow
public string[] ColumnNames { get; protected set; }
public Type[] ColumnTypes { get; protected set; }
protected static bool? _littleEndian = null; protected static bool? _littleEndian = null;
public static bool LittleEndian public static bool LittleEndian
{ {
@ -71,7 +68,6 @@ public class SaneTsv
public static CommentedTsv<T> ParseCommentedTsv<T>(byte[] inputBuffer) where T : CommentedTsvRecord, new() public static CommentedTsv<T> ParseCommentedTsv<T>(byte[] inputBuffer) where T : CommentedTsvRecord, new()
{ {
// TODO: add the file comment?
return (CommentedTsv<T>)Parse<T>(inputBuffer, FormatType.COMMENTED_TSV); return (CommentedTsv<T>)Parse<T>(inputBuffer, FormatType.COMMENTED_TSV);
} }
@ -88,22 +84,22 @@ public class SaneTsv
} }
parsed.Records = new List<T>(); parsed.Records = new List<T>();
var headerTypes = new List<Type>(); var columnTypes = new List<Type>();
var headerNames = new List<string>(); var columnNames = new List<string>();
var headerPropertyInfos = new List<PropertyInfo>(); var columnPropertyInfos = new List<PropertyInfo>();
int columnCount = 0; int columnCount = 0;
foreach (PropertyInfo property in typeof(T).GetProperties()) foreach (PropertyInfo property in typeof(T).GetProperties())
{ {
TypedTsvColumnAttribute attribute = (TypedTsvColumnAttribute)Attribute.GetCustomAttribute(property, typeof(TypedTsvColumnAttribute)); TsvColumnAttribute attribute = (TsvColumnAttribute)Attribute.GetCustomAttribute(property, typeof(TsvColumnAttribute));
if (attribute == null) if (attribute == null)
{ {
continue; continue;
} }
headerNames.Add(attribute.ColumnName ?? property.Name); columnNames.Add(attribute.ColumnName ?? property.Name);
headerTypes.Add(attribute.ColumnType ?? GetColumnFromType(property.PropertyType)); columnTypes.Add(attribute.ColumnType ?? GetColumnFromType(property.PropertyType));
headerPropertyInfos.Add(property); columnPropertyInfos.Add(property);
// TODO: Check that the property type and given column type are compatible // TODO: Check that the property type and given column type are compatible
columnCount++; columnCount++;
} }
@ -170,7 +166,7 @@ public class SaneTsv
} }
catch (Exception e) catch (Exception e)
{ {
throw new Exception($"Header {fields.Count} is not valid UTF-8", e); throw new Exception($"Header field {fields.Count} is not valid UTF-8", e);
} }
string columnTypeString; string columnTypeString;
@ -179,7 +175,7 @@ public class SaneTsv
{ {
if (format == FormatType.SIMPLE_TSV) if (format == FormatType.SIMPLE_TSV)
{ {
throw new Exception($"Header {fields.Count} contain ':', which is not allowed for column names"); throw new Exception($"Header field {j} contains ':', which is not allowed for column names");
} }
columnTypeString = columnString.Split(":").Last(); columnTypeString = columnString.Split(":").Last();
columnName = columnString.Substring(0, columnString.Length - columnTypeString.Length - 1); columnName = columnString.Substring(0, columnString.Length - columnTypeString.Length - 1);
@ -188,7 +184,7 @@ public class SaneTsv
{ {
if (format > FormatType.SIMPLE_TSV) if (format > FormatType.SIMPLE_TSV)
{ {
throw new Exception($"Header {fields.Count} has no type"); throw new Exception($"Header field {fields.Count} has no type");
} }
columnTypeString = ""; columnTypeString = "";
columnName = columnString; columnName = columnString;
@ -241,14 +237,14 @@ public class SaneTsv
// 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 (headerNames[j] != columnName) if (columnNames[j] != columnName)
{ {
throw new Exception($"Column {j} has name {columnName}, but expected {headerNames[j]}"); throw new Exception($"Column {j} has name {columnName}, but expected {columnNames[j]}");
