public void SetColumn(int columnIndex, BaseColumn column)
{
column = column ?? throw new ArgumentNullException(nameof(column));
if ((uint)columnIndex >= ColumnCount)
{
throw new ArgumentOutOfRangeException(nameof(columnIndex));
}
if (RowCount > 0 && column.Length != RowCount)
{
throw new ArgumentException(Strings.MismatchedColumnLengths, nameof(column));
}
if (_columnNameToIndexDictionary.ContainsKey(column.Name))
{
throw new ArgumentException($"Table already contains a column called {column.Name}");
}
_columnNameToIndexDictionary.Remove(_columnNames[columnIndex]);
_columnNames[columnIndex] = column.Name;
_columnNameToIndexDictionary[column.Name] = columnIndex;
_columns[columnIndex] = column;
}
public void InsertColumn(int columnIndex, BaseColumn column, DataFrame parent)
{
column = column ?? throw new ArgumentNullException(nameof(column));
if ((uint)columnIndex > _columns.Count)
{
throw new ArgumentOutOfRangeException(nameof(columnIndex));
}
if (RowCount > 0 && column.Length != RowCount)
{
throw new ArgumentException(Strings.MismatchedColumnLengths, nameof(column));
}
if (_columnNameToIndexDictionary.ContainsKey(column.Name))
{
throw new ArgumentException($"Table already contains a column called {column.Name}");
}
RowCount = column.Length;
_columnNames.Insert(columnIndex, column.Name);
_columnNameToIndexDictionary[column.Name] = columnIndex;
_columns.Insert(columnIndex, column);
ColumnCount++;
}
public override BaseColumn Clone(BaseColumn mapIndices = null, bool invertMapIndices = false, long numberOfNullsToAppend = 0)
{
PrimitiveColumn <T> clone;
if (!(mapIndices is null))
{
if (mapIndices.DataType != typeof(long) && mapIndices.DataType != typeof(bool))
{
throw new ArgumentException(String.Format(Strings.MultipleMismatchedValueType, typeof(long), typeof(bool)), nameof(mapIndices));
}
if (mapIndices.Length > Length)
{
throw new ArgumentException(Strings.MapIndicesExceedsColumnLenth, nameof(mapIndices));
}
if (mapIndices.DataType == typeof(long))
{
clone = Clone(mapIndices as PrimitiveColumn <long>, invertMapIndices);
}
else
{
clone = Clone(mapIndices as PrimitiveColumn <bool>);
}
}
public IEnumerable <RecordBatch> AsArrowRecordBatches()
{
Apache.Arrow.Schema.Builder schemaBuilder = new Apache.Arrow.Schema.Builder();
int columnCount = ColumnCount;
for (int i = 0; i < columnCount; i++)
{
BaseColumn column = Column(i);
Field field = column.Field();
schemaBuilder.Field(field);
}
Schema schema = schemaBuilder.Build();
List <Apache.Arrow.Array> arrays = new List <Apache.Arrow.Array>();
int recordBatchLength = Int32.MaxValue;
int numberOfRowsInThisRecordBatch = (int)Math.Min(recordBatchLength, RowCount);
long numberOfRowsProcessed = 0;
// Sometimes .NET for Spark passes in DataFrames with no rows. In those cases, we just return a RecordBatch with the right Schema and no rows
do
{
for (int i = 0; i < columnCount; i++)
{
BaseColumn column = Column(i);
numberOfRowsInThisRecordBatch = (int)Math.Min(numberOfRowsInThisRecordBatch, column.MaxRecordBatchLength(numberOfRowsProcessed));
}
for (int i = 0; i < columnCount; i++)
{
BaseColumn column = Column(i);
arrays.Add(column.AsArrowArray(numberOfRowsProcessed, numberOfRowsInThisRecordBatch));
}
numberOfRowsProcessed += numberOfRowsInThisRecordBatch;
