public void WriteDataWithAnnotationColumnsTest()
{
// main data
IDataWithAnnotationColumns mdata = PerseusFactory.CreateDataWithAnnotationColumns();
// annotation columns
mdata.AddStringColumn("strcol1", "this is stringcol1", new[] { "1", "2" });
mdata.AddStringColumn("strcol2", "", new[] { "", "hallo" });
mdata.AddNumericColumn("numcol", "", new[] { 1.0, 2.0 });
mdata.AddMultiNumericColumn("multnumcol", "this is multnumcol", new[] { new[] { -2.0, 2.0 }, new double[] {} });
mdata.AddCategoryColumn("catcol", "", new[] { new[] { "cat1", "cat1.1" }, new[] { "cat2", "cat1" } });
string mdataStr;
using (MemoryStream memstream = new MemoryStream())
using (StreamWriter writer = new StreamWriter(memstream)) {
PerseusUtils.WriteDataWithAnnotationColumns(mdata, writer);
writer.Flush();
mdataStr = Encoding.UTF8.GetString(memstream.ToArray());
}
IMatrixData mdata3 = PerseusFactory.CreateMatrixData();
PerseusUtils.ReadMatrix(mdata3, new ProcessInfo(new Settings(), status => { }, progress => { }, 1, i => { }),
() => {
StreamReader tmpStream = new StreamReader(new MemoryStream(Encoding.UTF8.GetBytes(mdataStr)));
return(tmpStream);
}, "matrix1", '\t');
IDataWithAnnotationColumns mdata2 = mdata3;
Assert.AreEqual(2, mdata2.RowCount);
Assert.AreEqual(2, mdata2.StringColumnCount);
Assert.AreEqual(1, mdata2.NumericColumnCount);
Assert.AreEqual(1, mdata2.CategoryColumnCount);
Assert.AreEqual(1, mdata2.MultiNumericColumnCount);
Assert.AreEqual("hallo", mdata2.StringColumns[mdata2.StringColumnNames.FindIndex(col => col.Equals("strcol2"))][1]);
}
public static void ProcessData(IDataWithAnnotationColumns mdata, IAnnotationProvider annotationProvider, Parameters para, ProcessInfo processInfo)
{
string[] baseIds = GetBaseIds(mdata, annotationProvider, para);
bool success = ProcessDataAddAnnotation(mdata.RowCount, annotationProvider, para, baseIds, processInfo, out string[] name,
out int[] catColInds,
out int[] textColInds, out int[] numColInds, out string[][][] catCols, out string[][] textCols,
out double[][] numCols);
if (!success)
{
return;
}
for (int i = 0; i < catCols.Length; i++)
{
mdata.AddCategoryColumn(name[catColInds[i]], "", catCols[i]);
}
for (int i = 0; i < textCols.Length; i++)
{
mdata.AddStringColumn(name[textColInds[i]], "", textCols[i]);
}
for (int i = 0; i < numCols.Length; i++)
{
mdata.AddNumericColumn(name[numColInds[i]], "", numCols[i]);
}
}
private static void AddNumericColumns(IDataWithAnnotationColumns mdata1, IDataWithAnnotationColumns mdata2,
Parameters parameters, IList <int[]> indexMap, IDataWithAnnotationColumns result)
{
Func <double[], double> avNumerical = GetAveraging(parameters.GetParam <int>("Combine numerical values").Value);
int[] numCols = parameters.GetParam <int[]>("Numerical columns").Value;
if (avNumerical != null)
{
double[][] newNumericalColumns = new double[numCols.Length][];
string[] newNumColNames = new string[numCols.Length];
for (int i = 0; i < numCols.Length; i++)
{
double[] oldCol = mdata2.NumericColumns[numCols[i]];
newNumColNames[i] = mdata2.NumericColumnNames[numCols[i]];
newNumericalColumns[i] = new double[mdata1.RowCount];
for (int j = 0; j < mdata1.RowCount; j++)
{
int[] inds = indexMap[j];
List <double> values = new List <double>();
foreach (int ind in inds)
{
double v = oldCol[ind];
if (!double.IsNaN(v))
{
values.Add(v);
}
}
newNumericalColumns[i][j] = values.Count == 0 ? double.NaN : avNumerical(values.ToArray());
}
}
for (int i = 0; i < numCols.Length; i++)
{
result.AddNumericColumn(newNumColNames[i], "", newNumericalColumns[i]);
}
}
else
{
double[][][] newMultiNumericalColumns = new double[numCols.Length][][];
string[] newMultiNumColNames = new string[numCols.Length];
for (int i = 0; i < numCols.Length; i++)
{
double[] oldCol = mdata2.NumericColumns[numCols[i]];
newMultiNumColNames[i] = mdata2.NumericColumnNames[numCols[i]];
newMultiNumericalColumns[i] = new double[mdata1.RowCount][];
for (int j = 0; j < mdata1.RowCount; j++)
{
int[] inds = indexMap[j];
List <double> values = new List <double>();
foreach (int ind in inds)
{
double v = oldCol[ind];
if (!double.IsNaN(v))
{
values.Add(v);
}
}
newMultiNumericalColumns[i][j] = values.ToArray();
}
}
for (int i = 0; i < numCols.Length; i++)
{
result.AddMultiNumericColumn(newMultiNumColNames[i], "", newMultiNumericalColumns[i]);
}
}
}