/// <summary>
/// Splits the string into an array of individual words
/// </summary>
/// <param name="value">The value to split into words</param>
/// <returns>An array of extracted words</returns>
public static string[] SplitIntoWords
(
this string value
)
{
if (String.IsNullOrEmpty(value))
{
return(new string[] { });
}
// Trim and remove double spaces to avoid empty words
value = value.Trim().Replace(" ", " ");
var matchingWords = new List <string>();
var wordBuilder = new StringBuilder();
var currentIndex = 0;
var previousChar = '\0';
foreach (char currentChar in value)
{
if (Char.IsLetterOrDigit(currentChar) || Char.IsSymbol(currentChar))
{
wordBuilder.Append(currentChar);
}
else if (Char.IsWhiteSpace(currentChar) || Char.IsPunctuation(currentChar))
{
if (wordBuilder.Length > 0)
{
var flushWord = true;
// Check ahead of the current position to see if the next character is a digit
// If the current digit is a number, the current punctuation is a full stop
// and the next digit is also a number, then treat it as part of the word
// (e.g. "1.50" would be treated as a whole word instead of "1" and "50")
if (Char.IsNumber(previousChar) && currentChar == '.' && (currentIndex + 1) < value.Length)
{
var nextChar = value[currentIndex + 1];
if (Char.IsNumber(nextChar))
{
wordBuilder.Append(currentChar);
flushWord = false;
}
}
// Flush the word into the matching words collection
if (flushWord)
{
matchingWords.Add(wordBuilder.ToString());
wordBuilder.Clear();
}
}
}
currentIndex++;
previousChar = currentChar;
}
// Add anything not yet flushed to the words list
if (wordBuilder.Length > 0)
{
matchingWords.Add(wordBuilder.ToString());
}
return(matchingWords.ToArray());
}