yield return(new RecordBatch(schema, arrays, numberOfRowsInThisRecordBatch));
} while (numberOfRowsProcessed < RowCount);
}
public DataFrameTable(BaseColumn column) : this(new List <BaseColumn> {
column
})
{
}
public DataFrame(RecordBatch recordBatch)
{
_table = new DataFrameTable();
Apache.Arrow.Schema arrowSchema = recordBatch.Schema;
int fieldIndex = 0;
IEnumerable <IArrowArray> arrowArrays = recordBatch.Arrays;
foreach (IArrowArray arrowArray in arrowArrays)
{
Field field = arrowSchema.GetFieldByIndex(fieldIndex);
IArrowType fieldType = field.DataType;
BaseColumn dataFrameColumn = null;
switch (fieldType.TypeId)
{
case ArrowTypeId.Boolean:
BooleanArray arrowBooleanArray = (BooleanArray)arrowArray;
ReadOnlyMemory <byte> valueBuffer = arrowBooleanArray.ValueBuffer.Memory;
ReadOnlyMemory <byte> nullBitMapBuffer = arrowBooleanArray.NullBitmapBuffer.Memory;
dataFrameColumn = new PrimitiveColumn <bool>(field.Name, valueBuffer, nullBitMapBuffer, arrowArray.Length, arrowArray.NullCount);
break;
case ArrowTypeId.Double:
PrimitiveArray <double> arrowDoubleArray = (PrimitiveArray <double>)arrowArray;
ReadOnlyMemory <byte> doubleValueBuffer = arrowDoubleArray.ValueBuffer.Memory;
ReadOnlyMemory <byte> doubleNullBitMapBuffer = arrowDoubleArray.NullBitmapBuffer.Memory;
dataFrameColumn = new PrimitiveColumn <double>(field.Name, doubleValueBuffer, doubleNullBitMapBuffer, arrowArray.Length, arrowArray.NullCount);
break;
case ArrowTypeId.Float:
PrimitiveArray <float> arrowFloatArray = (PrimitiveArray <float>)arrowArray;
ReadOnlyMemory <byte> floatValueBuffer = arrowFloatArray.ValueBuffer.Memory;
ReadOnlyMemory <byte> floatNullBitMapBuffer = arrowFloatArray.NullBitmapBuffer.Memory;
dataFrameColumn = new PrimitiveColumn <float>(field.Name, floatValueBuffer, floatNullBitMapBuffer, arrowArray.Length, arrowArray.NullCount);
break;
case ArrowTypeId.Int8:
PrimitiveArray <sbyte> arrowsbyteArray = (PrimitiveArray <sbyte>)arrowArray;
ReadOnlyMemory <byte> sbyteValueBuffer = arrowsbyteArray.ValueBuffer.Memory;
ReadOnlyMemory <byte> sbyteNullBitMapBuffer = arrowsbyteArray.NullBitmapBuffer.Memory;
dataFrameColumn = new PrimitiveColumn <sbyte>(field.Name, sbyteValueBuffer, sbyteNullBitMapBuffer, arrowArray.Length, arrowArray.NullCount);
break;
case ArrowTypeId.Int16:
PrimitiveArray <short> arrowshortArray = (PrimitiveArray <short>)arrowArray;
ReadOnlyMemory <byte> shortValueBuffer = arrowshortArray.ValueBuffer.Memory;
ReadOnlyMemory <byte> shortNullBitMapBuffer = arrowshortArray.NullBitmapBuffer.Memory;
dataFrameColumn = new PrimitiveColumn <short>(field.Name, shortValueBuffer, shortNullBitMapBuffer, arrowArray.Length, arrowArray.NullCount);
break;
case ArrowTypeId.Int32:
PrimitiveArray <int> arrowIntArray = (PrimitiveArray <int>)arrowArray;
ReadOnlyMemory <byte> intValueBuffer = arrowIntArray.ValueBuffer.Memory;
ReadOnlyMemory <byte> intNullBitMapBuffer = arrowIntArray.NullBitmapBuffer.Memory;
dataFrameColumn = new PrimitiveColumn <int>(field.Name, intValueBuffer, intNullBitMapBuffer, arrowArray.Length, arrowArray.NullCount);
break;
case ArrowTypeId.Int64:
PrimitiveArray <long> arrowLongArray = (PrimitiveArray <long>)arrowArray;
ReadOnlyMemory <byte> longValueBuffer = arrowLongArray.ValueBuffer.Memory;
ReadOnlyMemory <byte> longNullBitMapBuffer = arrowLongArray.NullBitmapBuffer.Memory;
