///<summary>
/// Complete the rest of incomplete word
///</summary>
/// <param name="subWord">Incomplete word</param>
/// <returns>Completed words start with incomplete word</returns>
public string[] Complete(string subWord)
{
NodeWithFreqandPOS node = IndexOf(subWord);
if (node == null)
{
return(new string[0]);
}
string[] restOfWord = TraverseFrom(node);
List <string> completeWords = new List <string>();
//StringBuilder sb = new StringBuilder(subWord.Substring(0, subWord.Length - 1));
//int start = subWord.Length - 1;
string origstr = subWord.Substring(0, subWord.Length - 1);
foreach (string str in restOfWord)
{
//sb.Append(str);
//completeWords.Add(sb.ToString());
completeWords.Add(origstr + str);
//sb.Remove(start, str.Length);
}
return(completeWords.ToArray());
}
///<summary>
/// Remove a child node
///</summary>
///<param name="node">Child node</param>
public void RemoveLink(NodeWithFreqandPOS node)
{
if (this.m_links.ContainsKey(node.m_value))
{
this.m_links.Remove(node.m_value);
}
}
private void RemoveFromMemory(string word)
{
NodeWithFreqandPOS leaf = IndexOf(word);
if (leaf != null)
{
leaf.LogicalRemove();
}
}
private static void Clear(NodeWithFreqandPOS node)
{
foreach (NodeWithFreqandPOS curNode in node.GetLinks())
{
Clear(curNode);
}
node.Clear();
}
/// <summary>
/// Check if a word exists
/// </summary>
/// <param name="word">Word</param>
/// <returns>If the dictionary contains the word, returns true, else returns false.</returns>
public bool Contain(string word)
{
NodeWithFreqandPOS leaf = IndexOf(word);
if (leaf == null || !leaf.IsEndOfWord)
{
return(false);
}
return(true);
}
/// <summary>
/// Word frequency
/// </summary>
/// <param name="word">word</param>
/// <returns>usage frequency</returns>
public int WordFrequency(string word)
{
NodeWithFreqandPOS leaf = IndexOf(word);
if (leaf == null || !leaf.IsEndOfWord)
{
return(0);
}
return(leaf.WordFrequency);
}
protected string[] TraverseFrom(NodeWithFreqandPOS node)
{
string[] words = DoDFSTraverse(node);
List <string> strList = new List <string>();
foreach (string str in words)
{
strList.Add(ReverseString(str));
}
return(strList.ToArray());
}
/// <summary>
/// Word pos
/// </summary>
/// <param name="word">word</param>
/// <returns>POS tag</returns>
public PersianPOSTag WordPOS(string word)
{
NodeWithFreqandPOS leaf = IndexOf(word);
if (leaf == null || !leaf.IsEndOfWord)
{
return(0);
}
return(leaf.POSTag.ToEnum <PersianPOSTag>());
}
/// <summary>
/// Check if a word exists
/// </summary>
/// <param name="word">Word</param>
/// <param name="freq">Word's usage frequency</param>
/// <returns>If the dictionary contains the word, returns true, else returns false.</returns>
public bool Contain(string word, out int freq)
{
freq = 0;
NodeWithFreqandPOS leaf = IndexOf(word);
if (leaf == null || !leaf.IsEndOfWord)
{
return(false);
}
freq = leaf.WordFrequency;
return(true);
}
/// <summary>
/// Check if a word exists
/// </summary>
/// <param name="word">Word</param>
/// <param name="posTag">Word's POS tag</param>
/// <returns>If the dictionary contains the word, returns true, else returns false.</returns>
public bool Contain(string word, out PersianPOSTag posTag)
{
posTag = PersianPOSTag.UserPOS;
NodeWithFreqandPOS leaf = IndexOf(word);
if (leaf == null || !leaf.IsEndOfWord)
{
return(false);
}
posTag = leaf.POSTag.ToEnum <PersianPOSTag>();
return(true);
}
/// <summary>
/// Teaverse Tree on Breath First Type
/// </summary>
/// <returns></returns>
private string[] DoBFSTraverse(NodeWithFreqandPOS node)
{
