public string GetFilteredText(string content)
{
// assuming that words in store are written in lowercase
string[] wordsToFilter = _wordsDataStore.GetWords();
// initial filtering
string cleanString = Regex.Replace(content, @"[^ .:a-zA-Z0-9\-]", "");
// word filtering
var re = new Regex(
@"\b("
+ string.Join("|", wordsToFilter.Select(word =>
string.Join(@"\s*", word.ToCharArray())))
+ @")\b", RegexOptions.IgnoreCase);
return(re.Replace(cleanString, match =>
{
if (match.Length > 2)
{
return match.Value[0] + new string('*', match.Length - 2) + match.Value[match.Length - 1];
}
else
{
return new string('*', match.Length);
}
}));
}
public int CountWordsFromDatastore(string content)
{
// assuming that words in store are written in lowercase
string[] wordsToCount = _wordsDataStore.GetWords();
string[] source = content.Split(new char[] { '.', '?', '!', ' ', ';', ':', ',' }, StringSplitOptions.RemoveEmptyEntries);
var matchQuery = from word in source
where wordsToCount.Contains(word.ToLowerInvariant())
select word;
int wordCount = matchQuery.Count();
return(wordCount);
}