/// <summary>
/// Returns the max value for the specified metadata kind.
/// The metadata type should be a KeyType with raw type U4.
/// colMax will be set to the first column that has the max value for the specified metadata.
/// If no column has the specified metadata, colMax is set to -1 and the method returns zero.
/// The filter function is called for each column, passing in the schema and the column index, and returns
/// true if the column should be considered, false if the column should be skipped.
/// </summary>
public static uint GetMaxMetadataKind(this Schema schema, out int colMax, string metadataKind, Func <Schema, int, bool> filterFunc = null)
{
uint max = 0;
colMax = -1;
for (int col = 0; col < schema.Count; col++)
{
var columnType = schema.GetMetadataTypeOrNull(metadataKind, col);
if (columnType == null || !columnType.IsKey || columnType.RawKind != DataKind.U4)
{
continue;
}
if (filterFunc != null && !filterFunc(schema, col))
{
continue;
}
uint value = 0;
schema.GetMetadata(metadataKind, col, ref value);
if (max < value)
{
max = value;
colMax = col;
}
}
return(max);
}
public ColumnType GetMetadataTypeOrNull(string kind, int col)
{
_ectx.Check(0 <= col && col < ColumnCount);
if (IsPivot(col) && !ShouldPreserveMetadata(kind))
{
return(null);
}
return(_inputSchema.GetMetadataTypeOrNull(kind, col));
}
/// <summary>
/// Tries to get the metadata kind of the specified type for a column.
/// </summary>
/// <typeparam name="T">The raw type of the metadata, should match the PrimitiveType type</typeparam>
/// <param name="schema">The schema</param>
/// <param name="type">The type of the metadata</param>
/// <param name="kind">The metadata kind</param>
/// <param name="col">The column</param>
/// <param name="value">The value to return, if successful</param>
/// <returns>True if the metadata of the right type exists, false otherwise</returns>
public static bool TryGetMetadata <T>(this Schema schema, PrimitiveType type, string kind, int col, ref T value)
{
Contracts.CheckValue(schema, nameof(schema));
Contracts.CheckValue(type, nameof(type));
var metadataType = schema.GetMetadataTypeOrNull(kind, col);
if (!type.Equals(metadataType))
{
return(false);
}
schema.GetMetadata(kind, col, ref value);
return(true);
}
public static bool HasKeyValues(this Schema schema, int col, int keyCount)
{
if (keyCount == 0)
{
return(false);
}
var type = schema.GetMetadataTypeOrNull(Kinds.KeyValues, col);
return
(type != null &&
type.IsVector &&
type.VectorSize == keyCount &&
type.ItemType.IsText);
}
/// <summary>
/// Returns <c>true</c> if the specified column:
/// * is a vector of length N (including 0)
/// * has a SlotNames metadata
/// * metadata type is VBuffer<ReadOnlyMemory<char>> of length N
/// </summary>
public static bool HasSlotNames(this Schema schema, int col, int vectorSize)
{
if (vectorSize == 0)
{
return(false);
}
var type = schema.GetMetadataTypeOrNull(Kinds.SlotNames, col);
return
(type != null &&
type.IsVector &&
type.VectorSize == vectorSize &&
type.ItemType.IsText);
}
/// <summary>
/// The categoricalFeatures is a vector of the indices of categorical features slots.
/// This vector should always have an even number of elements, and the elements should be parsed in groups of two consecutive numbers.
/// So if its value is the range of numbers: 0,2,3,4,8,9
/// look at it as [0,2],[3,4],[8,9].
/// The way to interpret that is: feature with indices 0, 1, and 2 are one categorical
/// Features with indices 3 and 4 are another categorical. Features 5 and 6 don't appear there, so they are not categoricals.
