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

Clarify terminology of 'header'/'column'

Browse Source 
The header is just the region containing the column name / types.
This commit is contained in:
Nathan McRae 2024-03-09 13:13:41 -08:00
parent d7720d8cde
commit 32393e704d
1 changed files with 61 additions and 61 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

122
SaneTsv.cs
View File

@ -87,9 +87,9 @@ public class SaneTsv
}
parsed.Records = new List<T>();
var headerTypes = new List<Type>();
var headerNames = new List<string>();
var headerPropertyInfos = new List<PropertyInfo>();
var columnTypes = new List<Type>();
var columnNames = new List<string>();
var columnPropertyInfos = new List<PropertyInfo>();
int columnCount = 0;
foreach (PropertyInfo property in typeof(T).GetProperties())
@ -100,9 +100,9 @@ public class SaneTsv
continue;
}
headerNames.Add(attribute.ColumnName ?? property.Name);
headerTypes.Add(attribute.ColumnType ?? GetColumnFromType(property.PropertyType));
headerPropertyInfos.Add(property);
columnNames.Add(attribute.ColumnName ?? property.Name);
columnTypes.Add(attribute.ColumnType ?? GetColumnFromType(property.PropertyType));
columnPropertyInfos.Add(property);
// TODO: Check that the property type and given column type are compatible
columnCount++;
}
@ -169,7 +169,7 @@ public class SaneTsv
}
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;
@ -178,7 +178,7 @@ public class SaneTsv
{
if (format == FormatType.SIMPLE_TSV)
{
throw new Exception($"Header {j} contains ':', which is not allowed for column names");
throw new Exception($"Header field {j} contains ':', which is not allowed for column names");
}
columnTypeString = columnString.Split(":").Last();
columnName = columnString.Substring(0, columnString.Length - columnTypeString.Length - 1);
@ -187,7 +187,7 @@ public class SaneTsv
{
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 = "";
columnName = columnString;
@ -240,14 +240,14 @@ public class SaneTsv
// 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]}");
}
}
@ -311,7 +311,7 @@ public class SaneTsv
// Complication: it probably depends on processor count
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;
}
else
@ -333,7 +333,7 @@ public class SaneTsv
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
@ -351,7 +351,7 @@ public class SaneTsv
// 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
// (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 fields = new List<byte[]>();
@ -421,9 +421,9 @@ public class SaneTsv
fields.Add(fieldBytes.ToArray());
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
{
@ -433,7 +433,7 @@ public class SaneTsv
comment = currentComment.ToString();
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();
}
@ -493,9 +493,9 @@ public class SaneTsv
// TODO
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
{
@ -505,7 +505,7 @@ public class SaneTsv
comment = currentComment.ToString();
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();
}
@ -846,7 +846,7 @@ public class SaneTsv
public static (string[] columns, string[][] data) ParseSimpleTsv(byte[] inputBuffer)
{
string[] columnNames = null;
var headers = new List<byte[]>();
var headerFields = new List<byte[]>();
var fieldBytes = new List<byte>();
int startOfData = -1;
for (int i = 0; i < inputBuffer.Count(); i++)
@ -885,33 +885,33 @@ public class SaneTsv
else if (inputBuffer[i] == '\t')
{
// end of field
headers.Add(fieldBytes.ToArray());
headerFields.Add(fieldBytes.ToArray());
fieldBytes.Clear();
}
else if (inputBuffer[i] == '\n')
{
// This is the end of the header
headers.Add(fieldBytes.ToArray());
headerFields.Add(fieldBytes.ToArray());
startOfData = i + 1;
columnNames = new string[headers.Count];
columnNames = new string[headerFields.Count];
fieldBytes.Clear();
for (int j = 0; j < headers.Count; j++)
for (int j = 0; j < headerFields.Count; j++)
{
string columnString;
try
{
columnString = Encoding.UTF8.GetString(headers[j]);
columnString = Encoding.UTF8.GetString(headerFields[j]);
}
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(':'))
{
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;
@ -1184,9 +1184,9 @@ public class SaneTsv
{
var bytes = new List<byte>();
var headerTypes = new List<Type>();
var headerNames = new List<string>();
var headerPropertyInfos = new List<PropertyInfo>();
var columnTypes = new List<Type>();
var columnNames = new List<string>();
