public List<Token> correct(string word)
{
var token = new Token( word );
token.origTerm = word;
var result = new List<Token>();
Console.WriteLine("correct : " + word );
// if its an number, just leave it alone.
if (!word.IsInt() )
{
var hs = Edits1(new List<Token>() { token });
// add original word
hs.Add(token);
var hs2 = EditAgain(hs, 2);
// cant just add them? tried union, but returns IEnumerable.
foreach( var s in hs2)
{
hs.Add( s ) ;
}
var finalHS = hs;
//Console.WriteLine("dictionary size " + dictionary.Count.ToString());
//Console.WriteLine("result size " + finalHS.Count.ToString() );
var wordDict = new Dictionary<string, List<Token>>();
// old fashioned way
// get every single word that is legit.
// add to dictionary.
// determine "best" version of each word (ie, highest score for each of the tokens)
foreach (Token s in finalHS)
{
//Console.WriteLine("testing {0}", s.term );
if (dictionary.ContainsKey(s.term ))
{
//Console.WriteLine("{0} in dictionary", s.term );
// initial score.
s.score = dictionary.getOrElse(s.term, 1.0);
//Console.WriteLine("term score :" + s.term + " : " + s.score.ToString() );
// this will create a new list for every single call... v v v wasteful. Need to modify.
var l = wordDict.getOrElseAssign(s.term, new List<Token>() );
l.Add(s);
}
}
// now for each collection of tokens that all represent the same word, go and rank them, and only take "the best"
foreach (var k in wordDict.Keys)
{
// all same word.
var l = wordDict[k];
var t = ranker.RankWordTokens(l);
// should just be BEST version of the word.
result.Add(t);
}
}
else
{
token.score = 1.0;
result.Add( token );
}
return result.ToList<Token>();
}
private int GetModificationTypeCount(Token t, ModificationType modType )
{
int c = 0;
foreach (var m in t.modifications)
{
if (m.modType == modType)
{
++c;
}
}
return c;
}
HashSet<Token> Edits1(List<Token> tokenList)
{
var hs = new HashSet<Token>();
// make sure original is added.
foreach (var w in tokenList)
{
hs.Add(w);
}
// deletion.
foreach (var w in tokenList)
{
//Console.WriteLine("word: " + w.term);
foreach (Tuple<string, string> t in split( w.term ))
{
if (t.Item2 != null && t.Item2 != "")
{
var newWord = t.Item1 + t.Item2.Substring(1);
var token = new Token( newWord );
token.origTerm = w.origTerm;
var mod = new Modification();
mod.modType = ModificationType.Delete;
mod.origChar = t.Item2[0];
token.modifications = w.modifications.GetRange(0, w.modifications.Count);
token.modifications.Add( mod );
hs.Add(token);
}
}
}
// transposes
foreach (var w in tokenList)
{
foreach (Tuple<string, string> t in split(w.term))
{
if (t.Item2.Length > 1)
{
// only if two letters aren't the same
if ( t.Item2[0] != t.Item2[1] )
{
var newWord = t.Item1 + t.Item2[1] + t.Item2[0] + t.Item2.Substring(2);
var token = new Token(newWord);
token.origTerm = w.origTerm;
var mod = new Modification();
mod.modType = ModificationType.Transpose;
mod.origChar = t.Item2[0];
mod.newChar = t.Item2[1];
token.modifications = w.modifications.GetRange(0, w.modifications.Count);
token.modifications.Add( mod );
hs.Add(token);
}
}
}
}
// replacement.
foreach (var w in tokenList)
{
foreach (Tuple<string, string> t in split(w.term))
{
if (t.Item2 != "")
{
foreach (char c in alphabet)
{
// only replace if different char.
if ( c != t.Item2[0] )
{
var newWord = t.Item1 + c + t.Item2.Substring(1);
var token = new Token(newWord);
token.origTerm = w.origTerm;
var mod = new Modification();
mod.modType = ModificationType.Replace;
mod.origChar = t.Item2[0];
mod.newChar = c;
token.modifications = w.modifications.GetRange(0, w.modifications.Count);
token.modifications.Add( mod );
hs.Add(token);
}
}
}
}
}
// inserts.
foreach (var w in tokenList)
{
foreach (Tuple<string, string> t in split(w.term))
{
if (t.Item2 != "")
{
foreach (char c in alphabet)
{
var newWord = t.Item1 + c + t.Item2;
var token = new Token(newWord);
token.origTerm = w.origTerm;
//Console.WriteLine("orig term " + token.origTerm );
//Console.WriteLine("new term " + token.term );
var mod = new Modification();
mod.modType = ModificationType.Insert;
mod.origChar = c;
token.modifications = w.modifications.GetRange(0, w.modifications.Count);
token.modifications.Add( mod );
hs.Add(token);
}
}
else
{
// just appending to end... seem legit.
foreach (char c in alphabet)
{
var newWord = t.Item1 + c;
var token = new Token(newWord);
token.origTerm = w.origTerm;
var mod = new Modification();
mod.modType = ModificationType.Insert;
mod.origChar = c;
//Console.WriteLine("orig term " + token.origTerm );
//Console.WriteLine("new term " + token.term );
token.modifications = w.modifications.GetRange(0, w.modifications.Count);
token.modifications.Add( mod );
hs.Add(token);
}
}
}
}
return hs;
}