dataFrameColumn = new PrimitiveColumn <long>(field.Name, longValueBuffer, longNullBitMapBuffer, arrowArray.Length, arrowArray.NullCount);
break;
case ArrowTypeId.String:
StringArray stringArray = (StringArray)arrowArray;
ReadOnlyMemory <byte> dataMemory = stringArray.ValueBuffer.Memory;
ReadOnlyMemory <byte> offsetsMemory = stringArray.ValueOffsetsBuffer.Memory;
ReadOnlyMemory <byte> nullMemory = stringArray.NullBitmapBuffer.Memory;
dataFrameColumn = new ArrowStringColumn(field.Name, dataMemory, offsetsMemory, nullMemory, stringArray.Length, stringArray.NullCount);
break;
case ArrowTypeId.UInt8:
PrimitiveArray <byte> arrowbyteArray = (PrimitiveArray <byte>)arrowArray;
ReadOnlyMemory <byte> byteValueBuffer = arrowbyteArray.ValueBuffer.Memory;
ReadOnlyMemory <byte> byteNullBitMapBuffer = arrowbyteArray.NullBitmapBuffer.Memory;
dataFrameColumn = new PrimitiveColumn <byte>(field.Name, byteValueBuffer, byteNullBitMapBuffer, arrowArray.Length, arrowArray.NullCount);
break;
case ArrowTypeId.UInt16:
PrimitiveArray <ushort> arrowUshortArray = (PrimitiveArray <ushort>)arrowArray;
ReadOnlyMemory <byte> ushortValueBuffer = arrowUshortArray.ValueBuffer.Memory;
ReadOnlyMemory <byte> ushortNullBitMapBuffer = arrowUshortArray.NullBitmapBuffer.Memory;
dataFrameColumn = new PrimitiveColumn <ushort>(field.Name, ushortValueBuffer, ushortNullBitMapBuffer, arrowArray.Length, arrowArray.NullCount);
break;
case ArrowTypeId.UInt32:
PrimitiveArray <uint> arrowUintArray = (PrimitiveArray <uint>)arrowArray;
ReadOnlyMemory <byte> uintValueBuffer = arrowUintArray.ValueBuffer.Memory;
ReadOnlyMemory <byte> uintNullBitMapBuffer = arrowUintArray.NullBitmapBuffer.Memory;
dataFrameColumn = new PrimitiveColumn <uint>(field.Name, uintValueBuffer, uintNullBitMapBuffer, arrowArray.Length, arrowArray.NullCount);
break;
case ArrowTypeId.UInt64:
PrimitiveArray <ulong> arrowUlongArray = (PrimitiveArray <ulong>)arrowArray;
ReadOnlyMemory <byte> ulongValueBuffer = arrowUlongArray.ValueBuffer.Memory;
ReadOnlyMemory <byte> ulongNullBitMapBuffer = arrowUlongArray.NullBitmapBuffer.Memory;
dataFrameColumn = new PrimitiveColumn <ulong>(field.Name, ulongValueBuffer, ulongNullBitMapBuffer, arrowArray.Length, arrowArray.NullCount);
break;
case ArrowTypeId.Decimal:
case ArrowTypeId.Binary:
case ArrowTypeId.Date32:
case ArrowTypeId.Date64:
case ArrowTypeId.Dictionary:
case ArrowTypeId.FixedSizedBinary:
case ArrowTypeId.HalfFloat:
case ArrowTypeId.Interval:
case ArrowTypeId.List:
case ArrowTypeId.Map:
case ArrowTypeId.Null:
case ArrowTypeId.Struct:
case ArrowTypeId.Time32:
case ArrowTypeId.Time64:
default:
throw new NotImplementedException(nameof(fieldType.Name));
}
_table.InsertColumn(ColumnCount, dataFrameColumn);
fieldIndex++;
}
}
public virtual PrimitiveColumn <bool> LessThan(BaseColumn column)
{
throw new NotImplementedException();
}
public void InsertColumn(int columnIndex, BaseColumn column)
{
_table.InsertColumn(columnIndex, column, this);
OnColumnsChanged();
}
// TODO: Merge API with an "On" parameter that merges on a column common to 2 dataframes
/// <summary>
/// Merge DataFrames with a database style join
/// </summary>
/// <param name="other"></param>
/// <param name="leftJoinColumn"></param>
/// <param name="rightJoinColumn"></param>
/// <param name="leftSuffix"></param>
