public static void MergePostings(this VectorNode target, VectorNode source)
{
if (source.PostingsOffsets != null)
{
((List <long>)target.PostingsOffsets).AddRange(source.PostingsOffsets);
}
}
public static void AddDocId(VectorNode target, VectorNode node)
{
foreach (var docId in node.DocIds)
{
target.DocIds.Add(docId);
}
}
public static void Serialize(this VectorNode node, Stream stream)
{
long terminator = 1;
if (node.Left == null && node.Right == null) // there are no children
{
terminator = 3;
}
else if (node.Left == null) // there is a right but no left
{
terminator = 2;
}
else if (node.Right == null) // there is a left but no right
{
terminator = 1;
}
else // there is a left and a right
{
terminator = 0;
}
Span <long> span = stackalloc long[5];
span[0] = node.VectorOffset;
span[1] = node.PostingsOffset;
span[2] = node.Vector.ComponentCount;
span[3] = node.Weight;
span[4] = terminator;
stream.Write(MemoryMarshal.Cast <long, byte>(span));
}
public static void MergeDocIds(this VectorNode target, VectorNode source)
{
if (source.DocIds != null)
{
target.DocIds.AddRange(source.DocIds);
}
}
public static void SetIdsOnAllNodes(VectorNode root)
{
var node = root.ComponentCount == 0 ? root.Right : root;
var stack = new Stack <VectorNode>();
long id = 0;
while (node != null)
{
node.PostingsOffset = id++;
if (node.Right != null)
{
stack.Push(node.Right);
}
node = node.Left;
if (node == null)
{
if (stack.Count > 0)
{
node = stack.Pop();
}
}
}
}
public static IEnumerable <VectorNode> All(VectorNode root)
{
var node = root.ComponentCount == 0 ? root.Right : root;
var stack = new Stack <VectorNode>();
while (node != null)
{
yield return(node);
if (node.Right != null)
{
stack.Push(node.Right);
}
node = node.Left;
if (node == null)
{
if (stack.Count > 0)
{
node = stack.Pop();
}
}
}
}
public static string Visualize(VectorNode root)
{
StringBuilder output = new StringBuilder();
Visualize(root, output, 0);
return(output.ToString());
}
public static Hit ClosestMatch(VectorNode root, IVector vector, IModel model)
{
var best = root;
var cursor = root;
double highscore = 0;
while (cursor != null)
{
var angle = cursor.Vector == null ? 0 : model.CosAngle(vector, cursor.Vector);
if (angle > model.FoldAngle)
{
if (angle > highscore)
{
highscore = angle;
best = cursor;
}
cursor = cursor.Left;
}
else
{
if (angle > highscore)
{
highscore = angle;
best = cursor;
}
cursor = cursor.Right;
}
}
return(new Hit(best, highscore));
}
public static float[][] AsOneHotMatrix(VectorNode root)
{
var node = root.Vector == null ? root.Right : root;
var stack = new Stack <VectorNode>();
var matrix = new float[root.Weight][];
var index = 0;
while (node != null)
{
var vector = new float[root.Weight];
vector[index] = 1;
matrix[index] = vector;
index++;
if (node.Right != null)
{
stack.Push(node.Right);
}
node = node.Left;
if (node == null)
{
if (stack.Count > 0)
{
node = stack.Pop();
}
}
}
return(matrix);
}
public (int depth, int width) CreatePage(VectorNode column, Stream vectorStream, Stream postingsStream, PageIndexWriter pageIndexWriter)
{
var page = column.SerializeTree(_ixStream, vectorStream, postingsStream);
pageIndexWriter.Put(page.offset, page.length);
return(PathFinder.Size(column));
}
public (int depth, int width) CreatePage(VectorNode column, Stream vectorStream, PageIndexWriter pageIndexWriter)
{
var page = GraphBuilder.SerializeTree(column, _ixStream, vectorStream, null);
pageIndexWriter.Put(page.offset, page.length);
return(PathFinder.Size(column));
}
public static long AppendSynchronized(
VectorNode root,
VectorNode node,
IDistanceCalculator model,
double foldAngle,
double identicalAngle)
{
var cursor = root;
while (true)
{
var angle = cursor.Vector == null ? 0 : model.CosAngle(node.Vector, cursor.Vector);
if (angle >= identicalAngle)
{
return(cursor.PostingsOffset);
}
else if (angle > foldAngle)
{
if (cursor.Left == null)
{
lock (cursor)
{
if (cursor.Left == null)
{
node.PostingsOffset = root.Weight;
cursor.Left = node;
return(node.PostingsOffset);
}
}
}
else
{
cursor = cursor.Left;
}
}
else
{
if (cursor.Right == null)
{
lock (cursor)
{
if (cursor.Right == null)
{
node.PostingsOffset = root.Weight;
