public NgramBufferBuilder(int ngramLength, int skipLength, int slotLim, NgramIdFinder finder)
{
Contracts.Assert(ngramLength > 0);
Contracts.Assert(skipLength >= 0);
Contracts.Assert(ngramLength <= MaxSkipNgramLength - skipLength);
Contracts.Assert(slotLim >= 0);
_ngramLength = ngramLength;
_skipLength = skipLength;
_slotLim = slotLim;
_ngram = new uint[_ngramLength];
_queue = new FixedSizeQueue <uint>(_ngramLength + _skipLength);
_bldr = BufferBuilder <Float> .CreateDefault();
_finder = finder;
}
/// <summary>
/// This is for the bagging case - vector input and outputs should be added.
/// </summary>
private ValueGetter <VBuffer <Float> > MakeGetterBag(IRow input, int iinfo)
{
Host.AssertValue(input);
Host.Assert(Infos[iinfo].TypeSrc.IsVector);
Host.Assert(Infos[iinfo].TypeSrc.ItemType.IsKey);
Host.Assert(_bag[iinfo]);
Host.Assert(Infos[iinfo].TypeSrc.ItemType.KeyCount == _types[iinfo].VectorSize);
var info = Infos[iinfo];
int size = info.TypeSrc.ItemType.KeyCount;
Host.Assert(size > 0);
int cv = info.TypeSrc.VectorSize;
Host.Assert(cv >= 0);
var getSrc = RowCursorUtils.GetVecGetterAs <uint>(NumberType.U4, input, info.Source);
var src = default(VBuffer <uint>);
var bldr = BufferBuilder <float> .CreateDefault();
return
((ref VBuffer <Float> dst) =>
{
bldr.Reset(size, false);
getSrc(ref src);
Host.Check(cv == 0 || src.Length == cv);
// The indices are irrelevant in the bagging case.
var values = src.Values;
int count = src.Count;
for (int slot = 0; slot < count; slot++)
{
uint key = values[slot] - 1;
if (key < size)
{
bldr.AddFeature((int)key, 1);
}
}
bldr.GetResult(ref dst);
});
}
public ArrayToSparseVBufferColumn(PrimitiveType itemType, Combiner <T> combiner, T[][] values)
: base(itemType, values, Utils.Size)
{
_bldr = new BufferBuilder <T>(combiner);
}
