public static string Normalizer(string input)
{
if (input == "" || input == null)
{
return("");
}
input = Utilizer.RemoveTroublesomeCharacters(input);
// split string based on size
// 5000 character is a safe size for hazm websevice
List <string> portions;
if (input.Length > safePrtionSize)
{
portions = Utilizer.SplitUsingUpssalaSentSegmenter(input, safePrtionSize);
}
else
{
portions = new List <string>()
{
input
};
}
for (int i = 0; i < portions.Count; i++)
{
// normalize input text
portions[i] = HazmWebService().Normalizer(portions[i]);
}
string output = portions.Aggregate((x, y) => x + y);
return(output);
}
public static string[] SentenceTokenizer(string input)
{
if (input == "" || input == null)
{
string[] a = new string[0];
return(a);
}
input = Utilizer.RemoveTroublesomeCharacters(input);
// split string based on size
// 5000 character is a safe size for hazm websevice
List <string> portions;
if (input.Length > safePrtionSize)
{
portions = Utilizer.SplitUsingUpssalaSentSegmenter(input, safePrtionSize);
}
else
{
portions = new List <string>()
{
input
};
}
List <string> finalTokens = new List <string>();
for (int i = 0; i < portions.Count; i++)
{
// normalize input text
finalTokens.AddRange(HazmWebService().SentenceTokenizer(portions[i]));
}
// trim all tokens
finalTokens = finalTokens.Select(s => s.Trim()).ToList();
// remove empty tokens.
finalTokens = finalTokens.Where(s => !string.IsNullOrWhiteSpace(s)).ToList();
return(finalTokens.ToArray());
}
