nathanmcrae/extra-tsv
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity

Add type parsing

Browse Source
This commit is contained in:
Nathan McRae 2024-02-14 14:30:36 -08:00
parent 695f98b67c
commit d199984364
2 changed files with 248 additions and 57 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

267
SaneTsv.cs
View File

@ -7,20 +7,37 @@ namespace NathanMcRae;
/// </summary>
public class SaneTsv
{
public enum ColumnType
{
STRING,
BOOLEAN,
FLOAT32,
FLOAT64,
UINT32,
UINT64,
INT32,
INT64,
BINARY,
}
// TODO: We need to be able to update all these in tandem somehow
public string[] ColumnNames { get; protected set; }
public Dictionary<string, List<string>> Columns { get; protected set; }
public ColumnType[] ColumnTypes { get; protected set; }
public Dictionary<string, List<object>> Columns { get; protected set; }
public List<SaneTsvRecord> Records { get; protected set; }
// TODO: Parse with specified columns / types
public static SaneTsv Parse(byte[] inputBuffer)
{
var parsed = new SaneTsv();
parsed.Columns = new Dictionary<string, List<string>>();
parsed.Columns = new Dictionary<string, List<object>>();
parsed.ColumnNames = new string[] { };
parsed.ColumnTypes = new ColumnType[] { };
parsed.Records = new List<SaneTsvRecord>();
var fieldBytes = new List<byte>();
var fields = new List<string>();
var fields = new List<byte[]>();
int numFields = -1;
for (int i = 0; i < inputBuffer.Count(); i++)
{
@ -53,26 +70,12 @@ public class SaneTsv
else if (inputBuffer[i] == '\t')
{
// end of field
try
{
fields.Add(Encoding.UTF8.GetString(fieldBytes.ToArray()));
}
catch (Exception e)
{
throw new Exception($"Field {fields.Count} on line {parsed.Records.Count + 1} is not valid UTF-8", e);
}
fields.Add(fieldBytes.ToArray());
fieldBytes.Clear();
}
else if (inputBuffer[i] == '\n')
{
try
{
fields.Add(Encoding.UTF8.GetString(fieldBytes.ToArray()));
}
catch (Exception e)
{
throw new Exception($"Field {fields.Count} on line {parsed.Records.Count + 1} is not valid UTF-8", e);
}
fields.Add(fieldBytes.ToArray());
fieldBytes.Clear();
if (numFields < 0)
@ -82,14 +85,76 @@ public class SaneTsv
numFields = fields.Count;
parsed.ColumnNames = new string[numFields];
parsed.ColumnTypes = new ColumnType[numFields];
int numTypesBlank = 0;
for (int j = 0; j < fields.Count; j++)
{
string columnName = fields[j];
string columnString;
try
{
columnString = Encoding.UTF8.GetString(fields[j]);
}
catch (Exception e)
{
throw new Exception($"Header {fields.Count} is not valid UTF-8", e);
}
string columnTypeString;
string columnName;
if (columnString.Contains(":")) {
columnTypeString = columnString.Split(":").Last();
columnName = columnString.Substring(0, columnString.Length - columnTypeString.Length - 1);
}
else
{
columnTypeString = "";
columnName = columnString;
}
ColumnType type;
switch (columnTypeString)
{
case "":
numTypesBlank++;
type = ColumnType.STRING;
break;
case "string":
type = ColumnType.STRING;
break;
case "boolean":
type = ColumnType.BOOLEAN;
break;
case "float32":
type = ColumnType.FLOAT32;
break;
case "float64":
type = ColumnType.FLOAT64;
break;
case "uint32":
type = ColumnType.UINT32;
break;
case "uint64":
type = ColumnType.UINT64;
break;
case "int32":
type = ColumnType.INT32;
break;
case "int64":
type = ColumnType.INT64;
break;
case "binary":
type = ColumnType.BINARY;
break;
default:
throw new Exception($"Invalid type '{columnTypeString}' for column {j}");
}
try
{
parsed.Columns.Add(columnName, new List<string>());
parsed.Columns.Add(columnName, new List<object>());
}
catch (Exception e)
{
@ -97,6 +162,12 @@ public class SaneTsv
}
parsed.ColumnNames[j] = columnName;
parsed.ColumnTypes[j] = type;
}
if (numTypesBlank != 0 && numTypesBlank != fields.Count)
{
throw new Exception("Types must be provided for all columns or none. Use 'string' for columns missing types.");
}
fields.Clear();
@ -107,13 +178,7 @@ public class SaneTsv
}
else
{
for (int j = 0; j < fields.Count; j++)
{
parsed.Columns[parsed.ColumnNames[j]].Add(fields[j]);
}
parsed.Records.Add(new SaneTsvRecord(parsed, fields.ToArray()));
fields.Clear();
AddRecord(parsed, fields);
}
}
else
@ -122,50 +187,146 @@ public class SaneTsv
}
}
try
{
fields.Add(Encoding.UTF8.GetString(fieldBytes.ToArray()));
}
catch (Exception e)
{