var columnPropertyInfos = new List<PropertyInfo>();
int columnCount = 0;
// Serialize header
@ -1199,30 +1199,30 @@ public class SaneTsv
}
string headerName = attribute.ColumnName ?? property.Name;
headerNames.Add(headerName);
columnNames.Add(headerName);
Type headerType = attribute.ColumnType ?? GetColumnFromType(property.PropertyType);
if (tsvFormat == FormatType.SIMPLE_TSV && headerType != typeof(StringType))
{
throw new Exception($"Serializing Simple TSV requires all columns be of type string, but column '{headerName}' has type '{headerType}'");
}
headerTypes.Add(headerType);
headerPropertyInfos.Add(property);
columnTypes.Add(headerType);
columnPropertyInfos.Add(property);
// TODO: Check that the property type and given column type are compatible
columnCount++;
}
// Serialize header
for (int i = 0; i < headerNames.Count; i++)
for (int i = 0; i < columnNames.Count; i++)
{
for (int j = i + 1; j < headerNames.Count; j++)
for (int j = i + 1; j < columnNames.Count; j++)
{
if (headerNames[i] == headerNames[j])
if (columnNames[i] == columnNames[j])
{
throw new Exception("Column names in header must be unique");
}
}
byte[] nameEncoded = Encoding.UTF8.GetBytes(headerNames[i]);
byte[] nameEncoded = Encoding.UTF8.GetBytes(columnNames[i]);
for (int j = 0; j < nameEncoded.Length; j++)
{
@ -1257,15 +1257,15 @@ public class SaneTsv
bytes.Add((byte)':');
try
{
bytes.AddRange(Encoding.UTF8.GetBytes(GetNameFromColumn(headerTypes[i])));
bytes.AddRange(Encoding.UTF8.GetBytes(GetNameFromColumn(columnTypes[i])));
}
catch (Exception e)
{
throw new Exception($"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');
}
@ -1276,19 +1276,19 @@ public class SaneTsv
}
// 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();
}
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
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
{
@ -1296,16 +1296,16 @@ public class SaneTsv
// Some fields definitely don't need escaping, so we add them directly to bytes
bool skipEscaping = false;
if (headerTypes[j] == typeof(StringType))
if (columnTypes[j] == typeof(StringType))
{
fieldEncoded = Encoding.UTF8.GetBytes((string)datum);
}
else if (headerTypes[j] == typeof(BooleanType))
else if (columnTypes[j] == typeof(BooleanType))
{
bytes.AddRange((bool)datum ? TrueEncoded : FalseEncoded);
skipEscaping = true;
}
else if (headerTypes[j] == typeof(Float32Type))
else if (columnTypes[j] == typeof(Float32Type))
{
if (datum is float f)
{
@ -1329,7 +1329,7 @@ public class SaneTsv
}
skipEscaping = true;
}
else if (headerTypes[j] == typeof(Float32LEType))
else if (columnTypes[j] == typeof(Float32LEType))
{
if (LittleEndian)
{
@ -1345,7 +1345,7 @@ public class SaneTsv
}
}
}
else if (headerTypes[j] == typeof(Float64Type))
else if (columnTypes[j] == typeof(Float64Type))
{
if (datum is double d)
{
@ -1369,7 +1369,7 @@ public class SaneTsv
}
skipEscaping = true;
}
else if (headerTypes[j] == typeof(Float64LEType))
else if (columnTypes[j] == typeof(Float64LEType))
{
if (LittleEndian)
{
@ -1385,33 +1385,33 @@ public class SaneTsv
}
}
}
else if (headerTypes[j] == typeof(UInt32Type))
else if (columnTypes[j] == typeof(UInt32Type))
{
bytes.AddRange(Encoding.UTF8.GetBytes(((UInt32)datum).ToString()));
skipEscaping = true;
}
else if (headerTypes[j] == typeof(UInt64Type))
else if (columnTypes[j] == typeof(UInt64Type))
{
bytes.AddRange(Encoding.UTF8.GetBytes(((UInt64)datum).ToString()));
skipEscaping = true;
}
else if (headerTypes[j] == typeof(Int32Type))
else if (columnTypes[j] == typeof(Int32Type))
{
bytes.AddRange(Encoding.UTF8.GetBytes(((Int32)datum).ToString()));
skipEscaping = true;
}
else if (headerTypes[j] == typeof(Int64Type))
else if (columnTypes[j] == typeof(Int64Type))
{
bytes.AddRange(Encoding.UTF8.GetBytes(((Int64)datum).ToString()));
skipEscaping = true;
}
else if (headerTypes[j] == typeof(BinaryType))
else if (columnTypes[j] == typeof(BinaryType))
{
fieldEncoded = (byte[])datum;
}
else
{
throw new Exception($"Unexpected column type {headerTypes[j]} for column {j}");
throw new Exception($"Unexpected column type {columnTypes[j]} for column {j}");
}
if (!skipEscaping)
@ -1445,7 +1445,7 @@ public class SaneTsv
}
}
if (j < headerTypes.Length - 1)
if (j < columnTypes.Length - 1)
{
bytes.Add((byte)'\t');
}
@ -1456,7 +1456,7 @@ public class SaneTsv
}
catch (InvalidCastException e)
{
throw new Exception($"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);
}
}
}