org.apache.druid.query.groupby.epinephelinae.collection

Class MemoryOpenHashTable

java.lang.Object

org.apache.druid.query.groupby.epinephelinae.collection.MemoryOpenHashTable

public class MemoryOpenHashTable
extends Object

An open-addressed hash table with linear probing backed by WritableMemory. Does not offer a similar interface to Map because this is meant to be useful to lower-level, high-performance callers. There is no copying or serde of keys and values: callers access the backing memory of the table directly. This table will not grow itself. Callers must handle growing if required; the copyTo(org.apache.druid.query.groupby.epinephelinae.collection.MemoryOpenHashTable, org.apache.druid.query.groupby.epinephelinae.collection.MemoryOpenHashTable.BucketCopyHandler) method is provided to assist.

Nested Class Summary

Nested Classes

Modifier and Type	Class and Description
static interface	MemoryOpenHashTable.BucketCopyHandler Callback used by copyTo(org.apache.druid.query.groupby.epinephelinae.collection.MemoryOpenHashTable, org.apache.druid.query.groupby.epinephelinae.collection.MemoryOpenHashTable.BucketCopyHandler).

Constructor Summary

Constructors

Constructor and Description
MemoryOpenHashTable(org.apache.datasketches.memory.WritableMemory tableMemory, int numBuckets, int maxSize, int keySize, int valueSize) Create a new table.

Method Summary

Modifier and Type	Method and Description
it.unimi.dsi.fastutil.ints.IntIterator	bucketIterator() Iterates over all used buckets, returning bucket numbers for each one.
int	bucketKeyOffset() Returns the offset within each bucket where the key starts.
int	bucketMemoryPosition(int bucket) Returns the position within memory() where a particular bucket starts.
int	bucketSize() Returns the size in bytes of each bucket.
static int	bucketSize(int keySize, int valueSize) Returns the size of each bucket in a table.
int	bucketValueOffset() Returns the offset within each bucket where the value starts.
boolean	canInsertNewBucket() Returns whether this table can accept a new bucket.
void	clear() Clear the table, resetting size to zero.
void	copyTo(MemoryOpenHashTable other, MemoryOpenHashTable.BucketCopyHandler copyHandler) Copy this table into another one.
int	findBucket(int keyHash, org.apache.datasketches.memory.Memory keySpace, int keySpacePosition) Finds the bucket for a particular key.
void	initBucket(int bucket, org.apache.datasketches.memory.Memory keySpace, int keySpacePosition) Initialize a bucket with a particular key.
int	keySize() Returns the size of keys, in bytes.
org.apache.datasketches.memory.WritableMemory	memory() Returns the memory backing this table.
static int	memoryNeeded(int numBuckets, int bucketSize) Returns the amount of memory needed for a table.
int	numBuckets() Returns the number of buckets in this table.
int	size() Returns the number of elements currently in the table.
int	valueSize() Returns the size of values, in bytes.

Methods inherited from class java.lang.Object

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

Constructor Detail

MemoryOpenHashTable

public MemoryOpenHashTable(org.apache.datasketches.memory.WritableMemory tableMemory,
                           int numBuckets,
                           int maxSize,
                           int keySize,
                           int valueSize)

Create a new table.

Parameters:

tableMemory - backing memory for the table; must be exactly large enough to hold "numBuckets"

numBuckets - number of buckets for the table

maxSize - maximum number of elements for the table; must be less than numBuckets

keySize - key size in bytes

valueSize - value size in bytes

Method Detail

memoryNeeded

public static int memoryNeeded(int numBuckets,
                               int bucketSize)

Returns the amount of memory needed for a table. This is just a multiplication, which is easy enough to do on your own, but sometimes it's nice for clarity's sake to call a function with a name that indicates why the multiplication is happening.

Parameters:

numBuckets - number of buckets

bucketSize - size per bucket (in bytes)

Returns:

size of table (in bytes)

bucketSize

public static int bucketSize(int keySize,
                             int valueSize)

Returns the size of each bucket in a table.

Parameters:

keySize - size of keys (in bytes)

valueSize - size of values (in bytes)

Returns:

size of buckets (in bytes)

clear

public void clear()

Clear the table, resetting size to zero.

copyTo

public void copyTo(MemoryOpenHashTable other,
                   @Nullable
                   MemoryOpenHashTable.BucketCopyHandler copyHandler)

Copy this table into another one. The other table must be large enough to hold all the copied buckets. The other table will be cleared before the copy takes place.

Parameters:

other - the other table

copyHandler - a callback that is notified for each copied bucket

findBucket

public int findBucket(int keyHash,
                      org.apache.datasketches.memory.Memory keySpace,
                      int keySpacePosition)

Finds the bucket for a particular key.

Parameters:

keyHash - result of calling HashTableUtils.hashMemory(org.apache.datasketches.memory.Memory, long, int) on this key

keySpace - memory containing the key

keySpacePosition - position of the key within keySpace

Returns:

bucket number if currently occupied, or -bucket - 1 if not occupied (yet)

canInsertNewBucket

public boolean canInsertNewBucket()

Returns whether this table can accept a new bucket.

initBucket

public void initBucket(int bucket,
                       org.apache.datasketches.memory.Memory keySpace,
                       int keySpacePosition)

Initialize a bucket with a particular key. Do not call this method unless the bucket is currently unused and canInsertNewBucket() returns true.

Parameters:

bucket - bucket number

keySpace - memory containing the key

keySpacePosition - position of the key within keySpace

size

public int size()

Returns the number of elements currently in the table.

numBuckets

public int numBuckets()

Returns the number of buckets in this table. Note that not all of these can actually be used. The amount that can be used depends on the "maxSize" parameter provided during construction.

keySize

public int keySize()

Returns the size of keys, in bytes.

valueSize

public int valueSize()

Returns the size of values, in bytes.

bucketKeyOffset

public int bucketKeyOffset()

Returns the offset within each bucket where the key starts.

bucketValueOffset

public int bucketValueOffset()

Returns the offset within each bucket where the value starts.

bucketSize

public int bucketSize()

Returns the size in bytes of each bucket.

bucketMemoryPosition

public int bucketMemoryPosition(int bucket)

Returns the position within memory() where a particular bucket starts.

memory

public org.apache.datasketches.memory.WritableMemory memory()

Returns the memory backing this table.

bucketIterator

public it.unimi.dsi.fastutil.ints.IntIterator bucketIterator()

Iterates over all used buckets, returning bucket numbers for each one. The intent is that callers will pass the bucket numbers to bucketMemoryPosition(int) and then use bucketKeyOffset() and bucketValueOffset() to extract keys and values from the buckets as needed.