} }
if (headerTypes[j] != type) if (columnTypes[j] != type)
{ {
throw new Exception($"Column {j} has type {type}, but expected {headerTypes[j]}"); throw new Exception($"Column {j} has type {type}, but expected {columnTypes[j]}");
} }
} }
@ -312,7 +308,7 @@ public class SaneTsv
// Complication: it probably depends on processor count // Complication: it probably depends on processor count
if (inputBuffer.Length < 10000) if (inputBuffer.Length < 10000)
{ {
parsed.Records.AddRange(Parse<T>(inputBuffer, format, headerPropertyInfos.ToArray(), headerTypes.ToArray(), currentLineStart - 1, inputBuffer.Length)); parsed.Records.AddRange(Parse<T>(inputBuffer, format, columnPropertyInfos.ToArray(), columnTypes.ToArray(), currentLineStart - 1, inputBuffer.Length));
return parsed; return parsed;
} }
else else
@ -334,9 +330,10 @@ public class SaneTsv
endIndex = (i + 1) * splitCount + parseStart; endIndex = (i + 1) * splitCount + parseStart;
} }
parsedValues[i] = Parse<T>(inputBuffer, format, headerPropertyInfos.ToArray(), headerTypes.ToArray(), startIndex, endIndex); parsedValues[i] = Parse<T>(inputBuffer, format, columnPropertyInfos.ToArray(), columnTypes.ToArray(), startIndex, endIndex);
}); });
// TODO: Handle relative line numbers
for (int i = 0; i < tasks; i++) for (int i = 0; i < tasks; i++)
{ {
parsed.Records.AddRange(parsedValues[i]); parsed.Records.AddRange(parsedValues[i]);
@ -351,7 +348,7 @@ 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[] headerPropertyInfos, Type[] headerTypes, int startIndex, int endIndex) where T : TsvRecord, new() protected static T[] Parse<T>(byte[] inputBuffer, FormatType format, PropertyInfo[] columnPropertyInfos, Type[] 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[]>();
@ -421,9 +418,9 @@ public class SaneTsv
fields.Add(fieldBytes.ToArray()); fields.Add(fieldBytes.ToArray());
fieldBytes.Clear(); fieldBytes.Clear();
if (headerTypes.Length != fields.Count) if (columnTypes.Length != fields.Count)
{ {
throw new Exception($"Expected {headerTypes.Length} fields on line {relativeLine}, but found {fields.Count}"); throw new Exception($"Expected {columnTypes.Length} fields on line {relativeLine}, but found {fields.Count}");
} }
else else
{ {
@ -433,7 +430,7 @@ public class SaneTsv
comment = currentComment.ToString(); comment = currentComment.ToString();
currentComment.Clear(); currentComment.Clear();
} }
parsed.Add(ParseCurrentRecord<T>(headerTypes.ToArray(), headerPropertyInfos.ToArray(), fields, comment, relativeLine)); parsed.Add(ParseCurrentRecord<T>(columnTypes.ToArray(), columnPropertyInfos.ToArray(), fields, comment, relativeLine));
fields.Clear(); fields.Clear();
} }
@ -493,9 +490,9 @@ public class SaneTsv
// TODO // TODO
throw new Exception("Not sure when this will happen. THis might actuall be fine"); throw new Exception("Not sure when this will happen. THis might actuall be fine");
} }
if (fields.Count != headerTypes.Length) if (fields.Count != columnTypes.Length)
{ {
throw new Exception($"Expected {headerTypes} fields on line {relativeLine}, but found {fields.Count}"); throw new Exception($"Expected {columnTypes} fields on line {relativeLine}, but found {fields.Count}");
} }
else else
{ {
@ -505,7 +502,7 @@ public class SaneTsv
comment = currentComment.ToString(); comment = currentComment.ToString();
currentComment.Clear(); currentComment.Clear();
} }
parsed.Add(ParseCurrentRecord<T>(headerTypes.ToArray(), headerPropertyInfos.ToArray(), fields, comment, relativeLine)); parsed.Add(ParseCurrentRecord<T>(columnTypes.ToArray(), columnPropertyInfos.ToArray(), fields, comment, relativeLine));
