/// <summary>
/// Add a particular word to node
/// </summary>
/// <param name="node">Node to which the new nodes to apped as child</param>
/// <param name="word">Word to be converted in the form of trie nodes</param>
private void AddWord(TrieNode node, string word, Dictionary <int, int> dicDocuFrequcy)
{
for (int _index = 0, len = word.Length; _index < len; _index++)
{
string _characterToAdd = word[_index].ToString();
TrieNode _childNode = node.GetChild(_characterToAdd);
if (_childNode == null)
{
_wordIndex++;
_childNode = TrieFactory.CreateTrieNode(_characterToAdd);
_childNode.WordIndex = _wordIndex;
_childNode.DocReferences = dicDocuFrequcy;
node.SetChild(_childNode);
}
node = _childNode;
}
node.WordCount++;
}
/// <summary>
/// Creates the trie from flattend body contents. Stores the frequency of the word in the particular document by it's id
/// </summary>
/// <param name="webResponses"></param>
public void CreateTrie(Dictionary <int, string> webResponses)
{
try
{
if (_trie == null)
{
_trie = TrieFactory.CreateTrie();
}
foreach (var _webString in webResponses)
{
List <string> _lstSorted = _webString.Value.Split(' ', StringSplitOptions.RemoveEmptyEntries).ToList();
_lstSorted.Sort();
var numberOfTestcasesWithDuplicates = (from word in _lstSorted
where !Constants._lstExclusion.Contains(word.ToLower())
select word.ToLower())
.GroupBy(x => x)
.ToDictionary(x => x.First(), x => x.Count());
foreach (var item in numberOfTestcasesWithDuplicates)
{
var _trieNode = _trie.ContainsWord(item.Key.Trim());
if (_trieNode != null)
{
_trieNode.DocReferences[_webString.Key] = item.Value;
}
else
{
_trie.AddWord(item.Key.Trim(), new Dictionary <int, int>()
{
{ _webString.Key, item.Value }
});
}
}
}
}
catch (System.Exception ex)
{
Console.WriteLine($"An exception occurred while creating trie {ex}");
}
}
