Compare commits

...

10 Commits

Author SHA1 Message Date
Nathan McRae
8e3332b484 Remove unused SaneTsv fields 2024-03-10 12:55:21 -07:00
Nathan McRae
32393e704d 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
d7720d8cde Flesh out tests 2024-03-09 09:58:25 -08:00
Nathan McRae
aff4b353bb Update some comments 2024-03-09 09:58:09 -08:00
Nathan McRae
203458fdf7 Fix some end-of-file simple TSV parsing 2024-03-09 09:57:56 -08:00
Nathan McRae
4148475031 Fix some minor line/column numbering issues 2024-03-09 09:57:27 -08:00
Nathan McRae
e38baa9167 Fix column attribute bug 2024-03-09 09:57:00 -08:00
Nathan McRae
a66f6a1368 Make parallel implementations the default 2024-03-08 12:31:40 -08:00
Nathan McRae
96af5ae82c Complete parallel parsing implementation 2024-03-08 11:43:06 -08:00
Nathan McRae
11c7daec8e Start parallel versions of general TSV serialization/parsing
They mostly work, but are not actually parallelized yet and likely have some edge cases.
Also, the soon-to-be parallel version of parsing is very slow compared to the original.
2024-02-25 22:35:56 -08:00
2 changed files with 731 additions and 511 deletions

File diff suppressed because it is too large Load Diff

View File

@ -1,7 +1,8 @@
using NathanMcRae;
using System.Reflection;
using System.Text;
internal class Program
internal class Program : SaneTsv
{
public class TestRecord : SaneTsv.TsvRecord
{
@ -80,6 +81,18 @@ internal class Program
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")]
@ -349,45 +362,26 @@ internal class Program
}
{
string testName = "Check parallel serialization";
string testName = "With and without file comment";
int N = 100000;
var records = new StringTestRecord[N];
var rand = new Random(1);
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";
for (int i = 0; i < N; i++)
{
records[i] = new StringTestRecord()
{
Column1 = rand.Next().ToString(),
column2 = rand.Next().ToString(),
Column3 = rand.Next().ToString(),
};
}
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";
string[][] recordStrings = records.Select(record => new string[] { record.Column1, record.column2, record.Column3 }).ToArray();
CommentedTsv<BoolTestRecord2> parsed = SaneTsv.ParseCommentedTsv<BoolTestRecord2>(Encoding.UTF8.GetBytes(testString1));
CommentedTsv<BoolTestRecord2> parsed2 = SaneTsv.ParseCommentedTsv<BoolTestRecord2>(Encoding.UTF8.GetBytes(testString2));
DateTime lastTime = DateTime.Now;
byte[] serialized1 = SaneTsv.SerializeSimpleTsv(new string[] { "column1", "column2", "columnthree\nyep" }, recordStrings);
TimeSpan unparallelTime = DateTime.Now - lastTime;
lastTime = DateTime.Now;
byte[] serialized2 = SaneTsv.SerializeSimpleTsvParallel(new string[] { "column1", "column2", "columnthree\nyep" }, recordStrings);
TimeSpan parallelTime = DateTime.Now - lastTime;
Console.WriteLine($"Unparallel serialization time: {unparallelTime}");
Console.WriteLine($"Parallel serialization time: {parallelTime}");
bool matching = true;
for (int i = 0; i < Math.Min(serialized1.Length, serialized2.Length); i++)
{
if (serialized1[i] != serialized2[i])
{
matching = false;
break;
}
}
if (matching)
if (parsed.FileComment == "This is a file comment\nOne more file comment line" && parsed2.FileComment == null)
{
Console.WriteLine($"Passed {testName}");
}
@ -398,63 +392,370 @@ internal class Program
}
{
string testName = "Check parallel parsing";
string testName = "With and without types";
int N = 100000;
var records = new StringTestRecord[N];
var rand = new Random(1);
string testString1 = "column1:type:boolean\tcolumn2:binary\tcolumnthree\\nyep:string" +
"\nTRUE\tvalue\\\\twoo\tvaluetrhee" +
"\nFALSE\tnother\tno\\ther";
for (int i = 0; i < N; i++)
try
{
records[i] = new StringTestRecord()
{
Column1 = rand.Next().ToString(),
column2 = rand.Next().ToString(),
Column3 = rand.Next().ToString(),
};
Tsv<BoolTestRecord2> parsed = SaneTsv.ParseTypedTsv<BoolTestRecord2>(Encoding.UTF8.GetBytes(testString1));
Console.WriteLine($"Passed {testName} 1A");
}
catch (Exception e)
{
Console.WriteLine($"Failed {testName} 1A");
}
byte[] serialized = SaneTsv.SerializeSimpleTsv<StringTestRecord>(records);
DateTime lastTime = DateTime.Now;
(string[] headers2, string[][] data2) = SaneTsv.ParseSimpleTsv(serialized);
TimeSpan unparallelTime = DateTime.Now - lastTime;
lastTime = DateTime.Now;
(string[] headers, string[][] data) = SaneTsv.ParseSimpleTsvParallel(serialized);
TimeSpan parallelTime = DateTime.Now - lastTime;
Console.WriteLine($"Unparallel serialization time: {unparallelTime}");
Console.WriteLine($"Parallel serialization time: {parallelTime}");
bool matching = true;
for (int j = 0; j < Math.Min(headers2.Length, headers.Length); j++)
try
{
if (headers[j] != headers2[j])
Tsv<BoolTestRecord2> parsed2 = SaneTsv.ParseSimpleTsv<BoolTestRecord2>(Encoding.UTF8.GetBytes(testString1));
Console.WriteLine($"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($"Passed {testName} 1D");
}
}
{
string testName = "Partial parsing";
string line1 = "column1\tcolumn2\tcolumnthree\\nyep";
string line2 = "\nTRUE\tvalue\\\\t\0woo\tvaluetrhee";
string line3 = "\nFALSE\tnother\tno\\ther";
byte[] inputBuffer = Encoding.UTF8.GetBytes(line1 + line2 + line3);
var headerTypes = new List<Type>();
var headerNames = new List<string>();
var headerPropertyInfos = new List<PropertyInfo>();
int columnCount = 0;
foreach (PropertyInfo property in typeof(BoolTestRecord3).GetProperties())
{
TsvColumnAttribute attribute = (TsvColumnAttribute)Attribute.GetCustomAttribute(property, typeof(TsvColumnAttribute));
if (attribute == null)
{
matching = false;
break;
continue;
}
headerNames.Add(attribute.ColumnName ?? property.Name);
headerTypes.Add(attribute.ColumnType ?? GetColumnFromType(property.PropertyType));
headerPropertyInfos.Add(property);
// TODO: Check that the property type and given column type are compatible
columnCount++;
}
for (int i = 0; i < Math.Min(data.Length, data2.Length) && matching; i++)
{
for (int j = 0; j < data[0].Length; j++)
{
if (data[i][j] != data2[i][j])
{
matching = false;
break;
}
}
}
BoolTestRecord3[] records = SaneTsv.Parse<BoolTestRecord3>(inputBuffer,
FormatType.SIMPLE_TSV,
headerPropertyInfos.ToArray(),
headerTypes.ToArray(),
line1.Length + line2.Length + 1,
inputBuffer.Length);
if (matching)
if (records.Length == 0 )
{
Console.WriteLine($"Passed {testName}");
Console.WriteLine($"Passed {testName} 1");
}
else
{
Console.WriteLine($"Failed {testName}");
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");
}
}