/// <param name="rightSuffix"></param>
/// <param name="joinAlgorithm"></param>
/// <returns></returns>
public DataFrame Merge <TKey>(DataFrame other, string leftJoinColumn, string rightJoinColumn, string leftSuffix = "_left", string rightSuffix = "_right", JoinAlgorithm joinAlgorithm = JoinAlgorithm.Left)
{
// A simple hash join
DataFrame ret = new DataFrame();
DataFrame leftDataFrame = this;
DataFrame rightDataFrame = other;
// The final table size is not known until runtime
long rowNumber = 0;
PrimitiveColumn <long> leftRowIndices = new PrimitiveColumn <long>("LeftIndices");
PrimitiveColumn <long> rightRowIndices = new PrimitiveColumn <long>("RightIndices");
if (joinAlgorithm == JoinAlgorithm.Left)
{
// First hash other dataframe on the rightJoinColumn
BaseColumn otherColumn = other[rightJoinColumn];
Dictionary <TKey, ICollection <long> > multimap = otherColumn.GroupColumnValues <TKey>();
// Go over the records in this dataframe and match with the dictionary
BaseColumn thisColumn = this[leftJoinColumn];
for (long i = 0; i < thisColumn.Length; i++)
{
var thisColumnValue = thisColumn[i];
TKey thisColumnValueOrDefault = (TKey)(thisColumnValue == null ? default(TKey) : thisColumnValue);
if (multimap.TryGetValue(thisColumnValueOrDefault, out ICollection <long> rowNumbers))
{
foreach (long row in rowNumbers)
{
if (thisColumnValue == null)
{
// Match only with nulls in otherColumn
if (otherColumn[row] == null)
{
leftRowIndices.Append(i);
rightRowIndices.Append(row);
}
}
else
{
// Cannot match nulls in otherColumn
if (otherColumn[row] != null)
{
leftRowIndices.Append(i);
rightRowIndices.Append(row);
}
}
}
}
else
{
leftRowIndices.Append(i);
rightRowIndices.Append(null);
}
}
}
else if (joinAlgorithm == JoinAlgorithm.Right)
{
BaseColumn thisColumn = this[leftJoinColumn];
Dictionary <TKey, ICollection <long> > multimap = thisColumn.GroupColumnValues <TKey>();
BaseColumn otherColumn = other[rightJoinColumn];
for (long i = 0; i < otherColumn.Length; i++)
{
var otherColumnValue = otherColumn[i];
TKey otherColumnValueOrDefault = (TKey)(otherColumnValue == null ? default(TKey) : otherColumnValue);
if (multimap.TryGetValue(otherColumnValueOrDefault, out ICollection <long> rowNumbers))
{
foreach (long row in rowNumbers)
{
if (otherColumnValue == null)
{
if (thisColumn[row] == null)
{
leftRowIndices.Append(row);
rightRowIndices.Append(i);
}
}
else
{
if (thisColumn[row] != null)
{
leftRowIndices.Append(row);
rightRowIndices.Append(i);
}
}
}
}
else
{
leftRowIndices.Append(null);
rightRowIndices.Append(i);
}
}
}
else if (joinAlgorithm == JoinAlgorithm.Inner)
{
// Hash the column with the smaller RowCount
long leftRowCount = RowCount;
long rightRowCount = other.RowCount;
DataFrame longerDataFrame = leftRowCount < rightRowCount ? other : this;
DataFrame shorterDataFrame = ReferenceEquals(longerDataFrame, this) ? other : this;
BaseColumn hashColumn = (leftRowCount < rightRowCount) ? this[leftJoinColumn] : other[rightJoinColumn];
BaseColumn otherColumn = ReferenceEquals(hashColumn, this[leftJoinColumn]) ? other[rightJoinColumn] : this[leftJoinColumn];
Dictionary <TKey, ICollection <long> > multimap = hashColumn.GroupColumnValues <TKey>();
for (long i = 0; i < otherColumn.Length; i++)
{
var otherColumnValue = otherColumn[i];
