public string Typoize(OptChar[] input,
Dictionary<Tuple<OptChar, string>, int> typos,
int guar)
{
var output = "";
// for each character in the string, sample from the typo dict
for (int i = 0; i < input.Length; i++)
{
KeyValuePair<Tuple<OptChar, String>, int>[] dist;
// handle case where the input character is an empty char
// and condition on the possible typos for this optchar
if (OptChar.get_IsNone(input[i]))
{
// the input character is the empty char, so condition on empty chars
var dist_1 = typos.Where(kvp => kvp.Key.Item1 == null);
// if the current character is a guaranteed typo, ensure that
// an empty character does not appear in the output
dist = (i == guar ? dist_1.Where(kvp => !kvp.Key.Item2.Equals("")) : dist_1).ToArray();
}
else
{
// condition on the possible typos for this particular OptChar
var dist_1 = typos.Where(kvp => input[i].Equals(kvp.Key.Item1));
// get the string corresponding to the current OptChar
var str_i = OptChar.get_IsNone(input[i]) ? "" : input[i].Value.ToString();
// if the current character is a guaranteed typo, ensure that
// the conditioned OptChar does not appear in the output
dist = (i == guar ? dist_1.Where(kvp => !kvp.Key.Item2.Equals(str_i)) : dist_1).ToArray();
}
var total = dist.Select(kvp => kvp.Value).Sum();
var prs = dist.Select(kvp => (double)kvp.Value / total).ToArray();
if (prs.Length == 0)
{
output += OptCharToString(input[i]);
}
else
{
// sample
var j = MultinomialSample(prs);
// j corresponds to what typo string?
output += dist[j].Key.Item2;
}
}
return output;
}