cursor.Right = node;
return(node.PostingsOffset);
}
}
}
else
{
cursor = cursor.Right;
}
}
}
}
public static void MergeDocIdsConcurrent(this VectorNode target, VectorNode source)
{
lock (target.Sync)
{
if (source.DocIds != null)
{
target.DocIds.AddRange(source.DocIds);
}
}
}
public static IEnumerable <VectorNode> LeftList(VectorNode root)
{
var node = root.Left;
while (node != null)
{
yield return(node);
node = node.Left;
}
}
public static void AddRight(VectorNode parent, VectorNode node)
{
var target = parent;
while (target.Right != null)
{
target = target.Right;
}
node.Right = target.Right;
target.Right = node;
}
public static int Depth(VectorNode node)
{
var count = 0;
node = node.Left;
while (node != null)
{
count++;
node = node.Left;
}
return(count);
}
public static VectorNode Train <T>(IModel <T> model, params T[] data)
{
var root = new VectorNode();
var unclassified = new Queue <VectorNode>();
foreach (var item in data)
{
foreach (var vector in model.Tokenize(item))
{
VectorNode node;
if (!TryMergeOrAddSupervised(root, new VectorNode(vector), model, out node))
{
unclassified.Enqueue(node);
}
}
}
var batchSize = unclassified.Count;
var numOfIterations = 0;
var lastCount = 0;
while (unclassified.Count > 0)
{
VectorNode node;
if (!TryMergeOrAddSupervised(root, unclassified.Dequeue(), model, out node))
{
unclassified.Enqueue(node);
}
if (++numOfIterations % batchSize == 0)
{
if (lastCount == unclassified.Count)
{
break;
}
else
{
lastCount = unclassified.Count;
}
}
}
foreach (var node in unclassified)
{
MergeOrAdd(root, node, model);
}
return(root);
}
private static void Visualize(VectorNode node, StringBuilder output, int depth)
{
if (node == null)
{
return;
}
output.Append('\t', depth);
output.AppendFormat($"{node} w:{node.Weight}");
output.AppendLine();
Visualize(node.Left, output, depth + 1);
Visualize(node.Right, output, depth);
}
public static VectorNode CreateTree <T>(IModel <T> model, params T[] data)
{
var root = new VectorNode();
foreach (var item in data)
{
foreach (var vector in model.Tokenize(item))
{
MergeOrAdd(root, new VectorNode(vector), model);
}
}
return(root);
}
public static VectorNode DeserializeNode(
long vecOffset,
long postingsOffset,
long componentCount,
long weight,
long terminator,
Stream vectorStream,
IVectorSpaceConfig model)
{
var vector = VectorOperations.DeserializeVector(vecOffset, (int)componentCount, model.VectorWidth, vectorStream);
var node = new VectorNode(postingsOffset, vecOffset, terminator, weight, vector);
return(node);
}
public static VectorNode DeserializeNode(
long vecOffset,
long postingsOffset,
long componentCount,
long weight,
long terminator,
Stream vectorStream,
IDistanceCalculator model)
{
var vector = VectorOperations.DeserializeVector(vecOffset, (int)componentCount, model.NumOfDimensions, vectorStream);
var node = new VectorNode(postingsOffset, vecOffset, terminator, weight, vector);
return(node);
}
public static VectorNode CreateTree <T>(this IModel <T> model, IIndexingStrategy indexingStrategy, params T[] data)
{
var root = new VectorNode();
foreach (var item in data)
{
foreach (var vector in model.Tokenize(item))
{
indexingStrategy.ExecutePut <T>(root, new VectorNode(vector));
}
}
return(root);
}
public static void MergeOrAdd(
VectorNode root,
VectorNode node,
IModel model,
out VectorNode parent)
{
var cursor = root;
while (true)
{
var angle = cursor.Vector == null ? 0 : model.CosAngle(node.Vector, cursor.Vector);
if (angle >= model.IdenticalAngle)
{
parent = cursor;
break;
}
else if (angle > model.FoldAngle)
{
if (cursor.Left == null)
{
cursor.Left = node;
parent = cursor;
break;
}
else
{
cursor = cursor.Left;
}
}
else
{
if (cursor.Right == null)
{
cursor.Right = node;
parent = cursor;
break;
}
else
{
cursor = cursor.Right;
}
}
}
}
public static bool TryMerge(
VectorNode root,
VectorNode node,
IDistanceCalculator model,
double foldAngle,
double identicalAngle,
out VectorNode parent)
{
var cursor = root;
while (true)
{
var angle = cursor.Vector == null ? 0 : model.CosAngle(node.Vector, cursor.Vector);
if (angle >= identicalAngle)
{
parent = cursor;
return(true);
}
else if (angle > foldAngle)
{
if (cursor.Left == null)
{