throw new Exception($"Field {fields.Count} on line {parsed.Records.Count + 1} is not valid UTF-8", e);
}
fields.Add(fieldBytes.ToArray());
if (numFields == 0)
{
throw new Exception("Found 0 fields on last line. Possibly because of extra \\n after last record");
}
if (numFields != fields.Count)
{
throw new Exception($"Expected {numFields} fields on line {parsed.Records.Count + 2}, but found {fields.Count}");
}
else
{
for (int j = 0; j < fields.Count; j++)
{
try
{
parsed.Columns[parsed.ColumnNames[j]].Add(fields[j]);
}
catch (Exception e)
{
throw new Exception($"Field {j} on line {parsed.Records.Count + 1} is not valid UTF-8", e);
}
}
parsed.Records.Add(new SaneTsvRecord(parsed, fields.ToArray()));
fields.Clear();
AddRecord(parsed, fields);
}
return parsed;
}
protected static void AddRecord(SaneTsv parsed, List<byte[]> fields)
{
var parsedFields = new object[fields.Count];
for (int j = 0; j < fields.Count; j++)
{
// All other types require the content to be UTF-8. Binary fields can ignore that.
if (parsed.ColumnTypes[j] == ColumnType.BINARY)
{
parsedFields[j] = fields[j];
parsed.Columns[parsed.ColumnNames[j]].Add(fields[j]);
continue;
}
string fieldString;
try
{
fieldString = Encoding.UTF8.GetString(fields[j]);
}
catch (Exception e)
{
throw new Exception($"Field {j} on line {parsed.Records.Count + 2} is not valid UTF-8", e);
}
switch (parsed.ColumnTypes[j])
{
case ColumnType.STRING:
parsedFields[j] = fieldString;
parsed.Columns[parsed.ColumnNames[j]].Add(fieldString);
break;
case ColumnType.BOOLEAN:
bool parsedBool;
if (fieldString == "TRUE")
{
parsedBool = true;
}
else if (fieldString == "FALSE")
{
parsedBool = false;
}
else
{
throw new Exception($"Field {j} on line {parsed.Records.Count + 2} is not valid boolean. Must be 'TRUE' or 'FALSE' exactly");
}
parsedFields[j] = parsedBool;
parsed.Columns[parsed.ColumnNames[j]].Add(parsedBool);
break;
case ColumnType.FLOAT32:
if (!float.TryParse(fieldString, out float parsedFloat))
{
throw new Exception($"Field {j} on line {parsed.Records.Count + 2} is not valid single-precision float");
}
parsedFields[j] = parsedFloat;
parsed.Columns[parsed.ColumnNames[j]].Add(parsedFloat);
break;
case ColumnType.FLOAT64:
if (!double.TryParse(fieldString, out double parsedDouble))
{
throw new Exception($"Field {j} on line {parsed.Records.Count + 2} is not valid double-precision float");
}
parsedFields[j] = parsedDouble;
parsed.Columns[parsed.ColumnNames[j]].Add(parsedDouble);
break;
case ColumnType.UINT32:
if (!UInt32.TryParse(fieldString, out UInt32 parsedUInt32))
{
throw new Exception($"Field {j} on line {parsed.Records.Count + 2} is not valid UInt32");
}
parsedFields[j] = parsedUInt32;
parsed.Columns[parsed.ColumnNames[j]].Add(parsedUInt32);
break;
case ColumnType.UINT64:
if (!UInt64.TryParse(fieldString, out UInt64 parsedUInt64))
{
throw new Exception($"Field {j} on line {parsed.Records.Count + 2} is not valid UInt64");
}
parsedFields[j] = parsedUInt64;
parsed.Columns[parsed.ColumnNames[j]].Add(parsedUInt64);
break;
case ColumnType.INT32:
if (!Int32.TryParse(fieldString, out Int32 parsedInt32))
{
throw new Exception($"Field {j} on line {parsed.Records.Count + 2} is not valid Int32");
}
parsedFields[j] = parsedInt32;
parsed.Columns[parsed.ColumnNames[j]].Add(parsedInt32);
break;
case ColumnType.INT64:
if (!Int64.TryParse(fieldString, out Int64 parsedInt64))
{
throw new Exception($"Field {j} on line {parsed.Records.Count + 2} is not valid Int64");
}
parsedFields[j] = parsedInt64;
parsed.Columns[parsed.ColumnNames[j]].Add(parsedInt64);
break;
case ColumnType.BINARY:
throw new Exception($"Unexpected type {parsed.ColumnTypes[j]}");
default:
throw new Exception($"Unexpected type {parsed.ColumnTypes[j]}");
}
}
parsed.Records.Add(new SaneTsvRecord(parsed, parsedFields));
fields.Clear();
}
public SaneTsvRecord this[int i] => Records[i];
public class SaneTsvRecord
{
public SaneTsv Parent { get; }
public string[] Fields { get; }
public object[] Fields { get; }
public string this[string columnName] => Fields[Array.IndexOf(Parent.ColumnNames, columnName)];
public object this[string columnName] => Fields[Array.IndexOf(Parent.ColumnNames, columnName)];
public SaneTsvRecord(SaneTsv parent, string[] fields)
public SaneTsvRecord(SaneTsv parent, object[] fields)
{
Parent = parent;
Fields = fields;

