/// <summary>
/// constructor that takes relative text position
/// </summary>
/// <param name="start"></param>
/// <param name="end"></param>
public Edge(Node node, int start, int end = -1)
{
if (node == null)
{
throw new ArgumentNullException("node");
}
if (start < 0)
{
throw new ArgumentOutOfRangeException("start", "start cannot be negative");
}
// pretend that "end" can be infinite, and then compare with start
if (start > (uint)end)
{
throw new ArgumentOutOfRangeException("start", "cannot start the string after its end");
}
// infinity is just -1
if (end < 0)
{
end = -1;
}
this.Start = start;
this.End = end;
this.EndNode = node;
}
/// <summary>
/// Construct the tree for a given string of text
/// </summary>
/// <param name="text">text from which the tree is constructed</param>
public SuffixTree(string text)
{
if (string.IsNullOrWhiteSpace(text))
{
throw new ArgumentNullException(text);
}
Text = text;
_activeNode = new Node(_currentNodeNumber);
RootNode = _activeNode;
}
/// <summary>
/// Creates the actual suffix tree
/// </summary>
public void Create()
{
// some of our loop iterations actually constitute one route
// in that case we should not chose to accidentally slip and
// follow the suffix node
bool followSuffixNode = false;
for (int i=0; i < Text.Length;)
{
// make sure the lower bound remains within its boundaries
ValidateAndUpdateMinDistance(i);
var nodeEdge = _activeNode.FindNextRoute(i + _activeLength, followSuffixNode);
//if we have terminated in a non-leaf node we are done
if (i + _activeLength >= Text.Length && nodeEdge == null)
{
break;
}
// we could not find anything, add to the tree
if (nodeEdge == null)
{
_activeNode.AddNode(++_currentNodeNumber, i + _activeLength);
_lastBranchIndex = i + _activeLength;
i++;
followSuffixNode = true;
continue;
}
var node = nodeEdge.Item1;
var edge = nodeEdge.Item2;
if (edge == null)
{
//we found a suffix node
_activeNode = node;
_activeLength--;
followSuffixNode = false;
continue;
}
else if(node != null)
{
//we found a new active
_activeNode = node;
_activeLength++;
followSuffixNode = false;
continue;
}
// now walk the chosen path and see where the current suffix diverges
var edgePosTuple = edge.WalkTheEdge(i, ref _activeLength, ref _minDistance, ref _activeNode);
edge = edgePosTuple.Item1;
int j = edgePosTuple.Item2;
if (j == edge.Route.Length)
{
_activeNode = edge.EndNode;
_activeLength += edge.Route.Length;
followSuffixNode = false;
continue;
}
// we now need to insert a new branch node
_minDistance = j;
_lastBranchIndex = i + j + _activeLength;
if (_lastBranchIndex >= Text.Length)
{
i++;
followSuffixNode = true;
continue;
}
// we are inserting a new branch node
var newBranchNode = edge.Split(edge.Start + j - 1, ++_currentNodeNumber);
// if we have reached this branch node through a route of just
// one character - the last branch node should be set as the
if (edge.Route.Length == 1)
{
newBranchNode.SuffixPointer = _activeNode;
}
// the second check is because of the root-node suffix pointer special case
// above
if (null != _lastBranchNode && _lastBranchNode.SuffixPointer == null)
{
_lastBranchNode.SuffixPointer = newBranchNode;
}
newBranchNode.AddNode(++_currentNodeNumber, _lastBranchIndex);
_lastBranchNode = newBranchNode;
i++;
followSuffixNode = true;
}
}
/// <summary>
/// Adds a new node to the tree
/// </summary>
/// <param name="label">Node label</param>
/// <param name="start">Start position in the text</param>
/// <param name="end">End position in the text</param>
internal void AddNode(uint label, int start, int end = -1)
{
var newNode = new Node(label);
var newEdge = new Edge(newNode, start, end);
this.Edges.Add(newEdge.Route[0], newEdge);
}
/// <summary>
/// Splits the edge into two new edges.
/// </summary>
/// <param name="end">Index of the end of the old edge</param>
/// <returns></returns>
internal Node Split(int end, uint currentNodeNumber)
{
int nextStart = end + 1;
var oldNode = this.EndNode;
var newEdge = new Edge(oldNode, nextStart, this.End);
Node newNode = new Node(currentNodeNumber);
this.End = end;
this.EndNode = newNode;
newNode.Edges.Add(newEdge.Route[0], newEdge);
return newNode;
}
/// <summary>
/// Keep comparing original text from position i
/// with what is in the edge
/// </summary>
/// <param name="i">Index of comparison start in the original text</param>
/// <param name="skipCharacters"> How many characters are guaranteed equal</param>
/// <returns>(edge, index) - the edje the character in it where the walk ended</returns>
internal Tuple<Edge, int> WalkTheEdge(int i, ref int activeLength, ref int minDistance, ref Node activeNode)
{
string text = SuffixTree.Text;
int skipCharacters = minDistance;
int index = i + activeLength;
// we know we do not need any comparisons on this edge
if (skipCharacters >= this.Route.Length)
{
var edge = this.EndNode.FindEdgeByChar(i + this.Route.Length);
activeLength += this.Route.Length;
minDistance -= this.Route.Length;
activeNode = this.EndNode;
return edge.WalkTheEdge(i, ref activeLength, ref minDistance, ref activeNode);
}
int j = Walk(text, index, skipCharacters);
return new Tuple<Edge, int>(this, j);
}