private void Test(bool isCaseSensitive)
{
var comparer = isCaseSensitive ? StringComparer.Ordinal : StringComparer.OrdinalIgnoreCase;
var strings = GenerateStrings(2000).Skip(500).Take(1000).ToSet(comparer);
var testStrings = GenerateStrings(100000);
for (var d = 0.1; d >= 0.0001; d /= 10)
{
var filter = new BloomFilter(strings.Count, d, isCaseSensitive);
filter.AddRange(strings);
var correctCount = 0.0;
var incorrectCount = 0.0;
foreach (var test in testStrings)
{
var actualContains = strings.Contains(test);
var filterContains = filter.ProbablyContains(test);
if (!filterContains)
{
// if the filter says no, then it can't be in the real set.
Assert.False(actualContains);
}
if (actualContains == filterContains)
{
correctCount++;
}
else
{
incorrectCount++;
}
}
var falsePositivePercentage = incorrectCount / (correctCount + incorrectCount);
Assert.True(falsePositivePercentage < (d * 1.5), string.Format("falsePositivePercentage={0}, d={1}", falsePositivePercentage, d));
}
}
public void TestEmpty()
{
for (var d = 0.1; d >= 0.0001; d /= 10)
{
var filter = new BloomFilter(0, d, isCaseSensitive: true);
Assert.False(filter.ProbablyContains(string.Empty));
Assert.False(filter.ProbablyContains("a"));
Assert.False(filter.ProbablyContains("b"));
Assert.False(filter.ProbablyContains("c"));
var testStrings = GenerateStrings(100000);
foreach (var test in testStrings)
{
Assert.False(filter.ProbablyContains(test));
}
}
}