/// <summary>
/// Adds pattern ending in this node
/// </summary>
/// <param name="ngram">Pattern</param>
public void AddNgram(ngram_t ngram)
{
if (_Ngrams == null)
{
_Ngrams = new HashSet <ngram_t>(ngram_t_IEqualityComparer.Instance);
}
_Ngrams.Add(ngram);
}
public AhoCorasick(ngram_t ngram)
{
_Root = TreeNode.BuildTree(ngram);
}
public static TreeNode BuildTree(ngram_t ngram)
{
var root = new TreeNode();
// Build keyword tree and transition function
{
// add pattern to tree
var node = root;
foreach (var nerOutputType in ngram.NerOutputTypes)
{
var nodeNew = node.GetTransition(nerOutputType);
if (nodeNew == null)
{
nodeNew = new TreeNode(node, nerOutputType);
node.AddTransition(nodeNew);
}
node = nodeNew;
}
node.AddNgram(ngram);
}
// Find failure functions
var nodes = new List <TreeNode>();
// level 1 nodes - fail to root node
var transitions_root_nodes = root.Transitions;
if (transitions_root_nodes != null)
{
nodes.Capacity = transitions_root_nodes.Count;
foreach (var node in transitions_root_nodes)
{
node.Failure = root;
var transitions_nodes = node.Transitions;
if (transitions_nodes != null)
{
foreach (var trans in transitions_nodes)
{
nodes.Add(trans);
}
}
}
}
// other nodes - using BFS
while (nodes.Count != 0)
{
var newNodes = new List <TreeNode>(nodes.Count);
foreach (var node in nodes)
{
var r = node.Parent.Failure;
var nerOutputType = node.NerOutputType;
while ((r != null) && !r.ContainsTransition(nerOutputType))
{
r = r.Failure;
}
if (r == null)
{
node.Failure = root;
}
else
{
node.Failure = r.GetTransition(nerOutputType);
var failure_ngrams = node.Failure.Ngrams;
if (failure_ngrams != null)
{
foreach (var ng in failure_ngrams)
{
node.AddNgram(ng);
}
}
}
// add child nodes to BFS list
var transitions_nodes = node.Transitions;
if (transitions_nodes != null)
{
foreach (var child in transitions_nodes)
{
newNodes.Add(child);
}
}
}
nodes = newNodes;
}
root.Failure = root;
return(root);
}
private void BuildTree(ngram_t ngram)
{
// Build keyword tree and transition function
{
// add pattern to tree
TreeNode node = _Root;
foreach (var nerOutputType in ngram.NerOutputTypes)
{
TreeNode nodeNew = null;
foreach (TreeNode trans in node.Transitions)
{
if (trans.NerOutputType == nerOutputType)
{
nodeNew = trans;
break;
}
}
if (nodeNew == null)
{
nodeNew = new TreeNode(node, nerOutputType);
node.AddTransition(nodeNew);
}
node = nodeNew;
}
node.AddNgram(ngram);
}
// Find failure functions
var nodes = new List <TreeNode>();
// level 1 nodes - fail to root node
foreach (TreeNode node in _Root.Transitions)
{
node.Failure = _Root;
foreach (TreeNode trans in node.Transitions)
{
nodes.Add(trans);
}
}
// other nodes - using BFS
while (nodes.Count != 0)
{
var newNodes = new List <TreeNode>();
foreach (TreeNode node in nodes)
{
TreeNode r = node.Parent.Failure;
var nerOutputType = node.NerOutputType;
while (r != null && !r.ContainsTransition(nerOutputType))
{
r = r.Failure;
}
if (r == null)
{
node.Failure = _Root;
}
else
{
node.Failure = r.GetTransition(nerOutputType);
foreach (ngram_t result in node.Failure.Ngrams)
{
node.AddNgram(result);
}
}
// add child nodes to BFS list
foreach (TreeNode child in node.Transitions)
{
newNodes.Add(child);
}
}
nodes = newNodes;
}
_Root.Failure = _Root;
}
public AhoCorasick(ngram_t ngram)
{
_Root = new TreeNode();
Count = 1;
BuildTree(ngram);
}
/// <summary>
/// Adds pattern ending in this node
/// </summary>
/// <param name="ngram">Pattern</param>
public void AddNgram(ngram_t ngram)
{
Ngrams.Add(ngram);
}