Skip to content

Make BFHash take on non negative values only, issue #229 #243

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
merged 3 commits into from
Dec 13, 2013
Merged
Show file tree
Hide file tree
Changes from 2 commits
Commits
File filter

Filter by extension

Filter by extension

Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
Original file line number Diff line number Diff line change
Expand Up @@ -333,7 +333,7 @@ case class BFHash(numHashes: Int, width: Int, seed: Long = 0L) extends Function1
}else
digested

Stream.cons(d(0) % width, nextHash(bytes, k - 1, d.drop(1)))
Stream.cons(math.abs(d(0) % width), nextHash(bytes, k - 1, d.drop(1)))
Copy link
Collaborator

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

How about moving the extra abs into the splitLong function so all the abs/uglyness is together?

Also adding a comment beside it as to why we have two math.abs might be good too?

Copy link
Contributor Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Moving abs into the splitLong wouldn't help because of the Integer.MIN_VALUE issue.
I'd rather use @david206's solution.

}
}
}
Expand Down
Original file line number Diff line number Diff line change
Expand Up @@ -24,9 +24,68 @@ object BloomFilterLaws extends Properties("BloomFilter") {
property("BloomFilter is a Monoid") = monoidLaws[BF]
}

object BFHashIndices extends Properties("BFHash") {
import org.scalacheck.Prop.forAll

val NUM_HASHES = 10
val WIDTH = 4752800

val SEED = 1

class BloomFilterTest extends Specification {
implicit val bfHashIndices: Arbitrary[Stream[Int]] =
Arbitrary {
for {
hashes <- choose(1, 10)
width <- choose(100, 5000000)
v <- choose(0, 100000)
} yield {BFHash(hashes, width, SEED).apply(v.toString)}
}

property("Indices are non negative") = forAll{ hashIndices: Stream[Int] => hashIndices.forall(_ >= 0)}

/**
* This is the version of the BFHash before the negative values fix
*/
case class NegativeBFHash(numHashes: Int, width: Int, seed: Long = 0L) extends Function1[String, Iterable[Int]]{
val size = numHashes

def apply(s: String) = nextHash(s.getBytes, numHashes)

private def splitLong(x: Long) = {
val upper = math.abs(x >> 32).toInt
val lower = math.abs((x << 32) >> 32).toInt
(upper, lower)
}

private def nextHash(bytes: Array[Byte], k: Int, digested: Seq[Int] = Seq.empty): Stream[Int] = {
if(k == 0)
Stream.empty
else{
val d = if(digested.isEmpty){
val (a, b) = MurmurHash128(k)(bytes)
val (x1, x2) = splitLong(a)
val (x3, x4) = splitLong(b)
Seq(x1, x2, x3, x4)
}else
digested

Stream.cons(d(0) % width, nextHash(bytes, k - 1, d.drop(1)))
}
}
}

val negativeBFHash = NegativeBFHash(NUM_HASHES, WIDTH, SEED)
val bfHash = BFHash(NUM_HASHES, WIDTH, SEED)
Copy link
Collaborator

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Use scalacheck to generate these rather than being fixed magic values

Copy link
Collaborator

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Well, ALSO have fixed magic values. These were found to have the bug, so let's keep them to prevent regression.

Copy link
Collaborator

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Well for the Specification section rather than the scalacheck one we should try generate the exact problem case, not sure it would make any benefit here.

Copy link
Contributor Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

There is a test for the problem case in the Specification already.
I'll make the property generic with an arbitrary "negative" hash.


property("Indices of the two versions of BFHashes are the same, unless the first one contains negative index") = forAll{ long: Long =>
val s = long.toString
val indices = negativeBFHash.apply(s)
indices == bfHash.apply(s) || indices.exists(_ < 0)
Copy link
Collaborator

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Given how rare the bug was without the or in place here has it ever thrown for you?

Copy link
Contributor Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

It's thrown exactly when (x >> 32) happens to be the Integer.MIN_VALUE, and then its absolute value is negative.

}
}


class BloomFilterTest extends Specification {

val SEED = 1
val RAND = new scala.util.Random
Expand Down Expand Up @@ -119,5 +178,18 @@ class BloomFilterTest extends Specification {
val bytesAfterSizeCalled = new String(serialize(bf))
bytesBeforeSizeCalled mustEqual bytesAfterSizeCalled
}

/**
* this test failed before the fix for https://github.com/twitter/algebird/issues/229
*/
"not have negative hash values" in {
val NUM_HASHES = 2
val WIDTH = 4752800
val bfHash = BFHash(NUM_HASHES, WIDTH, SEED)
val s = "7024497610539761509"
val index = bfHash.apply(s).head

index must be_>=(0)
}
}
}