TKey otherColumnValueOrDefault = (TKey)(otherColumnValue == null ? default(TKey) : otherColumnValue);
if (multimap.TryGetValue(otherColumnValueOrDefault, out ICollection <long> rowNumbers))
{
foreach (long row in rowNumbers)
{
if (otherColumnValue == null)
{
if (hashColumn[row] == null)
{
leftRowIndices.Append(row);
rightRowIndices.Append(i);
}
}
else
{
if (hashColumn[row] != null)
{
leftRowIndices.Append(row);
rightRowIndices.Append(i);
}
}
}
}
}
leftDataFrame = shorterDataFrame;
rightDataFrame = longerDataFrame;
}
else if (joinAlgorithm == JoinAlgorithm.FullOuter)
{
BaseColumn otherColumn = other[rightJoinColumn];
Dictionary <TKey, ICollection <long> > multimap = otherColumn.GroupColumnValues <TKey>();
Dictionary <TKey, long> intersection = new Dictionary <TKey, long>(EqualityComparer <TKey> .Default);
// Go over the records in this dataframe and match with the dictionary
BaseColumn thisColumn = this[leftJoinColumn];
for (long i = 0; i < thisColumn.Length; i++)
{
var thisColumnValue = thisColumn[i];
TKey thisColumnValueOrDefault = (TKey)(thisColumnValue == null ? default(TKey) : thisColumnValue);
if (multimap.TryGetValue(thisColumnValueOrDefault, out ICollection <long> rowNumbers))
{
foreach (long row in rowNumbers)
{
if (thisColumnValue == null)
{
// Has to match only with nulls in otherColumn
if (otherColumn[row] == null)
{
leftRowIndices.Append(i);
rightRowIndices.Append(row);
if (!intersection.ContainsKey(thisColumnValueOrDefault))
{
intersection.Add(thisColumnValueOrDefault, rowNumber);
}
}
}
else
{
// Cannot match to nulls in otherColumn
if (otherColumn[row] != null)
{
leftRowIndices.Append(i);
rightRowIndices.Append(row);
if (!intersection.ContainsKey(thisColumnValueOrDefault))
{
intersection.Add(thisColumnValueOrDefault, rowNumber);
}
}
}
}
}
else
{
leftRowIndices.Append(i);
rightRowIndices.Append(null);
}
}
for (long i = 0; i < otherColumn.Length; i++)
{
TKey value = (TKey)(otherColumn[i] ?? default(TKey));
if (!intersection.ContainsKey(value))
{
leftRowIndices.Append(null);
rightRowIndices.Append(i);
}
}
}
else
{
throw new NotImplementedException(nameof(joinAlgorithm));
}
for (int i = 0; i < leftDataFrame.ColumnCount; i++)
{
ret.InsertColumn(i, leftDataFrame.Column(i).Clone(leftRowIndices));
}
for (int i = 0; i < rightDataFrame.ColumnCount; i++)
{
BaseColumn column = rightDataFrame.Column(i).Clone(rightRowIndices);
SetSuffixForDuplicatedColumnNames(ret, column, leftSuffix, rightSuffix);
ret.InsertColumn(ret.ColumnCount, column);
}
return(ret);
}
public virtual BaseColumn Xor(BaseColumn column)
{
throw new NotImplementedException();
}
public virtual BaseColumn GreaterThan(BaseColumn column)
{
throw new NotImplementedException();
}
public virtual BaseColumn LessThanOrEqual(BaseColumn column)
{
throw new NotImplementedException();
}
private Delegate CreateGetterDelegate(int col)
{
BaseColumn column = _dataFrame.Column(col);
return(column.GetDataViewGetter(this));
}
// TODO: Merge API with an "On" parameter that merges on a column common to 2 dataframes
/// <summary>
/// Merge DataFrames with a database style join
/// </summary>
/// <param name="other"></param>
/// <param name="leftJoinColumn"></param>
/// <param name="rightJoinColumn"></param>
/// <param name="leftSuffix"></param>
/// <param name="rightSuffix"></param>
