/// <summary>
/// Convert pandas DataFrame to <see cref="DataTable"/>
/// </summary>
/// <param name="pyObject">Pandas DataFrame object</param>
/// <returns>Result table</returns>
public static DataTable DataFrameToDataTable(dynamic pyObject)
{
if (pyObject == null)
{
throw new ArgumentNullException(nameof(pyObject));
}
using (var utilities = new PythonUtils())
{
if (!utilities.IsDataFrame(pyObject))
{
throw new ArgumentException("Object is not a pandas dataframe", nameof(pyObject));
}
//Define columns info
var columnNames = utilities.GetDataFrameColumnNames(pyObject);
var columns = new ColumnInfo[columnNames.Length];
for (var i = 0; i < columns.Length; i++)
{
var name = columnNames[i];
var type = utilities.GetTypeCode(pyObject, i);
columns[i] = new ColumnInfo(utilities, name, type);
}
//Define result table columns
var dataTable = new DataTable("Result");
foreach (var column in columns)
{
var name = column.Name;
var dataColumn = new DataColumn(name)
{
Caption = name,
DataType = column.ManagedType,
AllowDBNull = true
};
dataTable.Columns.Add(dataColumn);
}
//Iterate for each row
var rowCount = ((PyObject)pyObject.index).Length();
for (long i = 0; i < rowCount; i++)
{
var row = dataTable.NewRow();
for (var j = 0; j < columns.Length; j++)
{
var column = columns[j];
var pyValue = utilities.GetDataFrameItem(pyObject, i, j);
var value = pyValue == null ? null : column.GetManagedValue(pyValue);
row[j] = value ?? DBNull.Value;
}
dataTable.Rows.Add(row);
}
return(dataTable);
}
}
/// <summary>
/// Create an instance of <see cref="ColumnInfo"/>
/// </summary>
/// <param name="utils">Python utilities</param>
/// <param name="name">Column name</param>
/// <param name="pyType">Column type</param>
public ColumnInfo(PythonUtils utils, string name, PyType pyType)
{
_utils = utils ?? throw new ArgumentNullException(nameof(utils));
if (string.IsNullOrWhiteSpace(name))
{
throw new ArgumentException("Name can't be null or empty", nameof(name));
}
Name = name;
PythonType = pyType;
switch (PythonType)
{
case PyType.Undefined:
ManagedType = typeof(string);
break;
case PyType.BoolType:
ManagedType = typeof(bool);
break;
case PyType.Int8Type:
ManagedType = typeof(sbyte);
break;
case PyType.Uint8Type:
ManagedType = typeof(byte);
break;
case PyType.Int16Type:
ManagedType = typeof(short);
break;
case PyType.Uint16Type:
ManagedType = typeof(ushort);
break;
case PyType.Uint32Type:
ManagedType = typeof(uint);
break;
case PyType.Int64Type:
ManagedType = typeof(long);
break;
case PyType.Uint64Type:
ManagedType = typeof(ulong);
break;
case PyType.Int32Type:
ManagedType = typeof(int);
break;
case PyType.Float16Type:
ManagedType = typeof(float);
break;
case PyType.Float32Type:
ManagedType = typeof(float);
break;
case PyType.Float64Type:
ManagedType = typeof(double);
break;
case PyType.Complex128Type:
ManagedType = typeof(string);
break;
case PyType.Complex64Type:
ManagedType = typeof(string);
break;
case PyType.ObjectType:
ManagedType = typeof(string);
break;
case PyType.Datetime64Type:
ManagedType = typeof(DateTime);
break;
case PyType.Timedelta64Type:
ManagedType = typeof(TimeSpan);
break;
default:
throw new ArgumentOutOfRangeException();
}
}
