org.apache.hadoop.hive.ql.exec.vector.mapjoin.fast

Class VectorMapJoinFastBytesHashKeyRef

java.lang.Object

org.apache.hadoop.hive.ql.exec.vector.mapjoin.fast.VectorMapJoinFastBytesHashKeyRef

public class VectorMapJoinFastBytesHashKeyRef
extends Object

Nested Class Summary

Nested Classes

Modifier and Type	Class and Description
static class	VectorMapJoinFastBytesHashKeyRef.KeyRef

Constructor Summary

Constructors

Constructor and Description
VectorMapJoinFastBytesHashKeyRef()

Method Summary

Modifier and Type	Method and Description
static int	calculateHashCode(long refWord, WriteBuffers writeBuffers, WriteBuffers.Position readPos) This method calculates the hash code of the key referred by the refWord.
static boolean	equalKey(long refWord, byte[] keyBytes, int keyStart, int keyLength, WriteBuffers writeBuffers, WriteBuffers.Position readPos)
static long	extractPartialHashCode(long hashCode) Extract partial hash code from the full hash code.
static long	getPartialHashCodeFromRefWord(long refWord) Get partial hash code from the reference word.

Methods inherited from class java.lang.Object

clone, equals, finalize, getClass, hashCode, notify, notifyAll, toString, wait, wait, wait

Constructor Detail

VectorMapJoinFastBytesHashKeyRef

public VectorMapJoinFastBytesHashKeyRef()

Method Detail

equalKey

public static boolean equalKey(long refWord,
                               byte[] keyBytes,
                               int keyStart,
                               int keyLength,
                               WriteBuffers writeBuffers,
                               WriteBuffers.Position readPos)

calculateHashCode

public static int calculateHashCode(long refWord,
                                    WriteBuffers writeBuffers,
                                    WriteBuffers.Position readPos)

This method calculates the hash code of the key referred by the refWord. The refWord has the offset of the key in the writeBuffers. In case of smaller key, of length less than 256 bytes, the length is stored in refWord itself. For larger keys, the length is stored in the writeBuffers along with the key. The length is stored first followed by the key. So in case of big key, readPos is used to get the length first. This will set the read position to the beginning of key. Then writeBuffers.hashCode is used to get the hash code. For smaller keys, unsafeHashCode is used, which will set the read position to keyAbsoluteOffset.

Parameters:

refWord - The long value containing the offset of the key in writeBuffers.

writeBuffers - Buffer used for storing the keys present in the hash table.

readPos - temporary position to be used in this method.

Returns:

hash code for the key pointed to by refWord

extractPartialHashCode

public static long extractPartialHashCode(long hashCode)

Extract partial hash code from the full hash code. Choose the high bits of the hash code KNOWING it was calculated as an int. We want the partial hash code to be different than the lower bits used for our hash table slot calculations.

Parameters:

hashCode -

Returns:

getPartialHashCodeFromRefWord

public static long getPartialHashCodeFromRefWord(long refWord)

Get partial hash code from the reference word.

Returns: