private void IdentifyShow(File file, FuzzyString showName, int keywordWeight, Keywords keywords)
{
var shows = this.Database.GetShows();
var showMatches = shows.Select(s => new WeightedGuess<Show>(s, s.Name.GetSimilarity(showName)) )
.Where(s => s.Weight >= (int) FuzzyString.Resemblance.VerySimilar);
foreach (var entry in showMatches)
{
entry.Weight *= keywordWeight;
file.ShowMatches.Add(entry);
}
}
private void AddKeyword(string word, int weight)
{
var fuzzy = new FuzzyString(word);
var existing = this.List.FirstOrDefault(k => k.Guess.Compare(fuzzy) >= FuzzyString.Resemblance.VerySimilar);
if (existing != null)
{
existing.Weight += weight;
}
else
{
this.List.Add(new WeightedGuess<FuzzyString>(fuzzy, weight));
}
}
public int GetSimilarity(FuzzyString other)
{
var key = new StringPair(this.CleanString, other.CleanString);
int value;
if (!CachedComparisons.TryGetValue(key, out value))
{
var similarity = SiftStringSimilarity.StringMetrics.SiftSimilarity(this.CleanString, other.CleanString);
value = (int)(similarity * 100);
CachedComparisons.Add(key, value);
}
return value;
}
public Resemblance Compare(FuzzyString other)
{
var similarity = GetSimilarity(other);
if (similarity == (int)Resemblance.Identical)
{
return Resemblance.Identical;
}
if (similarity >= (int)Resemblance.VerySimilar)
{
return Resemblance.VerySimilar;
}
if (similarity >= (int)Resemblance.Similar)
{
return Resemblance.Similar;
}
return Resemblance.Different;
}
public void ShouldReplaceAmpersandWithAnd()
{
var similarity = new FuzzyString("bob and bill").Compare(new FuzzyString("bob & bill"));
Assert.AreEqual(FuzzyString.Resemblance.Identical, similarity);
}
public void ShouldRecognizeIdenticalStrings()
{
var similarity = new FuzzyString("dot net").Compare(new FuzzyString("dot net"));
Assert.AreEqual(FuzzyString.Resemblance.Identical, similarity);
}
public void ShouldIgnorePunctuationAndNoiseCharacters()
{
var similarity = new FuzzyString("hello world").Compare(new FuzzyString(".+hello-'world_="));
Assert.AreEqual(FuzzyString.Resemblance.Identical, similarity);
}
public void ShouldIgnoreCommonWords()
{
var similarity = new FuzzyString("the code").Compare(new FuzzyString("a code"));
Assert.AreEqual(FuzzyString.Resemblance.Identical, similarity);
}
public void ShouldIdentifyVerySimilarWords()
{
var similarity = new FuzzyString("the simpsons").Compare(new FuzzyString("the simsons"));
Assert.AreEqual(FuzzyString.Resemblance.VerySimilar, similarity);
}
public void ShouldIdentifySimilarWords()
{
var similarity = new FuzzyString("south park").Compare(new FuzzyString("north park"));
Assert.AreEqual(FuzzyString.Resemblance.Similar, similarity);
}
public void ShouldIdentifyDifferentWords()
{
var similarity = new FuzzyString("will and grace").Compare(new FuzzyString("the sopranos"));
Assert.AreEqual(FuzzyString.Resemblance.Different, similarity);
}
public void ShouldBeCaseInsensitive()
{
var similarity = new FuzzyString("Dot nEt").Compare(new FuzzyString("dot Net"));
Assert.AreEqual(FuzzyString.Resemblance.Identical, similarity);
}
