public DataFrameViewGroupResults <ImmutableTuple <T1> > TGroupBy <T1>(IEnumerable <int> cols, bool sort = true)
where T1 : IEquatable <T1>, IComparable <T1>
{
int[] order = _rows.Select(c => c).ToArray();
var icols = cols.ToArray();
var scols = icols.Select(c => Schema.GetColumnName(c)).ToArray();
return(DataFrameGrouping.TGroupBy(this, order, _columns, icols, true, GetMultiGetterAt <T1>(icols),
ke => ke.ToImTuple(), ke => DataFrameGroupKey.Create(scols, ke)));
}
/// <summary>
/// Groupby.
/// </summary>
public IDataFrameViewGroupResults GroupBy(IEnumerable <int> cols, bool sort = true)
{
return(DataFrameGrouping.GroupBy(this, cols, sort));
}
public static IEnumerable <DataFrame> TJoin <TKey>(
IDataFrameView left, IDataFrameView right,
int[] orderLeft, int[] orderRight,
TKey[] keysLeft, TKey[] keysRight,
int[] icolsLeft, int[] icolsRight,
string leftSuffix, string rightSuffix,
JoinStrategy joinType,
Func <TKey, DataFrameGroupKey[]> funcLeft,
Func <TKey, DataFrameGroupKey[]> funcRight)
where TKey : IEquatable <TKey>, IComparable <TKey>
{
var groupLeft = DataFrameGrouping.TGroupBy <TKey>(left, orderLeft, keysLeft, icolsLeft, funcLeft);
var groupRight = DataFrameGrouping.TGroupBy <TKey>(right, orderRight, keysRight, icolsRight, funcRight);
var iterLeft = groupLeft.GetEnumerator();
var iterRight = groupRight.GetEnumerator();
bool contLeft = iterLeft.MoveNext();
bool contRight = iterRight.MoveNext();
leftSuffix = string.IsNullOrEmpty(leftSuffix) ? string.Empty : leftSuffix;
rightSuffix = string.IsNullOrEmpty(rightSuffix) ? string.Empty : rightSuffix;
var newColsLeft = left.Columns.Select(c => c + leftSuffix).ToArray();
var newColsRight = right.Columns.Select(c => c + rightSuffix).ToArray();
var existsCols = new HashSet <string>(newColsLeft);
for (int i = 0; i < newColsRight.Length; ++i)
{
while (existsCols.Contains(newColsRight[i]))
{
newColsRight[i] += "_y";
}
existsCols.Add(newColsRight[i]);
}
var newCols = newColsLeft.Concat(newColsRight).ToArray();
int r;
while (contLeft || contRight)
{
r = contLeft && contRight
? iterLeft.Current.Key.CompareTo(iterRight.Current.Key)
: (contRight ? 1 : -1);
if (r < 0)
{
if (joinType == JoinStrategy.Left || joinType == JoinStrategy.Outer)
{
var df = iterLeft.Current.Value.Copy();
if (!string.IsNullOrEmpty(leftSuffix))
{
df.RenameColumns(newColsLeft);
}
for (int i = 0; i < right.ColumnCount; ++i)
{
var kind = right.SchemaI.GetColumnType(i);
var col = df.AddColumn(newColsRight[i], kind, df.Length);
df.GetColumn(col).Set(DataFrameMissingValue.GetMissingOrDefaultMissingValue(kind));
}
yield return(df);
}
contLeft = iterLeft.MoveNext();
}
else if (r > 0)
{
if (joinType == JoinStrategy.Right || joinType == JoinStrategy.Outer)
{
var df = iterRight.Current.Value.Copy();
df.RenameColumns(newColsRight);
for (int i = 0; i < left.ColumnCount; ++i)
{
var kind = left.SchemaI.GetColumnType(i);
var col = df.AddColumn(newColsLeft[i], kind, df.Length);
df.GetColumn(col).Set(DataFrameMissingValue.GetMissingOrDefaultMissingValue(kind));
}
df.OrderColumns(newCols);
yield return(df);
}
contRight = iterRight.MoveNext();
}
else
{
var dfLeft = iterLeft.Current.Value.Copy();
var dfRight = iterRight.Current.Value.Copy();
if (!string.IsNullOrEmpty(leftSuffix))
{
dfLeft.RenameColumns(newColsLeft);
}
dfRight.RenameColumns(newColsRight);
var vleft = dfLeft.Multiply(dfRight.Length, MultiplyStrategy.Block).Copy();
var vright = dfRight.Multiply(dfLeft.Length, MultiplyStrategy.Row).Copy();
for (int i = 0; i < vright.ColumnCount; ++i)
{
vleft.AddColumn(newColsRight[i], vright.GetColumn(i));
}
yield return(vleft);
contLeft = iterLeft.MoveNext();
contRight = iterRight.MoveNext();
}
}
}