/// <param name="joinAlgorithm"></param>
/// <returns></returns>
public DataFrame Merge <TKey>(DataFrame other, string leftJoinColumn, string rightJoinColumn, string leftSuffix = "_left", string rightSuffix = "_right", JoinAlgorithm joinAlgorithm = JoinAlgorithm.Left)
{
// A simple hash join
DataFrame ret = new DataFrame();
PrimitiveColumn <long> emptyMap = new PrimitiveColumn <long>("Empty");
for (int i = 0; i < ColumnCount; i++)
{
// Create empty columns
BaseColumn column = Column(i).Clone(emptyMap);
ret.InsertColumn(ret.ColumnCount, column);
}
for (int i = 0; i < other.ColumnCount; i++)
{
// Create empty columns
BaseColumn column = other.Column(i).Clone(emptyMap);
SetSuffixForDuplicatedColumnNames(ret, column, leftSuffix, rightSuffix);
ret.InsertColumn(ret.ColumnCount, column);
}
// The final table size is not known until runtime
long rowNumber = 0;
if (joinAlgorithm == JoinAlgorithm.Left)
{
// First hash other dataframe on the rightJoinColumn
BaseColumn otherColumn = other[rightJoinColumn];
Dictionary <TKey, ICollection <long> > multimap = otherColumn.GroupColumnValues <TKey>();
// Go over the records in this dataframe and match with the dictionary
BaseColumn thisColumn = this[leftJoinColumn];
for (int c = 0; c < ret.ColumnCount; c++)
{
ret.Column(c).Resize(thisColumn.Length);
}
for (long i = 0; i < thisColumn.Length; i++)
{
if (rowNumber >= thisColumn.Length)
{
for (int c = 0; c < ret.ColumnCount; c++)
{
ret.Column(c).Resize(rowNumber + 1);
}
}
TKey value = (TKey)(thisColumn[i] ?? default(TKey));
if (multimap.TryGetValue(value, out ICollection <long> rowNumbers))
{
foreach (long row in rowNumbers)
{
if (thisColumn[i] == null)
{
// Match only with nulls in otherColumn
if (otherColumn[row] == null)
{
AppendForMerge(ret, rowNumber++, this, other, i, row);
}
}
else
{
// Cannot match nulls in otherColumn
if (otherColumn[row] != null)
{
AppendForMerge(ret, rowNumber++, this, other, i, row);
}
}
}
}
else
{
AppendForMerge(ret, rowNumber++, this, other, i, -1);
}
}
ret._table.RowCount = rowNumber;
}
else if (joinAlgorithm == JoinAlgorithm.Right)
{
BaseColumn thisColumn = this[leftJoinColumn];
Dictionary <TKey, ICollection <long> > multimap = thisColumn.GroupColumnValues <TKey>();
BaseColumn otherColumn = other[rightJoinColumn];
for (int c = 0; c < ret.ColumnCount; c++)
{
ret.Column(c).Resize(otherColumn.Length);
}
for (long i = 0; i < otherColumn.Length; i++)
{
if (rowNumber >= otherColumn.Length)
{
for (int c = 0; c < ret.ColumnCount; c++)
{
ret.Column(c).Resize(rowNumber + 1);
}
}
TKey value = (TKey)(otherColumn[i] ?? default(TKey));
if (multimap.TryGetValue(value, out ICollection <long> rowNumbers))
{
foreach (long row in rowNumbers)
{
if (otherColumn[i] == null)
{
if (thisColumn[row] == null)
{
AppendForMerge(ret, rowNumber++, this, other, row, i);
}
}
else
{
if (thisColumn[row] != null)
{
AppendForMerge(ret, rowNumber++, this, other, row, i);
}
}
}
}
else
{
AppendForMerge(ret, rowNumber++, this, other, -1, i);
}
}
ret._table.RowCount = rowNumber;
}
else if (joinAlgorithm == JoinAlgorithm.Inner)
{
// Hash the column with the smaller RowCount
long leftRowCount = RowCount;
long rightRowCount = other.RowCount;
DataFrame longerDataFrame = leftRowCount < rightRowCount ? other : this;
DataFrame shorterDataFrame = ReferenceEquals(longerDataFrame, this) ? other : this;
