summaryrefslogtreecommitdiff
diff options
context:
space:
mode:
authorRĂ¼diger Klaehn <rklaehn@gmail.com>2014-01-15 21:12:43 +0100
committerRĂ¼diger Klaehn <rklaehn@gmail.com>2014-01-15 21:12:43 +0100
commit47a91d76fc08dbac889f37915b7c08534bfa74a8 (patch)
treecb51e18fd452d16226cbbee10d5ab19c54f9f208
parentafcfba02edf342ea020f8c45c3ef83168c45e5ae (diff)
downloadscala-47a91d76fc08dbac889f37915b7c08534bfa74a8.tar.gz
scala-47a91d76fc08dbac889f37915b7c08534bfa74a8.tar.bz2
scala-47a91d76fc08dbac889f37915b7c08534bfa74a8.zip
SI-6200 - HashMap should implement filter
This is the exact same algorithm as in SI-6196, with only slight differences due to the two type arguments of HashMap. Filter is tested by the new comparative collection test by @Ichoran, but there is nevertheless a small correctness test in t6200 in addition to tests of the new desirable behavior.
-rw-r--r--src/library/scala/collection/immutable/HashMap.scala122
-rw-r--r--test/files/run/t6200.scala68
2 files changed, 190 insertions, 0 deletions
diff --git a/src/library/scala/collection/immutable/HashMap.scala b/src/library/scala/collection/immutable/HashMap.scala
index 0a8524c139..e7d87dd3fe 100644
--- a/src/library/scala/collection/immutable/HashMap.scala
+++ b/src/library/scala/collection/immutable/HashMap.scala
@@ -40,6 +40,8 @@ class HashMap[A, +B] extends AbstractMap[A, B]
with Serializable
with CustomParallelizable[(A, B), ParHashMap[A, B]]
{
+ import HashMap.{nullToEmpty, bufferSize}
+
override def size: Int = 0
override def empty = HashMap.empty[A, B]
@@ -63,6 +65,18 @@ class HashMap[A, +B] extends AbstractMap[A, B]
def - (key: A): HashMap[A, B] =
removed0(key, computeHash(key), 0)
+ override def filter(p: ((A, B)) => Boolean) = {
+ val buffer = new Array[HashMap[A, B]](bufferSize(size))
+ nullToEmpty(filter0(p, false, 0, buffer, 0))
+ }
+
+ override def filterNot(p: ((A, B)) => Boolean) = {
+ val buffer = new Array[HashMap[A, B]](bufferSize(size))
+ nullToEmpty(filter0(p, true, 0, buffer, 0))
+ }
+
+ protected def filter0(p: ((A, B)) => Boolean, negate: Boolean, level: Int, buffer: Array[HashMap[A, B @uV]], offset0: Int): HashMap[A, B] = null
+
protected def elemHashCode(key: A) = key.##
protected final def improve(hcode: Int) = {
@@ -200,6 +214,9 @@ object HashMap extends ImmutableMapFactory[HashMap] with BitOperations.Int {
override def removed0(key: A, hash: Int, level: Int): HashMap[A, B] =
if (hash == this.hash && key == this.key) HashMap.empty[A,B] else this
+ override protected def filter0(p: ((A, B)) => Boolean, negate: Boolean, level: Int, buffer: Array[HashMap[A, B @uV]], offset0: Int): HashMap[A, B] =
+ if (negate ^ p(ensurePair)) this else null
+
override def iterator: Iterator[(A,B)] = Iterator(ensurePair)
override def foreach[U](f: ((A, B)) => U): Unit = f(ensurePair)
// this method may be called multiple times in a multithreaded environment, but that's ok
@@ -241,6 +258,21 @@ object HashMap extends ImmutableMapFactory[HashMap] with BitOperations.Int {
new HashMapCollision1(hash, kvs1)
} else this
+ override protected def filter0(p: ((A, B)) => Boolean, negate: Boolean, level: Int, buffer: Array[HashMap[A, B @uV]], offset0: Int): HashMap[A, B] = {
+ val kvs1 = if(negate) kvs.filterNot(p) else kvs.filter(p)
+ kvs1.size match {
+ case 0 =>
+ null
+ case 1 =>
+ val kv@(k,v) = kvs1.head
+ new HashMap1(k, hash, v, kv)
+ case x if x == kvs.size =>
+ this
+ case _ =>
+ new HashMapCollision1(hash, kvs1)
+ }
+ }
+
override def iterator: Iterator[(A,B)] = kvs.iterator
override def foreach[U](f: ((A, B)) => U): Unit = kvs.foreach(f)
override def split: Seq[HashMap[A, B]] = {
@@ -336,6 +368,52 @@ object HashMap extends ImmutableMapFactory[HashMap] with BitOperations.Int {
}
}
+ override protected def filter0(p: ((A, B)) => Boolean, negate: Boolean, level: Int, buffer: Array[HashMap[A, B @uV]], offset0: Int): HashMap[A, B] = {
+ // current offset
+ var offset = offset0
+ // result size
+ var rs = 0
+ // bitmap for kept elems
+ var kept = 0
+ // loop over all elements
+ var i = 0
+ while (i < elems.length) {
+ val result = elems(i).filter0(p, negate, level + 5, buffer, offset)
+ if (result ne null) {
+ buffer(offset) = result
+ offset += 1
+ // add the result size
+ rs += result.size
+ // mark the bit i as kept
+ kept |= (1 << i)
+ }
+ i += 1
+ }
+ if (offset == offset0) {
+ // empty
+ null
+ } else if (rs == size0) {
+ // unchanged
+ this
+ } else if (offset == offset0 + 1 && !buffer(offset0).isInstanceOf[HashTrieMap[A, B]]) {
+ // leaf
+ buffer(offset0)
+ } else {
+ // we have to return a HashTrieMap
+ val length = offset - offset0
+ val elems1 = new Array[HashMap[A, B]](length)