fields.Clear(); fields.Clear();
} }
@ -846,7 +843,7 @@ public class SaneTsv
public static (string[] columns, string[][] data) ParseSimpleTsv(byte[] inputBuffer) public static (string[] columns, string[][] data) ParseSimpleTsv(byte[] inputBuffer)
{ {
string[] columnNames = null; string[] columnNames = null;
var headers = new List<byte[]>(); var headerFields = new List<byte[]>();
var fieldBytes = new List<byte>(); var fieldBytes = new List<byte>();
int startOfData = -1; int startOfData = -1;
for (int i = 0; i < inputBuffer.Count(); i++) for (int i = 0; i < inputBuffer.Count(); i++)
@ -885,33 +882,33 @@ public class SaneTsv
else if (inputBuffer[i] == '\t') else if (inputBuffer[i] == '\t')
{ {
// end of field // end of field
headers.Add(fieldBytes.ToArray()); headerFields.Add(fieldBytes.ToArray());
fieldBytes.Clear(); fieldBytes.Clear();
} }
else if (inputBuffer[i] == '\n') else if (inputBuffer[i] == '\n')
{ {
// This is the end of the header // This is the end of the header
headers.Add(fieldBytes.ToArray()); headerFields.Add(fieldBytes.ToArray());
startOfData = i + 1; startOfData = i + 1;
columnNames = new string[headers.Count]; columnNames = new string[headerFields.Count];
fieldBytes.Clear(); fieldBytes.Clear();
for (int j = 0; j < headers.Count; j++) for (int j = 0; j < headerFields.Count; j++)
{ {
string columnString; string columnString;
try try
{ {
columnString = Encoding.UTF8.GetString(headers[j]); columnString = Encoding.UTF8.GetString(headerFields[j]);
} }
catch (Exception e) catch (Exception e)
{ {
throw new Exception($"Column {headers.Count} name is not valid UTF-8", e); throw new Exception($"Column {headerFields.Count} name is not valid UTF-8", e);
} }
if (columnString.Contains(':')) if (columnString.Contains(':'))
{ {
throw new Exception($"Header {headers.Count} contain ':', which is not allowed for column names"); throw new Exception($"Header field {headerFields.Count} contain ':', which is not allowed for column names");
} }
columnNames[j] = columnString; columnNames[j] = columnString;
@ -939,7 +936,7 @@ public class SaneTsv
var fields = new List<byte[]>(); var fields = new List<byte[]>();
var records = new List<string[]>(); var records = new List<string[]>();
int line = 1; int line = 2;
int currentLineStart = 0; int currentLineStart = 0;
// Go back to the start of the current line // Go back to the start of the current line
@ -1033,13 +1030,20 @@ public class SaneTsv
fields.Add(fieldBytes.ToArray()); fields.Add(fieldBytes.ToArray());
if (fields.Count == 0) if (fields.Count == 0 && endIndex == inputBuffer.Length)
{ {
throw new Exception("Found 0 fields on last line. Possibly because of extra \\n after last record"); throw new Exception("Found 0 fields on last line. Possibly because of extra \\n after last record");
} }
if (numFields != fields.Count) if (numFields != fields.Count)
{ {
throw new Exception($"Expected {numFields} fields on line {line}, but found {fields.Count}"); if (endIndex == inputBuffer.Length)
{
throw new Exception($"Expected {numFields} fields on line {line}, but found {fields.Count}");
}
else
{
return records.ToArray();
}
} }
else else
{ {
@ -1177,9 +1181,9 @@ public class SaneTsv
{ {
var bytes = new List<byte>(); var bytes = new List<byte>();
var headerTypes = new List<Type>(); var columnTypes = new List<Type>();
var headerNames = new List<string>(); var columnNames = new List<string>();
var headerPropertyInfos = new List<PropertyInfo>(); var columnPropertyInfos = new List<PropertyInfo>();
int columnCount = 0; int columnCount = 0;
