/// <summary>
/// Using the dependencyMapping and included transforms, determines whether every
/// transform present only consumes columns produced by a lower- or same-level transform,
/// or existed in the original dataset. Note, a column could be produced by a
/// transform on the same level, such as in multipart (atomic group) transforms.
/// </summary>
public static bool AreColumnsConsistent(TransformInference.SuggestedTransform[] includedTransforms,
AutoInference.DependencyMap dependencyMapping)
{
foreach (var transform in includedTransforms)
{
foreach (var colConsumed in transform.RoutingStructure.ColumnsConsumed)
{
AutoInference.LevelDependencyMap ldm = dependencyMapping[transform.RoutingStructure.Level];
var colInfo = ldm.Keys.FirstOrDefault(k => k.Name == colConsumed.Name);
// Consumed column does not exist at this sublevel. Since we never drop columns
// it will not exist at any lower levels, either. Thus, problem with column consumption.
if (colInfo.Name == null)
{
return(false);
}
// If this column could have been produced by a transform, make sure at least one
// of the possible producer transforms in in our included transforms list.
if (ldm[colInfo].Count > 0 && !ldm[colInfo].Any(t => includedTransforms.Contains(t)))
{
return(false);
}
}
}
// Passed all tests
return(true);
}
/// <summary>
/// Creates a dictionary mapping column names to the transforms which could have produced them.
/// </summary>
public static AutoInference.LevelDependencyMap ComputeColumnResponsibilities(IDataView transformedData,
TransformInference.SuggestedTransform[] appliedTransforms)
{
var mapping = new AutoInference.LevelDependencyMap();
for (int i = 0; i < transformedData.Schema.ColumnCount; i++)
{
if (transformedData.Schema.IsHidden(i))
{
continue;
}
var colInfo = new AutoInference.ColumnInfo
{
IsHidden = false,
ItemType = transformedData.Schema.GetColumnType(i).ItemType,
Name = transformedData.Schema.GetColumnName(i)
};
mapping.Add(colInfo, appliedTransforms.Where(t =>
t.RoutingStructure.ColumnsProduced.Any(o => o.Name == colInfo.Name &&
o.IsNumeric == transformedData.Schema.GetColumnType(i).ItemType.IsNumber)).ToList());
}
return(mapping);
}