cursor.Left = node;
parent = cursor;
return(false);
}
else
{
cursor = cursor.Left;
}
}
else
{
if (cursor.Right == null)
{
cursor.Right = node;
parent = cursor;
return(false);
}
else
{
cursor = cursor.Right;
}
}
}
}
public static void MergeOrAddSupervised(
this VectorNode root,
VectorNode node,
IModel model)
{
var cursor = root;
while (true)
{
var angle = cursor.Vector == null ? 0 : model.CosAngle(node.Vector, cursor.Vector);
if (angle >= model.IdenticalAngle)
{
if (!cursor.Vector.Label.Equals(node.Vector.Label))
{
throw new InvalidOperationException($"IdenticalAngle {model.IdenticalAngle} is too low. Angle was {angle}");
}
MergeDocIds(cursor, node);
break;
}
else if (angle > model.FoldAngle)
{
if (cursor.Left == null)
{
cursor.Left = node;
break;
}
else
{
cursor = cursor.Left;
}
}
else
{
if (cursor.Right == null)
{
cursor.Right = node;
break;
}
else
{
cursor = cursor.Right;
}
}
}
}
private static void Visualize(VectorNode node, int depth, StringBuilder output)
{
if (node == null)
{
return;
}
output.Append('\t', depth);
output.AppendFormat($"{node} w:{node.Weight} ");
if (node.IsRoot)
{
output.AppendFormat($"{Size(node)}");
}
output.AppendLine();
}
public static bool TryAdd(
this VectorNode root,
VectorNode node,
IModel model)
{
var cursor = root;
while (true)
{
var angle = cursor.Vector == null ? 0 : model.CosAngle(node.Vector, cursor.Vector);
if (angle >= model.IdenticalAngle)
{
return(false);
}
else if (angle > model.FoldAngle)
{
if (cursor.Left == null)
{
cursor.Left = node;
return(true);
}
else
{
cursor = cursor.Left;
}
}
else
{
if (cursor.Right == null)
{
cursor.Right = node;
return(true);
}
else
{
cursor = cursor.Right;
}
}
}
}
/// <summary>
/// Persist tree to disk.
/// </summary>
/// <param name="node">Tree to perist.</param>
/// <param name="indexStream">stream to perist tree into</param>
/// <param name="vectorStream">stream to persist vectors in</param>
/// <param name="postingsStream">optional stream to persist any posting references into</param>
/// <returns></returns>
public static (long offset, long length) SerializeTree(
VectorNode node,
Stream indexStream,
Stream vectorStream,
Stream postingsStream = null)
{
var stack = new Stack <VectorNode>();
var offset = indexStream.Position;
var length = 0;
if (node.ComponentCount == 0)
{
node = node.Right;
}
while (node != null)
{
if (node.PostingsOffset == -1 && postingsStream != null)
{
SerializePostings(node, postingsStream);
}
node.VectorOffset = VectorOperations.SerializeVector(node.Vector, vectorStream);
SerializeNode(node, indexStream);
length += VectorNode.BlockSize;
if (node.Right != null)
{
stack.Push(node.Right);
}
node = node.Left;
if (node == null && stack.Count > 0)
{
node = stack.Pop();
}
}
return(offset, length);
}
public static bool TryMergeOrAddSupervised(
VectorNode root,
VectorNode node,
IModel model,
out VectorNode unclassified)
{
var cursor = root;
while (true)
{
var angle = cursor.Vector == null ? 0 : model.CosAngle(node.Vector, cursor.Vector);
if (angle > model.FoldAngle)
{
if (node.Vector.Label.Equals(cursor.Vector.Label))
{
AddDocId(cursor, node);
cursor.Vector.Average(node.Vector);
unclassified = null;
return(true);
}
else
{
unclassified = node;
return(false);
}
}
else
{
if (cursor.Right == null)
{
cursor.Right = node;
unclassified = null;
return(true);
}
else
{
cursor = cursor.Right;
}
}
}
}
public static VectorNode DeserializeTree(Stream indexStream, Stream vectorStream, long indexLength, IModel model)
{
VectorNode root = new VectorNode();
VectorNode cursor = root;
var tail = new Stack <VectorNode>();
int read = 0;
var buf = new byte[VectorNode.BlockSize];
while (read < indexLength)
{
indexStream.Read(buf);
var node = DeserializeNode(buf, vectorStream, model);
if (node.Terminator == 0) // there is both a left and a right child
{
cursor.Left = node;
tail.Push(cursor);
}
else if (node.Terminator == 1) // there is a left but no right child
{
cursor.Left = node;
}
else if (node.Terminator == 2) // there is a right but no left child
{
cursor.Right = node;
}
else // there are no children
{
if (tail.Count > 0)
{
tail.Pop().Right = node;
}
}
cursor = node;
read += VectorNode.BlockSize;
}
return(root);
}
