public Text Parse()
{
var text = new Text();
var sentence = new Sentence();
var word = new Word();
var punctuationMark = new PunctuationMark();
string line = String.Empty;
string[] splitted;
try
{
while ((line += streamReader.ReadLine()) != null)
{
sentence = GetSentence(line);
}
}
catch (Exception ex)
{
throw ex;
}
return(text);
}
public void Separate(char currentSymbol)
{
if (PunctuationInSetnenceSetted == false && PunctuationSentenceFinisherSetted == false)
{
SetWord();
InSentenseSeparator separator = new InSentenseSeparator(currentSymbol);
SentenceBuffer.Add(separator);
}
if (PunctuationInSetnenceSetted == true)
{
PunctuationMark punctuationMark = new PunctuationMark(CharBuffer);
CharBuffer.Clear();
InSentenseSeparator separator = new InSentenseSeparator(currentSymbol);
SentenceBuffer.Add(separator);
}
if (PunctuationSentenceFinisherSetted == true)
{
PunctuationMark punctuationMark = new PunctuationMark(CharBuffer);
CharBuffer.Clear();
PunctuationSentenceFinisherSetted = false;
Sentence sentence = new Sentence(SentenceBuffer);
ResultTextModel.Add(sentence);
SentenceBuffer.Clear();
SentenseSeparator separator = new SentenseSeparator(currentSymbol);
ResultTextModel.Add(sentence);
}
}
public ITextItem Create(List <Symbol> symbols, int pageNumber)
{
ITextItem textElement;
if (!Separators.SentenceSeparators.Contains(symbols.Last().Value.ToString()) &&
!Separators.InnerSentenceSeparators.Contains(symbols.Last().Value.ToString()))
{
textElement = new Word(symbols);
((Word)textElement).PageNumber = pageNumber;
}
else
{
if (Separators.InnerSentenceSeparators.Contains(symbols.Last().Value.ToString()) ||
Separators.SentenceSeparators.Contains(symbols.Last().Value.ToString()))
{
textElement = new PunctuationMark(symbols);
}
else
{
throw new Exception("Unknown data");
}
}
return(textElement);
}
public IEnumerable <ISentence> GetSentences(PunctuationMark punctuationMark)
{
IEnumerable <ISentence> retVal = new List <ISentence>();
foreach (Paragraph paragraph in _items)
{
retVal = retVal.Union(paragraph.Items.Where(item => item.GetEndSentence().Equals(punctuationMark)));
}
return(retVal);
}
public Sentence Create(List <ITextItem> elementsOfText)
{
Sentence sent;
PunctuationMark punct = elementsOfText.Last() as PunctuationMark;
if (punct != null && punct.SymbolEndOfSent)
{
sent = new Sentence(elementsOfText);
}
else
{
throw new Exception("Error in a sentece");
}
return(sent);
}
private static void SearchSeparator(string line, out int firstSentenceSeparatorOccurence, out PunctuationMark firstSentenceSeparator)
{
firstSentenceSeparatorOccurence = line.Length;
firstSentenceSeparator = default(PunctuationMark);
for (int i = 0; i < PunctuationMarkContainer.AllMarks.Count; i++)
{
int a = line.IndexOf(PunctuationMarkContainer.AllMarks.ElementAt(i).Value);
if (a >= 0 && firstSentenceSeparatorOccurence > a)
{
firstSentenceSeparatorOccurence = a;
firstSentenceSeparator = PunctuationMarkContainer.AllMarks.ElementAt(i);
}
}
}
public Punctuation(IEnumerable <char> item, PunctuationMark punctuationMark)
{
_items = item.ToArray();
_punctuationMark = punctuationMark;
}
public Punctuation(string item, PunctuationMark punctuationMark)
{
_items = item.Select(x => x).ToArray();
_punctuationMark = punctuationMark;
}
public bool TestCharacter(char c)
{
return(PunctuationMark.TestCharacter(c));
}