+ System.arraycopy(buffer, offset0, elems1, 0, length)
+ val bitmap1 = if (length == elems.length) {
+ // we can reuse the original bitmap
+ bitmap
+ } else {
+ // calculate new bitmap by keeping just bits in the kept bitmask
+ keepBits(bitmap, kept)
+ }
+ new HashTrieMap(bitmap1, elems1, rs)
+ }
+ }
+
override def iterator: Iterator[(A, B)] = new TrieIterator[(A, B)](elems.asInstanceOf[Array[Iterable[(A, B)]]]) {
final override def getElem(cc: AnyRef): (A, B) = cc.asInstanceOf[HashMap1[A, B]].ensurePair
}
@@ -439,6 +517,50 @@ object HashMap extends ImmutableMapFactory[HashMap] with BitOperations.Int {
}
}
+ /**
+ * Calculates the maximum buffer size given the maximum possible total size of the trie-based collection
+ * @param size the maximum size of the collection to be generated
+ * @return the maximum buffer size
+ */
+ @inline private def bufferSize(size: Int): Int = (size + 6) min (32 * 7)
+
+ /**
+ * In many internal operations the empty map is represented as null for performance reasons. This method converts
+ * null to the empty map for use in public methods
+ */
+ @inline private def nullToEmpty[A, B](m: HashMap[A, B]): HashMap[A, B] = if (m eq null) empty[A, B] else m
+
+ /**
+ * Utility method to keep a subset of all bits in a given bitmap
+ *
+ * Example
+ * bitmap (binary): 00000001000000010000000100000001
+ * keep (binary): 1010
+ * result (binary): 00000001000000000000000100000000
+ *
+ * @param bitmap the bitmap
+ * @param keep a bitmask containing which bits to keep
+ * @return the original bitmap with all bits where keep is not 1 set to 0
+ */
+ private def keepBits(bitmap: Int, keep: Int): Int = {
+ var result = 0
+ var current = bitmap
+ var kept = keep
+ while (kept != 0) {
+ // lowest remaining bit in current
+ val lsb = current ^ (current & (current - 1))
+ if ((kept & 1) != 0) {
+ // mark bit in result bitmap
+ result |= lsb
+ }
+ // clear lowest remaining one bit in abm
+ current &= ~lsb
+ // look at the next kept bit
+ kept >>>= 1
+ }
+ result
+ }
+
@SerialVersionUID(2L)
private class SerializationProxy[A,B](@transient private var orig: HashMap[A, B]) extends Serializable {
private def writeObject(out: java.io.ObjectOutputStream) {
diff --git a/test/files/run/t6200.scala b/test/files/run/t6200.scala
new file mode 100644
index 0000000000..02659c26bc
--- /dev/null
+++ b/test/files/run/t6200.scala
@@ -0,0 +1,68 @@
+import scala.collection.immutable.HashMap
+
+object Test extends App {
+
+ case class Collision(value: Int) extends Ordered[Collision] {
+ def compare(that: Collision) = value compare that.value
+
+ override def hashCode = value / 5
+ }
+
+ def testCorrectness[T: Ordering](n: Int, mkKey: Int => T) {
+ val o = implicitly[Ordering[T]]
+ val s = HashMap.empty[T, Unit] ++ (0 until n).map(x => mkKey(x) ->())
+ for (i <- 0 until n) {
+ val ki = mkKey(i)
+ val a = s.filter(kv => o.lt(kv._1, ki))
+ val b = s.filterNot(kv => o.lt(kv._1, ki))
+ require(a.size == i && (0 until i).forall(i => a.contains(mkKey(i))))
+ require(b.size == n - i && (i until n).forall(i => b.contains(mkKey(i))))
+ }
+ }
+
+ // this tests the structural sharing of the new filter
+ // I could not come up with a simple test that tests structural sharing when only parts are reused, but
+ // at least this fails with the old and passes with the new implementation
+ def testSharing() {
+ val s = HashMap.empty[Int, Unit] ++ (0 until 100).map(_ ->())
+ require(s.filter(_ => true) eq s)
+ require(s.filterNot(_ => false) eq s)
+ }
+
+ // this tests that neither hashCode nor equals are called during filter
+ def testNoHashing() {
+ var hashCount = 0
+ var equalsCount = 0
+ case class HashCounter(value: Int) extends Ordered[HashCounter] {
+ def compare(that: HashCounter) = value compare that.value
+
+ override def hashCode = {
+ hashCount += 1
+ value
+ }
+
+ override def equals(that: Any) = {
+ equalsCount += 1
+ this match {
+ case HashCounter(value) => this.value == value
+ case _ => false
+ }
+ }
+ }
+
+ val s = HashMap.empty[HashCounter, Unit] ++ (0 until 100).map(k => HashCounter(k) ->())
+ val hashCount0 = hashCount
+ val equalsCount0 = equalsCount
+ val t = s.filter(_._1 < HashCounter(50))
+ require(hashCount == hashCount0)
+ require(equalsCount == equalsCount0)
+ }
+
+ // this tests correctness of filter and filterNot for integer keys
+ testCorrectness[Int](100, identity _)
+ // this tests correctness of filter and filterNot for keys with lots of collisions
+ // this is necessary because usually collisions are rare so the collision-related code is not thoroughly tested
+ testCorrectness[Collision](100, Collision.apply _)
+ testSharing()
+ testNoHashing()
+}