if (node.HaveLinks())
{
List <string> suggestions = new List <string>();
StringBuilder sb = new StringBuilder();
foreach (NodeWithFreqandPOS curNode in node.GetLinks())
{
sb.Append(DoDFSTraverse(curNode));
suggestions.Add(sb.ToString());
sb.Remove(0, sb.Length);
}
return(suggestions.ToArray());
}
return(new string[] { node.Value.ToString() });
}
/// <summary>
/// Teaverse Tree on Depth First Type
/// </summary>
/// <returns></returns>
private string[] DoDFSTraverse(NodeWithFreqandPOS node)
{
if (node.HaveLinks())
{
List <string> retDFS = new List <string>();
foreach (NodeWithFreqandPOS curNode in node.GetLinks())
{
retDFS.AddRange(DoDFSTraverse(curNode));
}
List <string> suggestions = new List <string>();
if (node.Value != this.m_root.Value)
{
if (node.IsEndOfWord)
{
suggestions.Add(new StringBuilder().Append(node.Value).ToString());
}
StringBuilder sb = new StringBuilder();
foreach (string str in retDFS)
{
sb.Append(str);
sb.Append(node.Value);
suggestions.Add(sb.ToString());
sb.Remove(0, sb.Length);
}
return(suggestions.ToArray());
}
else
{
return(retDFS.ToArray());
}
}
if (node.IsEndOfWord)
{
return(new string[] { new StringBuilder().Append(node.Value).ToString() });
}
else
{
return(new string[0]);
}
}
protected NodeWithFreqandPOS IndexOf(string subWord)
{
NodeWithFreqandPOS curNode = this.m_root;
foreach (char c in subWord)
{
if (curNode != null)
{
curNode = curNode.GetNextNode(c);
}
}
if (curNode == this.m_root)
{
curNode = null;
}
return(curNode);
}
///<summary>
/// Add a child node
///</summary>
///<param name="node">Child node</param>
///<returns>Node pointer to added child node</returns>
public NodeWithFreqandPOS AddLink(NodeWithFreqandPOS node)
{
if (!this.m_links.ContainsKey(node.m_value))
{
this.m_links.Add(node.m_value, node);
}
else if (node.IsEndOfWord)
{
this.m_links[node.m_value].m_isEndLetter = node.IsEndOfWord;
this.m_links[node.m_value].m_freq = this.m_links[node.m_value].m_freq > node.WordFrequency ? this.m_links[node.m_value].m_freq : node.WordFrequency;
if (this.m_links[node.m_value].m_pos != node.m_pos)
{
this.m_links[node.m_value].m_pos = mergePOS(node.m_pos, this.m_links[node.m_value].m_pos);
}
}
//if (node.IsEndOfWord && !this.m_links[node.m_value].m_isEndLetter)
return(this.m_links[node.m_value]);
}
private void AddWordToMemory(string word, int freq, string pos)
{
NodeWithFreqandPOS curNode = this.m_root;
int i = 0, length = word.Length;
foreach (char c in word)
{
if (++i == length)
{
curNode = curNode.AddLink(GenerateNode(c, freq, true, pos));
//add nubmer of dictionary words
this.m_freqSummation += freq;
++this.m_wordsCount;
}
else
{
curNode = curNode.AddLink(GenerateNode(c, 0, false, pos));
}
}
}
///<summary>
/// Complete the rest of incomplete word
///</summary>
/// <param name="subWord">Incomplete word</param>
///<param name="count">Number of returned suggestions</param>
/// <returns>Completed words start with incomplete word</returns>
public string[] Complete(string subWord, int count)
{
NodeWithFreqandPOS node = IndexOf(subWord);
if (node == null)
{
return(new string[0]);
}
string[] restOfWord = TraverseFrom(node);
string origstr = subWord.Substring(0, subWord.Length - 1);
List <string> completeWords = new List <string>();
for (int i = 0; i < Math.Min(count, restOfWord.Length); ++i)
{
completeWords.Add(origstr + restOfWord[i]);
}
return(completeWords.ToArray());
}
///<summary>
/// Class Constructor
///</summary>
public WordFreqPOSContainerTree()
{
m_root = new NodeWithFreqandPOS('*', false, 0, PersianPOSTag.UserPOS.ToString());
}