/// </summary>
public static bool TryGetCategoricalFeatureIndices(Schema schema, int colIndex, out int[] categoricalFeatures)
{
Contracts.CheckValue(schema, nameof(schema));
Contracts.Check(colIndex >= 0, nameof(colIndex));
bool isValid = false;
categoricalFeatures = null;
if (!(schema.GetColumnType(colIndex) is VectorType vecType && vecType.Size > 0))
{
return(isValid);
}
var type = schema.GetMetadataTypeOrNull(MetadataUtils.Kinds.CategoricalSlotRanges, colIndex);
if (type?.RawType == typeof(VBuffer <int>))
{
VBuffer <int> catIndices = default(VBuffer <int>);
schema.GetMetadata(MetadataUtils.Kinds.CategoricalSlotRanges, colIndex, ref catIndices);
VBufferUtils.Densify(ref catIndices);
int columnSlotsCount = vecType.Size;
if (catIndices.Length > 0 && catIndices.Length % 2 == 0 && catIndices.Length <= columnSlotsCount * 2)
{
int previousEndIndex = -1;
isValid = true;
var catIndicesValues = catIndices.GetValues();
for (int i = 0; i < catIndicesValues.Length; i += 2)
{
if (catIndicesValues[i] > catIndicesValues[i + 1] ||
catIndicesValues[i] <= previousEndIndex ||
catIndicesValues[i] >= columnSlotsCount ||
catIndicesValues[i + 1] >= columnSlotsCount)
{
isValid = false;
break;
}
previousEndIndex = catIndicesValues[i + 1];
}
if (isValid)
{
categoricalFeatures = catIndicesValues.ToArray();
}
}
}
return(isValid);
}
/// <summary>
/// Returns the set of column ids which match the value of specified metadata kind.
/// The metadata type should be of type text.
/// </summary>
public static IEnumerable <int> GetColumnSet(this Schema schema, string metadataKind, string value)
{
for (int col = 0; col < schema.Count; col++)
{
var columnType = schema.GetMetadataTypeOrNull(metadataKind, col);
if (columnType != null && columnType.IsText)
{
ReadOnlyMemory <char> val = default;
schema.GetMetadata(metadataKind, col, ref val);
if (ReadOnlyMemoryUtils.EqualsStr(value, val))
{
yield return(col);
}
}
}
}
/// <summary>
/// Returns the set of column ids which match the value of specified metadata kind.
/// The metadata type should be a KeyType with raw type U4.
/// </summary>
public static IEnumerable <int> GetColumnSet(this Schema schema, string metadataKind, uint value)
{
for (int col = 0; col < schema.Count; col++)
{
var columnType = schema.GetMetadataTypeOrNull(metadataKind, col);
if (columnType != null && columnType.IsKey && columnType.RawKind == DataKind.U4)
{
uint val = 0;
schema.GetMetadata(metadataKind, col, ref val);
if (val == value)
{
yield return(col);
}
}
}
}
private static void PrintSchema(TextWriter writer, Arguments args, Schema schema, ITransposeSchema tschema)
{
Contracts.AssertValue(writer);
Contracts.AssertValue(args);
Contracts.AssertValue(schema);
Contracts.AssertValueOrNull(tschema);
#if !CORECLR
if (args.ShowJson)
{
writer.WriteLine("Json Schema not supported.");
return;
}
#endif
int colLim = schema.ColumnCount;
var itw = new IndentedTextWriter(writer, " ");
itw.WriteLine("{0} columns:", colLim);
using (itw.Nest())
{
var names = default(VBuffer <ReadOnlyMemory <char> >);
for (int col = 0; col < colLim; col++)
{
var name = schema.GetColumnName(col);
var type = schema.GetColumnType(col);
var slotType = tschema == null ? null : tschema.GetSlotType(col);
itw.WriteLine("{0}: {1}{2}", name, type, slotType == null ? "" : " (T)");
bool metaVals = args.ShowMetadataValues;
if (metaVals || args.ShowMetadataTypes)
{
ShowMetadata(itw, schema, col, metaVals);
continue;
}
if (!args.ShowSlots)
{
continue;
}
if (!type.IsKnownSizeVector)
{
continue;
}
ColumnType typeNames;
if ((typeNames = schema.GetMetadataTypeOrNull(MetadataUtils.Kinds.SlotNames, col)) == null)
{
continue;
}
if (typeNames.VectorSize != type.VectorSize || !typeNames.ItemType.IsText)
{
Contracts.Assert(false, "Unexpected slot names type");
continue;
}
schema.GetMetadata(MetadataUtils.Kinds.SlotNames, col, ref names);
if (names.Length != type.VectorSize)
{
Contracts.Assert(false, "Unexpected length of slot names vector");
continue;
}
using (itw.Nest())
{
bool verbose = args.Verbose ?? false;
foreach (var kvp in names.Items(all: verbose))
{
if (verbose || !kvp.Value.IsEmpty)
{
itw.WriteLine("{0}:{1}", kvp.Key, kvp.Value);
}
}
}
}
}
}
public ColumnType GetMetadataTypeOrNull(string kind, int col)
{
Contracts.CheckNonEmpty(kind, nameof(kind));
Contracts.CheckParam(0 <= col && col < ColumnCount, nameof(col));
return(_input.GetMetadataTypeOrNull(kind, Sources[col]));
}