public TextProcessor(string text)
{
Input = text;
Input = HttpUtility.HtmlDecode(Input);
CurrentText = Input;
CurrentText = CurrentText.StripHtml().ToStandard();
var _models = CurrentText.ToWords();
_models = _models.Where(t => t.IsWord()).ToArray();
for (int i = 0; i < _models.Length; i++)
{
var _word = new Word(_models[i]);
if (_word.CanRead())
{
WordObject _object = new WordObject(_models[i], i);
_object.Indexs = CurrentText.IndexAll(_word.Value);
FullWordObjects.Add(_object);
_object.ToClean();
if (_object.IsClean())
{
CleanWordObjects.Add(_object);
}
}
}
CountOfWords = CleanWordObjects.Count;
CountOfFullWords = FullWordObjects.Count;
TimeToRead = CountOfWords / 200; //Normal read speed
LoadFull();
}
public string[] GetVnKeywords()
{
var _keywords = CleanWordObjects
.Where(t => t.IsVnKeyword())
.Select(t => t.Value)
.Distinct()
.ToArray();
return(_keywords);
}
public void LoadFull()
{
Words = FullWordObjects.Select(t => t.Value).ToList();
EnglihsWords = CleanWordObjects.Where(t => t.Lang == WordLang.English)
.Select(t => t.Value.ToLower())
.Distinct()
.ToList();
VietnameseWords = CleanWordObjects.Where(t => t.Lang == WordLang.Vietnamese)
.Select(t => t.Value.ToLower())
.Distinct()
.ToList();
Numbers = CleanWordObjects.Where(t => t.IsNumber())
.Select(t => t.Value.ToLower())
.Distinct()
.ToList();
UrlLinks = CleanWordObjects.Where(t => t.Value.IsUrl())
.Select(t => t.Value.ToLower())
.Distinct()
.ToList();
Emails = CleanWordObjects.Where(t => t.Value.IsEmail())
.Select(t => t.Value.ToLower())
.Distinct()
.ToList();
Sentences = CurrentText.GetSentences()
.Distinct()
.ToList();
QuestionSentences = GetQuestionSentences();
CountOfQuestions = QuestionSentences.Count();
try
{
Phrases = GetPhrases().ToList();
MathExprs = GetMathExprs().ToList();
Commands = GetCommands();
MakeOutput();
}
catch
{
return;
}
}
public IEnumerable <WordObject> GetWordsBy(WordLang lang)
{
return(CleanWordObjects.Where(t => t.Lang == lang));
}
