public static ICube Pivot(this ICube cube, string[] fieldsToAggregateBy, AggregationAction aggregationAction)
{
//for now, aggregate only works on numerical fields and performs a sum
foreach (var fieldToAggregateBy in fieldsToAggregateBy)
{
if (!cube.DataTypes.ContainsKey(fieldToAggregateBy))
{
throw new Exception($"Cannot aggregate on field {fieldToAggregateBy} as it is not present");
}
}
var types = cube.DataTypes.Keys.ToList();
var ixs = fieldsToAggregateBy.Select(f => types.IndexOf(f)).ToArray();
var rows = cube.GetAllRows();
var distinctValues = rows.Select(x => string.Join("~", ixs.Select(ix => x.MetaData[ix]?.ToString() ?? string.Empty))).Distinct();
var outCube = new ResultCube();
var oT = new Dictionary <string, Type>();
foreach (var fieldToAggregateBy in fieldsToAggregateBy)
{
oT.Add(fieldToAggregateBy, cube.DataTypes[fieldToAggregateBy]);
}
outCube.Initialize(oT);
var aggData = new Dictionary <string, double>();
var aggDataCount = new Dictionary <string, int>();
var metaDict = new Dictionary <string, object[]>();
foreach (var row in rows)
{
var rowKey = string.Join("~", ixs.Select(i => row.MetaData[i]?.ToString() ?? string.Empty));
if (!aggData.ContainsKey(rowKey))
{
if (aggregationAction == AggregationAction.Min)
{
aggData[rowKey] = double.MaxValue;
}
else if (aggregationAction == AggregationAction.Max)
{
aggData[rowKey] = double.MinValue;
}
else
{
aggData[rowKey] = 0;
}
aggDataCount[rowKey] = 0;
var filetedMetaData = new object[ixs.Length];
for (var i = 0; i < ixs.Length; i++)
{
filetedMetaData[i] = row.MetaData[ixs[i]];
}
metaDict[rowKey] = filetedMetaData;
}
switch (aggregationAction)
{
case AggregationAction.Sum:
aggData[rowKey] += row.Value;
break;
case AggregationAction.Average:
aggData[rowKey] += row.Value;
aggDataCount[rowKey]++;
break;
case AggregationAction.Min:
aggData[rowKey] = System.Math.Min(aggData[rowKey], row.Value);
break;
case AggregationAction.Max:
aggData[rowKey] = System.Math.Max(aggData[rowKey], row.Value);
break;
}
}
//final post-processing for average
foreach (var rowKey in aggData.Keys.ToList())
{
var rowDict = new Dictionary <string, object>();
if (aggregationAction == AggregationAction.Average)
{
aggData[rowKey] /= aggDataCount[rowKey];
}
outCube.AddRow(metaDict[rowKey], aggData[rowKey]);
}
return(outCube);
}
public static ICube Merge(this ICube baseCube, ICube otherCube)
{
//add check that common fields are of them same type...
foreach (var kv in baseCube.DataTypes.Where(x => otherCube.DataTypes.Keys.Contains(x.Key)))
{
if (kv.Value != otherCube.DataTypes[kv.Key])
{
throw new Exception($"Data types dont match for field {kv.Key}");
}
}
var newDataTypes = new Dictionary <string, Type>();
foreach (var kv in baseCube.DataTypes)
{
newDataTypes[kv.Key] = kv.Value;
}
foreach (var kv in otherCube.DataTypes)
{
newDataTypes[kv.Key] = kv.Value;
}
var o = new ResultCube();
o.Initialize(newDataTypes);
var baseRows = baseCube.GetAllRows().ToArray();
var otherRows = otherCube.GetAllRows().ToArray();
var baseIx = baseCube.DataTypes.Keys.Select(k => o.GetColumnIndex(k)).ToArray();
var otherIx = otherCube.DataTypes.Keys.Select(k => o.GetColumnIndex(k)).ToArray();
var cleanRow = o.DataTypes.Select(kv => GetDefaultValue(kv.Value)).ToArray();
for (var i = 0; i < baseRows.Length; i++)
{
var row = new object[cleanRow.Length];
Array.Copy(cleanRow, row, row.Length);
var br = baseRows[i];
for (var j = 0; j < baseIx.Length; j++)
{
row[baseIx[j]] = br.MetaData[j];
}
o.AddRow(row, br.Value);
}
for (var i = 0; i < otherRows.Length; i++)
{
var br = otherRows[i];
var row = new object[cleanRow.Length];
Array.Copy(cleanRow, row, row.Length);
for (var j = 0; j < baseIx.Length; j++)
{
row[baseIx[j]] = br.MetaData[j];
}
o.AddRow(row, br.Value);
}
return(o);
}
