private void CheckWordAndPunct(WordAndPunct wordAndPunct, string word, string punct, int offset)
{
Assert.AreEqual(word, wordAndPunct.Word, "The word is not correct");
Assert.AreEqual(punct, wordAndPunct.Punct, "The punctuation is not correct");
Assert.AreEqual(offset, wordAndPunct.Offset, "The offset is not correct");
}
//public string PunctuationCharacters { get { return punctuationCharacters; } }
public virtual List <WordAndPunct> WordAndPuncts(string text)
{
char cc;
int punctOffset;
List <WordAndPunct> waps = new List <WordAndPunct>();
for (int i = 0; i < text.Length;)
{
WordAndPunct wap = new WordAndPunct();
// Ignore any initial separator characters
while (i < text.Length && char.IsSeparator(text[i]))
{
i++;
}
wap.Offset = i;
bool isFirstCharacterInWord = true;
while (i < text.Length)
{
cc = text[i];
if (IsSingleCharacterWord(cc))
{
if (isFirstCharacterInWord)
{
// Single Character key is the first character in the key.
// It forms a key all by itself.
i = i + 1;
}
else
{
// Single Character key is NOT the first character in the key.
// It ends the key currently being formed.
// 'i' is not incremented
}
break;
}
else if (IsWordMedialPunctuation(cc))
{
// Word medial punctuation only counts if this is not the first
// character in a key AND the next character is key forming.
//! can we have multiple key medial punctuation?
if (isFirstCharacterInWord)
{
break;
}
if (i + 1 >= text.Length || !IsWordFormingCharacter(text[i + 1]))
{
break;
}
}
else if (char.IsDigit(cc))
{
// allow digits in words
}
else if (!IsWordFormingCharacter(cc))
{
break;
}
i = i + 1;
isFirstCharacterInWord = false;
}
wap.Word = text.Substring(wap.Offset, i - wap.Offset);
punctOffset = i;
while (i < text.Length)
{
cc = text[i];
if (IsWordFormingCharacter(cc) || char.IsDigit(cc))
{
break;
}
i = i + 1;
}
wap.Punct = text.Substring(punctOffset, i - punctOffset);
waps.Add(wap);
}
return(waps);
}