38
SaneTsvTest/Program.cs
View File

@ -1,8 +1,38 @@
using NathanMcRae;
using System.Text;
string testString1 = "column1\tcolumn2\tcolumnthree\\nyep\nvalue1\tvalue\\\\twoo\tvaluetrhee\nthis\\nis\\na\\nvalue\tnother\tno\\ther";
{
string testName = "Bool test";
string testString1 = "column1:type:boolean\tcolumn2:binary\tcolumnthree\\nyep:string" +
"\nTRUE\tvalue\\\\t\0woo\tvaluetrhee" +
"\nFALSE\tnother\tno\\ther";
SaneTsv parsed = SaneTsv.Parse(Encoding.UTF8.GetBytes(testString1));
// See https://aka.ms/new-console-template for more information
Console.WriteLine("Hello, World!");
SaneTsv parsed = SaneTsv.Parse(Encoding.UTF8.GetBytes(testString1));
if (parsed.Records[0]["column1:type"] is bool result && result)
{
Console.WriteLine($"Passed {testName}");
}
else
{
Console.WriteLine($"Failed {testName}");
}
}
{
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";
SaneTsv parsed = SaneTsv.Parse(Encoding.UTF8.GetBytes(testString1));
Console.WriteLine($"Failed {testName}");
}
catch (Exception)
{
Console.WriteLine($"Passed {testName}");
}
Console.WriteLine("Done with tests");
}