// Serialize header // Serialize header
@ -1192,30 +1196,30 @@ public class SaneTsv
} }
string headerName = attribute.ColumnName ?? property.Name; string headerName = attribute.ColumnName ?? property.Name;
headerNames.Add(headerName); columnNames.Add(headerName);
Type headerType = attribute.ColumnType ?? GetColumnFromType(property.PropertyType); Type headerType = attribute.ColumnType ?? GetColumnFromType(property.PropertyType);
if (tsvFormat == FormatType.SIMPLE_TSV && headerType != typeof(StringType)) if (tsvFormat == FormatType.SIMPLE_TSV && headerType != 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}'");
} }
headerTypes.Add(headerType); columnTypes.Add(headerType);
headerPropertyInfos.Add(property); columnPropertyInfos.Add(property);
// TODO: Check that the property type and given column type are compatible // TODO: Check that the property type and given column type are compatible
columnCount++; columnCount++;
} }
// Serialize header // Serialize header
for (int i = 0; i < headerNames.Count; i++) for (int i = 0; i < columnNames.Count; i++)
{ {
for (int j = i + 1; j < headerNames.Count; j++) for (int j = i + 1; j < columnNames.Count; j++)
{ {
if (headerNames[i] == headerNames[j]) if (columnNames[i] == columnNames[j])
{ {
throw new Exception("Column names in header must be unique"); throw new Exception("Column names in header must be unique");
} }
} }
byte[] nameEncoded = Encoding.UTF8.GetBytes(headerNames[i]); byte[] nameEncoded = Encoding.UTF8.GetBytes(columnNames[i]);
for (int j = 0; j < nameEncoded.Length; j++) for (int j = 0; j < nameEncoded.Length; j++)
{ {
@ -1250,15 +1254,15 @@ public class SaneTsv
bytes.Add((byte)':'); bytes.Add((byte)':');
try try
{ {
bytes.AddRange(Encoding.UTF8.GetBytes(GetNameFromColumn(headerTypes[i]))); bytes.AddRange(Encoding.UTF8.GetBytes(GetNameFromColumn(columnTypes[i])));
} }
catch (Exception e) catch (Exception e)
{ {
throw new Exception($"Invalid header type for column {i}", e); throw new Exception($"Invalid column type for column {i}", e);
} }
} }
if (i == headerNames.Count - 1) if (i == columnNames.Count - 1)
{ {
bytes.Add((byte)'\n'); bytes.Add((byte)'\n');
} }
@ -1269,19 +1273,19 @@ public class SaneTsv
} }
// Serialize data // Serialize data
SerializeTsv<T>(data, bytes, headerPropertyInfos.ToArray(), headerTypes.ToArray(), tsvFormat, 0, data.Count); SerializeTsv<T>(data, bytes, columnPropertyInfos.ToArray(), columnTypes.ToArray(), tsvFormat, 0, data.Count);
return bytes.ToArray(); return bytes.ToArray();
} }
protected static void SerializeTsv<T>(IList<T> data, List<byte> bytes, PropertyInfo[] headerPropertyInfos, Type[] headerTypes, FormatType tsvFormat, int startIndex, int endIndex) protected static void SerializeTsv<T>(IList<T> data, List<byte> bytes, PropertyInfo[] columnPropertyInfos, Type[] 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++)
{ {
for (int j = 0; j < headerTypes.Length; j++) for (int j = 0; j < columnTypes.Length; j++)
{ {
object datum = headerPropertyInfos[j].GetValue(data[i]); object datum = columnPropertyInfos[j].GetValue(data[i]);
try try
{ {
@ -1289,16 +1293,16 @@ 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 (headerTypes[j] == typeof(StringType)) if (columnTypes[j] == typeof(StringType))
{ {
fieldEncoded = Encoding.UTF8.GetBytes((string)datum); fieldEncoded = Encoding.UTF8.GetBytes((string)datum);
} }
else if (headerTypes[j] == typeof(BooleanType)) else if (columnTypes[j] == typeof(BooleanType))
{ {
bytes.AddRange((bool)datum ? TrueEncoded : FalseEncoded); bytes.AddRange((bool)datum ? TrueEncoded : FalseEncoded);
