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 AddCategoricalColumns(IDataWithAnnotationColumns mdata1, IDataWithAnnotationColumns mdata2,
Parameters parameters, IList <int[]> indexMap, IDataWithAnnotationColumns result)
{
int[] catCols = parameters.GetParam <int[]>("Categorical columns").Value;
string[][][] newCatColumns = new string[catCols.Length][][];
string[] newCatColNames = new string[catCols.Length];
for (int i = 0; i < catCols.Length; i++)
{
string[][] oldCol = mdata2.GetCategoryColumnAt(catCols[i]);
newCatColNames[i] = mdata2.CategoryColumnNames[catCols[i]];
newCatColumns[i] = new string[mdata1.RowCount][];
for (int j = 0; j < mdata1.RowCount; j++)
{
int[] inds = indexMap[j];
List <string[]> values = new List <string[]>();
foreach (int ind in inds)
{
string[] v = oldCol[ind];
if (v.Length > 0)
{
values.Add(v);
}
}
newCatColumns[i][j] = values.Count == 0
? new string[0]
: ArrayUtils.UniqueValues(ArrayUtils.Concat(values.ToArray()));
}
}
for (int i = 0; i < catCols.Length; i++)
{
result.AddCategoryColumn(newCatColNames[i], "", newCatColumns[i]);
}
}
private static void AddIndicator(IDataWithAnnotationColumns result, IData mdata2, int[][] indexMap)
{
string[][] indicatorCol = new string[indexMap.Length][];
for (int i = 0; i < indexMap.Length; i++)
{
indicatorCol[i] = indexMap[i].Length > 0 ? new[] { "+" } : new string[0];
}
result.AddCategoryColumn(mdata2.Name, "", indicatorCol);
}
