public void AnalyzeHashFunctionDistribution()
{
var seenHashes = new HashSet <int>();
var collisions = new List <int>();
const int hashCount = 200_000;
const int sourceWidth = 40;
const int minLength = 3;
var remaining = hashCount;
while (remaining > 0)
{
for (var start = 0; start < sourceWidth; start++)
{
var maxLength = sourceWidth - start;
for (var length = minLength; length < maxLength && remaining != 0; length++, remaining--)
{
var s = CreateRandomString(sourceWidth);
var hash = StringPool.GetSubstringHashCode(s, start, length);
if (!seenHashes.Add(hash))
{
collisions.Add(hash);
}
}
}
}
var crowdedBucketCount = collisions.Distinct().Count();
var max = collisions.GroupBy(hash => hash).Select(g => g.Count()).Max();
Console.Out.WriteLine(
$"Strings hashed {hashCount:#,0} with minimum length {minLength}\n" +
$"{collisions.Count} collisions over {crowdedBucketCount} indices\n" +
$"Most crowded bucket had {max} items\n" +
$"Average clashing bucket length {(double)collisions.Count / crowdedBucketCount:0.###}");
}
public void GetSubstringHashCode()
{
const string s = "01234567890abcdefghijklmnopqrstuvwxyzABCDEFGHIJKLMNOPQRSTUVWXYZ`¬¦!\"£$€%^&*()-_=+[]{};:'@#~/?,.<>\\|";
for (var i = 0; i < 10; i++)
{
Console.Out.WriteLine($"0x{StringPool.GetSubstringHashCode(s, 0, i):X2}");
}
Assert.AreEqual(0x162A16FE, (uint)StringPool.GetSubstringHashCode(s, 0, 0));
Assert.AreEqual(0x05E19FCE, (uint)StringPool.GetSubstringHashCode(s, 0, 1));
Assert.AreEqual(0x05E4405D, (uint)StringPool.GetSubstringHashCode(s, 0, 2));
Assert.AreEqual(0xFC2C8D57, (uint)StringPool.GetSubstringHashCode(s, 0, 3));
Assert.AreEqual(0xFC833FEA, (uint)StringPool.GetSubstringHashCode(s, 0, 4));
Assert.AreEqual(0x36D35466, (uint)StringPool.GetSubstringHashCode(s, 0, 5));
Assert.AreEqual(0x42005971, (uint)StringPool.GetSubstringHashCode(s, 0, 6));
Assert.AreEqual(0x4B54D52B, (uint)StringPool.GetSubstringHashCode(s, 0, 7));
Assert.AreEqual(0xBC227B3E, (uint)StringPool.GetSubstringHashCode(s, 0, 8));
Assert.AreEqual(0xCB2B1F36, (uint)StringPool.GetSubstringHashCode(s, 0, 9));
Assert.AreEqual(3779978672, (uint)StringPool.GetSubstringHashCode(s, 0, s.Length));
}