internal bool IsBetterHash(PropertyHash otherHash)
{
if (otherHash.CollisionCount < CollisionCount)
{
return(true);
}
if (otherHash.CollisionCount > CollisionCount)
{
return(false);
}
//same number collisions, use the one with the smallest mod
if (otherHash.ModValue > ModValue)
{
return(false);
}
return(true);
}
internal PropertyHash FindBestHash(string[] stringProperties, bool utf8)
{
var properties = stringProperties.Select(p => new Property(p, utf8)).ToArray();
var bestHash = new PropertyHash()
{
Column = 0,
ModValue = int.MaxValue,
CollisionCount = int.MaxValue,
Utf8 = utf8
};
//try column values
var sortedColumnCollisions = FindColumnCollisions(properties).OrderBy(collisions => collisions.NumberOfCollisions);
foreach (var columnCollision in sortedColumnCollisions.Take(3))
{
(var bestMod, var collisionCount) = FindBestMod(properties, property => property[columnCollision.ColumnIndex % property.Length]);
var candidateHash = new PropertyHash()
{
Column = columnCollision.ColumnIndex,
ModValue = bestMod,
CollisionCount = collisionCount,
};
if (bestHash.IsBetterHash(candidateHash))
{
bestHash = candidateHash;
}
}
//try length
(var bestModForLength, var collisionCountLength) = FindBestMod(properties, property => property.Length);
var lengthCandidateHash = new PropertyHash()
{
UseLength = true,
ModValue = bestModForLength,
CollisionCount = collisionCountLength
};
if (bestHash.IsBetterHash(lengthCandidateHash))
{
bestHash = lengthCandidateHash;
}
return(bestHash);
}