BaseColumn hashColumn = (leftRowCount < rightRowCount) ? this[leftJoinColumn] : other[rightJoinColumn];
BaseColumn otherColumn = ReferenceEquals(hashColumn, this[leftJoinColumn]) ? other[rightJoinColumn] : this[leftJoinColumn];
Dictionary <TKey, ICollection <long> > multimap = hashColumn.GroupColumnValues <TKey>();
for (int c = 0; c < ret.ColumnCount; c++)
{
ret.Column(c).Resize(1);
}
for (long i = 0; i < otherColumn.Length; i++)
{
if (rowNumber >= ret.Column(0).Length)
{
for (int c = 0; c < ret.ColumnCount; c++)
{
ret.Column(c).Resize(rowNumber + 1);
}
}
TKey value = (TKey)(otherColumn[i] ?? default(TKey));
if (multimap.TryGetValue(value, out ICollection <long> rowNumbers))
{
foreach (long row in rowNumbers)
{
if (otherColumn[i] == null)
{
if (hashColumn[row] == null)
{
AppendForMerge(ret, rowNumber++, this, other, ReferenceEquals(this, shorterDataFrame) ? row : i, ReferenceEquals(this, shorterDataFrame) ? i : row);
}
}
else
{
if (hashColumn[row] != null)
{
AppendForMerge(ret, rowNumber++, this, other, ReferenceEquals(this, shorterDataFrame) ? row : i, ReferenceEquals(this, shorterDataFrame) ? i : row);
}
}
}
}
}
ret._table.RowCount = rowNumber;
}
else if (joinAlgorithm == JoinAlgorithm.FullOuter)
{
BaseColumn otherColumn = other[rightJoinColumn];
Dictionary <TKey, ICollection <long> > multimap = otherColumn.GroupColumnValues <TKey>();
Dictionary <TKey, long> intersection = new Dictionary <TKey, long>(EqualityComparer <TKey> .Default);
// Go over the records in this dataframe and match with the dictionary
BaseColumn thisColumn = this[leftJoinColumn];
for (int c = 0; c < ret.ColumnCount; c++)
{
ret.Column(c).Resize(thisColumn.Length + 1);
}
for (long i = 0; i < thisColumn.Length; i++)
{
if (rowNumber >= thisColumn.Length)
{
for (int c = 0; c < ret.ColumnCount; c++)
{
ret.Column(c).Resize(rowNumber + 1);
}
}
TKey value = (TKey)(thisColumn[i] ?? default(TKey));
if (multimap.TryGetValue(value, out ICollection <long> rowNumbers))
{
foreach (long row in rowNumbers)
{
if (thisColumn[i] == null)
{
// Has to match only with nulls in otherColumn
if (otherColumn[row] == null)
{
AppendForMerge(ret, rowNumber++, this, other, i, row);
if (!intersection.ContainsKey(value))
{
intersection.Add(value, rowNumber);
}
}
}
else
{
// Cannot match to nulls in otherColumn
if (otherColumn[row] != null)
{
AppendForMerge(ret, rowNumber++, this, other, i, row);
if (!intersection.ContainsKey(value))
{
intersection.Add(value, rowNumber);
}
}
}
}
}
else
{
AppendForMerge(ret, rowNumber++, this, other, i, -1);
}
}
for (long i = 0; i < otherColumn.Length; i++)
{
if (rowNumber >= ret.Column(0).Length)
{
for (int c = 0; c < ret.ColumnCount; c++)
{
ret.Column(c).Resize(rowNumber + 1);
}
}
TKey value = (TKey)(otherColumn[i] ?? default(TKey));
if (!intersection.ContainsKey(value))
{
if (rowNumber >= otherColumn.Length)
{
for (int c = 0; c < ret.ColumnCount; c++)
{
ret.Column(c).Resize(rowNumber + 1);
}
}
AppendForMerge(ret, rowNumber++, this, other, -1, i);
}
}
ret._table.RowCount = rowNumber;
}
return(ret);
}
public virtual BaseColumn Xor(BaseColumn column, bool inPlace = false)
{
throw new NotImplementedException();
}
public void InsertColumn(int columnIndex, BaseColumn column) => _table.InsertColumn(columnIndex, column);