skipEscaping = true; skipEscaping = true;
} }
else if (headerTypes[j] == typeof(Float32Type)) else if (columnTypes[j] == typeof(Float32Type))
{ {
if (datum is float f) if (datum is float f)
{ {
@ -1322,7 +1326,7 @@ public class SaneTsv
} }
skipEscaping = true; skipEscaping = true;
} }
else if (headerTypes[j] == typeof(Float32LEType)) else if (columnTypes[j] == typeof(Float32LEType))
{ {
if (LittleEndian) if (LittleEndian)
{ {
@ -1338,7 +1342,7 @@ public class SaneTsv
} }
} }
} }
else if (headerTypes[j] == typeof(Float64Type)) else if (columnTypes[j] == typeof(Float64Type))
{ {
if (datum is double d) if (datum is double d)
{ {
@ -1362,7 +1366,7 @@ public class SaneTsv
} }
skipEscaping = true; skipEscaping = true;
} }
else if (headerTypes[j] == typeof(Float64LEType)) else if (columnTypes[j] == typeof(Float64LEType))
{ {
if (LittleEndian) if (LittleEndian)
{ {
@ -1378,33 +1382,33 @@ public class SaneTsv
} }
} }
} }
else if (headerTypes[j] == typeof(UInt32Type)) else if (columnTypes[j] == typeof(UInt32Type))
{ {
bytes.AddRange(Encoding.UTF8.GetBytes(((UInt32)datum).ToString())); bytes.AddRange(Encoding.UTF8.GetBytes(((UInt32)datum).ToString()));
skipEscaping = true; skipEscaping = true;
} }
else if (headerTypes[j] == typeof(UInt64Type)) else if (columnTypes[j] == typeof(UInt64Type))
{ {
bytes.AddRange(Encoding.UTF8.GetBytes(((UInt64)datum).ToString())); bytes.AddRange(Encoding.UTF8.GetBytes(((UInt64)datum).ToString()));
skipEscaping = true; skipEscaping = true;
} }
else if (headerTypes[j] == typeof(Int32Type)) else if (columnTypes[j] == typeof(Int32Type))
{ {
bytes.AddRange(Encoding.UTF8.GetBytes(((Int32)datum).ToString())); bytes.AddRange(Encoding.UTF8.GetBytes(((Int32)datum).ToString()));
skipEscaping = true; skipEscaping = true;
} }
else if (headerTypes[j] == typeof(Int64Type)) else if (columnTypes[j] == typeof(Int64Type))
{ {
bytes.AddRange(Encoding.UTF8.GetBytes(((Int64)datum).ToString())); bytes.AddRange(Encoding.UTF8.GetBytes(((Int64)datum).ToString()));
skipEscaping = true; skipEscaping = true;
} }
else if (headerTypes[j] == typeof(BinaryType)) else if (columnTypes[j] == typeof(BinaryType))
{ {
fieldEncoded = (byte[])datum; fieldEncoded = (byte[])datum;
} }
else else
{ {
throw new Exception($"Unexpected column type {headerTypes[j]} for column {j}"); throw new Exception($"Unexpected column type {columnTypes[j]} for column {j}");
} }
if (!skipEscaping) if (!skipEscaping)
@ -1438,7 +1442,7 @@ public class SaneTsv
} }
} }
if (j < headerTypes.Length - 1) if (j < columnTypes.Length - 1)
{ {
bytes.Add((byte)'\t'); bytes.Add((byte)'\t');
} }
@ -1449,7 +1453,7 @@ public class SaneTsv
} }
catch (InvalidCastException e) catch (InvalidCastException e)
{ {
throw new Exception($"Record {i}, field {j} expected type compatible with {GetNameFromColumn(headerTypes[j])}", e); throw new Exception($"Record {i}, field {j} expected type compatible with {GetNameFromColumn(columnTypes[j])}", e);
} }
} }
} }

View File

@ -1,4 +1,5 @@
using NathanMcRae; using NathanMcRae;
using System.Reflection;
using System.Text; using System.Text;
internal class Program : SaneTsv internal class Program : SaneTsv
@ -80,6 +81,18 @@ internal class Program : SaneTsv
public string Column3 { get; set; } 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 public class SerdeTestRecord : SaneTsv.CommentedTsvRecord
{ {
[SaneTsv.TypedTsvColumn("column1")] [SaneTsv.TypedTsvColumn("column1")]
@ -348,6 +361,404 @@ internal class Program : SaneTsv
Console.WriteLine($"Unspecced parse time: {unspeccedParseTime}"); Console.WriteLine($"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");
} }
} }