public Dictionary <string, string[]> GetMatchedPrefix(string word, string[] prePronunciation)
{
Dictionary <string, string[]> matchedPrefix = null;
matchedPrefix = _wordStructureRule.GetMatchedPrefix(word, prePronunciation);
if (matchedPrefix == null)
{
matchedPrefix = new Dictionary <string, string[]>();
}
matchedPrefix.Add("", prePronunciation);
return(matchedPrefix);
}
public string[] ToThaiPhoneme(string[] thaiPronunciations)
{
if (!this.IsInitialized)
{
this.InitializePronunciationRules();
}
string[] thaiPhonemes = new string[thaiPronunciations.Length];
for (int count = 0; count < thaiPronunciations.Length; count++)
{
string s = thaiPronunciations[count];
Dictionary <string, string[]> matchedPrefix = _thaiPronunciationStructureRule.GetMatchedPrefix(s, new string[] { "" });
thaiPhonemes[count] = matchedPrefix.ContainsKey(s) ? matchedPrefix[s][0] : "$-$-$-$";
}
return(thaiPhonemes);
}
//private bool ToThaiPronunciation(string word,
// List<KeyValuePair<string, string[]>> currentSegmentedWordList, List<string> currentPronunciationList, int currentScore,
// List<KeyValuePair<string, string[]>> resultSegmentedWordList, List<string> resultPronunciationList, ref int resultLeftWordLength, ref double resultScore,
// ref int numSolutionFound)
//{
// if (!_cachedThaiDictMatchedPrefix.ContainsKey(word))
// _cachedThaiDictMatchedPrefix.Add(word, WordStructureThaiDictRule.Instance.GetMatchedPrefix(word, new string[] { "" }));
// if (!_cachedThaiRuleMatchedPrefix.ContainsKey(word))
// _cachedThaiRuleMatchedPrefix.Add(word, _thaiWordStructureRule.GetMatchedPrefix(word, new string[] { "" }));
// Dictionary<string, string[]> thaiDictMatchedPrefix = _cachedThaiDictMatchedPrefix[word];
// Dictionary<string, string[]> thaiRuleMatchedPrefix = _cachedThaiRuleMatchedPrefix[word];
// Dictionary<string, string[]> matchedPrefix = new Dictionary<string, string[]>(thaiDictMatchedPrefix);
// foreach (KeyValuePair<string, string[]> keyValuePair in thaiRuleMatchedPrefix)
// if (!matchedPrefix.ContainsKey(keyValuePair.Key))
// matchedPrefix.Add(keyValuePair.Key, keyValuePair.Value);
// if (matchedPrefix != null && matchedPrefix.Count > 0)
// {
// string[] keys = new string[matchedPrefix.Count];
// matchedPrefix.Keys.CopyTo(keys, 0);
// Array.Sort<string>(keys, new Comparison<string>(delegate(string x, string y)
// {
// int firstCompareResult = (thaiDictMatchedPrefix.ContainsKey(y) ? 1 : 0) - (thaiDictMatchedPrefix.ContainsKey(x) ? 1 : 0);
// return firstCompareResult != 0 ? firstCompareResult : (y.Length - x.Length);
// }));
// foreach (string prefix in keys)
// {
// int nextScore = currentScore;
// if (thaiDictMatchedPrefix.ContainsKey(prefix))
// nextScore += prefix.Length;
// if (nextScore + word.Length - prefix.Length < resultScore)
// continue;
// string[] prefixPronunciations = matchedPrefix[prefix];
// currentSegmentedWordList.Add(new KeyValuePair<string, string[]>(prefix, prefixPronunciations));
// currentPronunciationList.AddRange(prefixPronunciations);
// if (this.ToThaiPronunciation(word.Substring(prefix.Length),
// currentSegmentedWordList, currentPronunciationList, nextScore,
// resultSegmentedWordList, resultPronunciationList, ref resultLeftWordLength,
// ref resultScore, ref numSolutionFound))
// return true;
// currentSegmentedWordList.RemoveAt(currentSegmentedWordList.Count - 1);
// currentPronunciationList.RemoveRange(currentPronunciationList.Count - prefixPronunciations.Length, prefixPronunciations.Length);