public void SetColumn(int columnIndex, BaseColumn column) => _table.SetColumn(columnIndex, column);
public DataFrame Join(DataFrame other, string leftSuffix = "_left", string rightSuffix = "_right", JoinAlgorithm joinAlgorithm = JoinAlgorithm.Left)
{
DataFrame ret = new DataFrame();
if (joinAlgorithm == JoinAlgorithm.Left)
{
for (int i = 0; i < ColumnCount; i++)
{
BaseColumn newColumn = Column(i).Clone();
ret.InsertColumn(ret.ColumnCount, newColumn);
}
long minLength = Math.Min(RowCount, other.RowCount);
PrimitiveColumn <long> mapIndices = new PrimitiveColumn <long>("mapIndices", minLength);
for (long i = 0; i < minLength; i++)
{
mapIndices[i] = i;
}
for (int i = 0; i < other.ColumnCount; i++)
{
BaseColumn newColumn;
if (other.RowCount < RowCount)
{
newColumn = other.Column(i).Clone(numberOfNullsToAppend: RowCount - other.RowCount);
}
else
{
newColumn = other.Column(i).Clone(mapIndices);
}
SetSuffixForDuplicatedColumnNames(ret, newColumn, leftSuffix, rightSuffix);
ret.InsertColumn(ret.ColumnCount, newColumn);
}
}
else if (joinAlgorithm == JoinAlgorithm.Right)
{
long minLength = Math.Min(RowCount, other.RowCount);
PrimitiveColumn <long> mapIndices = new PrimitiveColumn <long>("mapIndices", minLength);
for (long i = 0; i < minLength; i++)
{
mapIndices[i] = i;
}
for (int i = 0; i < ColumnCount; i++)
{
BaseColumn newColumn;
if (RowCount < other.RowCount)
{
newColumn = Column(i).Clone(numberOfNullsToAppend: other.RowCount - RowCount);
}
else
{
newColumn = Column(i).Clone(mapIndices);
}
ret.InsertColumn(ret.ColumnCount, newColumn);
}
for (int i = 0; i < other.ColumnCount; i++)
{
BaseColumn newColumn = other.Column(i).Clone();
SetSuffixForDuplicatedColumnNames(ret, newColumn, leftSuffix, rightSuffix);
ret.InsertColumn(ret.ColumnCount, newColumn);
}
}
else if (joinAlgorithm == JoinAlgorithm.FullOuter)
{
long newRowCount = Math.Max(RowCount, other.RowCount);
long numberOfNulls = newRowCount - RowCount;
for (int i = 0; i < ColumnCount; i++)
{
BaseColumn newColumn = Column(i).Clone(numberOfNullsToAppend: numberOfNulls);
ret.InsertColumn(ret.ColumnCount, newColumn);
}
numberOfNulls = newRowCount - other.RowCount;
for (int i = 0; i < other.ColumnCount; i++)
{
BaseColumn newColumn = other.Column(i).Clone(numberOfNullsToAppend: numberOfNulls);
SetSuffixForDuplicatedColumnNames(ret, newColumn, leftSuffix, rightSuffix);
ret.InsertColumn(ret.ColumnCount, newColumn);
}
}
else if (joinAlgorithm == JoinAlgorithm.Inner)
{
long newRowCount = Math.Min(RowCount, other.RowCount);
PrimitiveColumn <long> mapIndices = new PrimitiveColumn <long>("mapIndices", newRowCount);
for (long i = 0; i < newRowCount; i++)
{
mapIndices[i] = i;
}
for (int i = 0; i < ColumnCount; i++)
{
BaseColumn newColumn = Column(i).Clone(mapIndices);
ret.InsertColumn(ret.ColumnCount, newColumn);
}
for (int i = 0; i < other.ColumnCount; i++)
{
BaseColumn newColumn = other.Column(i).Clone(mapIndices);
SetSuffixForDuplicatedColumnNames(ret, newColumn, leftSuffix, rightSuffix);
ret.InsertColumn(ret.ColumnCount, newColumn);
}
}
return(ret);
}
public void SetColumn(int columnIndex, BaseColumn column)
{
_table.SetColumn(columnIndex, column);
OnColumnsChanged();
}
public virtual PrimitiveColumn <bool> GreaterThanOrEqual(BaseColumn column)
{
throw new NotImplementedException();
}
