public IEnumerable<IWordsTokenizer> Parse(string text)
{
if (string.IsNullOrEmpty(text))
{
throw new ArgumentException("Value cannot be null or empty.", nameof(text));
}
string[] sentences = splitter.Split(text).ToArray();
string saved = string.Empty;
for (int i = 0; i < sentences.Length; i++)
{
string currentSentence = sentences[i].Trim();
while (currentSentence.Length > 1 &&
currentSentence[currentSentence.Length - 2] == ' ')
{
currentSentence = currentSentence.Remove(currentSentence.Length - 2, 1);
}
if (string.IsNullOrWhiteSpace(currentSentence))
{
continue;
}
if (i < sentences.Length - 1)
{
string nextSentence = sentences[i + 1];
bool found = currentSentence.Count(char.IsLetterOrDigit) <= 2;
if (!found)
{
for (int j = 0; j < nextSentence.Length && j <= 3; j++)
{
if (nextSentence[j] == '.')
{
found = true;
break;
}
}
}
if (found)
{
saved += currentSentence;
continue;
}
}
if (!string.IsNullOrWhiteSpace(saved))
{
currentSentence = saved + " " + currentSentence;
}
IWordsTokenizer wordsTokenizer = TokenizerFactory.Create(currentSentence);
saved = string.Empty;
if (wordsTokenizer != NullWordsTokenizer.Instance)
{
yield return wordsTokenizer;
}
}
}
public void CreateNull()
{
WordsTokenizerFactory tokenizerFactory = new WordsTokenizerFactory(
WordsTokenizerFactory.Grouped,
new SimpleWordItemFactory(Global.PosTagger, Global.Raw),
new CombinedPipeline <string>(),
new CombinedPipeline <WordEx>());
IWordsTokenizer tokenizer = tokenizerFactory.Create(null);
Assert.IsInstanceOf <NullWordsTokenizer>(tokenizer);
}
public void Create()
{
WordsTokenizerFactory tokenizerFactory = new WordsTokenizerFactory(
WordsTokenizerFactory.Grouped,
new SimpleWordItemFactory(Global.PosTagger, Global.Raw),
new CombinedPipeline <string>(),
new CombinedPipeline <WordEx>());
IWordsTokenizer tokenizer = tokenizerFactory.Create("Test words");
string[] words = tokenizer.GetWords().ToArray();
Assert.AreEqual(2, words.Length);
Assert.AreEqual("Test", words[0]);
Assert.AreEqual("words", words[1]);
}