// }
// } else
// {
// if (word.Length < resultLeftWordLength
// || word.Length == resultLeftWordLength
// && (currentScore > resultScore
// || currentScore == resultScore && currentSegmentedWordList.Count < resultSegmentedWordList.Count))
// {
// resultSegmentedWordList.Clear();
// resultSegmentedWordList.AddRange(currentSegmentedWordList);
// resultPronunciationList.Clear();
// resultPronunciationList.AddRange(currentPronunciationList);
// resultLeftWordLength = word.Length;
// resultScore = currentScore;
// }
// if (++numSolutionFound >= this.MaxSearchingResult)
// return true;
// }
// return false;
//}
private void SegmentThaiSentence(string word,
out List <KeyValuePair <string, string[]> > resultSegmentedWordList, out int resultScore, out int segmentedWordsLength)
{
if (_cachedSegmentedWordList.ContainsKey(word))
{
resultSegmentedWordList = _cachedSegmentedWordList[word];
resultScore = _cachedResultScore[word];
segmentedWordsLength = _cachedSegmentedWordsLength[word];
return;
}
resultSegmentedWordList = new List <KeyValuePair <string, string[]> >();
resultScore = 0;
segmentedWordsLength = 0;
Dictionary <string, string[]> thaiDictMatchedPrefix = WordStructureThaiDictRule.Instance.GetMatchedPrefix(word, new string[] { "" });
Dictionary <string, string[]> thaiRuleMatchedPrefix = _thaiWordStructureRule.GetMatchedPrefix(word, new string[] { "" });
Dictionary <string, string[]> matchedPrefix = new Dictionary <string, string[]>(thaiDictMatchedPrefix);
foreach (KeyValuePair <string, string[]> keyValuePair in thaiRuleMatchedPrefix)
{
if (!matchedPrefix.ContainsKey(keyValuePair.Key))
{
matchedPrefix.Add(keyValuePair.Key, keyValuePair.Value);
}
}
if (matchedPrefix != null && matchedPrefix.Count > 0)
{
string[] keys = new string[matchedPrefix.Count];
matchedPrefix.Keys.CopyTo(keys, 0);
Array.Sort <string>(keys, new Comparison <string>(delegate(string x, string y)
{
int firstCompareResult = (thaiDictMatchedPrefix.ContainsKey(y) ? 1 : 0) - (thaiDictMatchedPrefix.ContainsKey(x) ? 1 : 0);
return(firstCompareResult != 0 ? firstCompareResult : (y.Length - x.Length));
}));
foreach (string prefix in keys)
{
List <KeyValuePair <string, string[]> > subResultSegmentedWordList;
int subResultScore;
int subResultSegmentedWordLength;
this.SegmentThaiSentence(word.Substring(prefix.Length), out subResultSegmentedWordList, out subResultScore, out subResultSegmentedWordLength);
int currentScore = (thaiDictMatchedPrefix.ContainsKey(prefix) ? prefix.Length : 0) + subResultScore;
int currentSegmentedWordsLength = prefix.Length + subResultSegmentedWordLength;
if (currentSegmentedWordsLength > segmentedWordsLength ||
currentSegmentedWordsLength == segmentedWordsLength &&
(currentScore > resultScore ||
currentScore == resultScore &&
subResultSegmentedWordList.Count < resultSegmentedWordList.Count))
{
string[] prefixPronunciations = matchedPrefix[prefix];
resultSegmentedWordList = new List <KeyValuePair <string, string[]> >();
resultSegmentedWordList.Add(new KeyValuePair <string, string[]>(prefix, prefixPronunciations));
resultSegmentedWordList.AddRange(subResultSegmentedWordList);
resultScore = currentScore;
segmentedWordsLength = currentSegmentedWordsLength;
}
}
}
_cachedSegmentedWordList.Add(word, resultSegmentedWordList);
_cachedResultScore.Add(word, resultScore);
_cachedSegmentedWordsLength.Add(word, segmentedWordsLength);
}