org.rocksdb

Class WriteBatchWithIndex

java.lang.Object

org.rocksdb.AbstractNativeReference

org.rocksdb.AbstractImmutableNativeReference

org.rocksdb.RocksObject

org.rocksdb.AbstractWriteBatch

org.rocksdb.WriteBatchWithIndex

All Implemented Interfaces:

java.lang.AutoCloseable, WriteBatchInterface

public class WriteBatchWithIndex
extends AbstractWriteBatch

Similar to WriteBatch but with a binary searchable index built for all the keys inserted. Calling put, merge, remove or putLogData calls the same function as with WriteBatch whilst also building an index. A user can call newIterator() to create an iterator over the write batch or newIteratorWithBase(org.rocksdb.RocksIterator) to get an iterator for the database with Read-Your-Own-Writes like capability

Field Summary

Fields inherited from class org.rocksdb.RocksObject

nativeHandle_

Fields inherited from class org.rocksdb.AbstractImmutableNativeReference

owningHandle_

Constructor Summary

Constructors

Constructor and Description
WriteBatchWithIndex() Creates a WriteBatchWithIndex where no bytes are reserved up-front, bytewise comparison is used for fallback key comparisons, and duplicate keys operations are retained
WriteBatchWithIndex(AbstractComparator fallbackIndexComparator, int reservedBytes, boolean overwriteKey) Creates a WriteBatchWithIndex
WriteBatchWithIndex(boolean overwriteKey) Creates a WriteBatchWithIndex where no bytes are reserved up-front, bytewise comparison is used for fallback key comparisons, and duplicate key assignment is determined by the constructor argument

Method Summary

Modifier and Type	Method and Description
protected void	disposeInternal(long handle)
byte[]	getFromBatch(ColumnFamilyHandle columnFamilyHandle, DBOptions options, byte[] key) Similar to RocksDB.get(ColumnFamilyHandle, byte[]) but will only read the key from this batch.
byte[]	getFromBatch(DBOptions options, byte[] key) Similar to RocksDB.get(byte[]) but will only read the key from this batch.
byte[]	getFromBatchAndDB(RocksDB db, ColumnFamilyHandle columnFamilyHandle, ReadOptions options, byte[] key) Similar to RocksDB.get(ColumnFamilyHandle, byte[]) but will also read writes from this batch.
byte[]	getFromBatchAndDB(RocksDB db, ReadOptions options, byte[] key) Similar to RocksDB.get(byte[]) but will also read writes from this batch.
WBWIRocksIterator	newIterator() Create an iterator of the default column family.
WBWIRocksIterator	newIterator(ColumnFamilyHandle columnFamilyHandle) Create an iterator of a column family.
RocksIterator	newIteratorWithBase(ColumnFamilyHandle columnFamilyHandle, RocksIterator baseIterator) Provides Read-Your-Own-Writes like functionality by creating a new Iterator that will use WBWIRocksIterator as a delta and baseIterator as a base Updating write batch with the current key of the iterator is not safe.
RocksIterator	newIteratorWithBase(ColumnFamilyHandle columnFamilyHandle, RocksIterator baseIterator, ReadOptions readOptions) Provides Read-Your-Own-Writes like functionality by creating a new Iterator that will use WBWIRocksIterator as a delta and baseIterator as a base Updating write batch with the current key of the iterator is not safe.
RocksIterator	newIteratorWithBase(RocksIterator baseIterator) Provides Read-Your-Own-Writes like functionality by creating a new Iterator that will use WBWIRocksIterator as a delta and baseIterator as a base.
RocksIterator	newIteratorWithBase(RocksIterator baseIterator, ReadOptions readOptions) Provides Read-Your-Own-Writes like functionality by creating a new Iterator that will use WBWIRocksIterator as a delta and baseIterator as a base.

Methods inherited from class org.rocksdb.AbstractWriteBatch

clear, count, delete, delete, delete, delete, deleteRange, deleteRange, getWriteBatch, merge, merge, popSavePoint, put, put, put, put, putLogData, rollbackToSavePoint, setMaxBytes, setSavePoint, singleDelete, singleDelete

Methods inherited from class org.rocksdb.RocksObject

disposeInternal, getNativeHandle

Methods inherited from class org.rocksdb.AbstractImmutableNativeReference

close, disOwnNativeHandle, isOwningHandle

Methods inherited from class java.lang.Object

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

Constructor Detail

WriteBatchWithIndex

public WriteBatchWithIndex()

Creates a WriteBatchWithIndex where no bytes are reserved up-front, bytewise comparison is used for fallback key comparisons, and duplicate keys operations are retained

WriteBatchWithIndex

public WriteBatchWithIndex(boolean overwriteKey)

Creates a WriteBatchWithIndex where no bytes are reserved up-front, bytewise comparison is used for fallback key comparisons, and duplicate key assignment is determined by the constructor argument

Parameters:

overwriteKey - if true, overwrite the key in the index when inserting a duplicate key, in this way an iterator will never show two entries with the same key.

WriteBatchWithIndex

public WriteBatchWithIndex(AbstractComparator fallbackIndexComparator,
                           int reservedBytes,
                           boolean overwriteKey)

Creates a WriteBatchWithIndex

Parameters:

fallbackIndexComparator - We fallback to this comparator to compare keys within a column family if we cannot determine the column family and so look up it's comparator.

reservedBytes - reserved bytes in underlying WriteBatch

overwriteKey - if true, overwrite the key in the index when inserting a duplicate key, in this way an iterator will never show two entries with the same key.

Method Detail

newIterator

public WBWIRocksIterator newIterator(ColumnFamilyHandle columnFamilyHandle)

Create an iterator of a column family. User can call RocksIteratorInterface.seek(byte[]) to search to the next entry of or after a key. Keys will be iterated in the order given by index_comparator. For multiple updates on the same key, each update will be returned as a separate entry, in the order of update time.

Parameters:

columnFamilyHandle - The column family to iterate over

Returns:

An iterator for the Write Batch contents, restricted to the column family

newIterator

public WBWIRocksIterator newIterator()

Create an iterator of the default column family. User can call RocksIteratorInterface.seek(byte[]) to search to the next entry of or after a key. Keys will be iterated in the order given by index_comparator. For multiple updates on the same key, each update will be returned as a separate entry, in the order of update time.

Returns:

An iterator for the Write Batch contents

newIteratorWithBase

public RocksIterator newIteratorWithBase(ColumnFamilyHandle columnFamilyHandle,
                                         RocksIterator baseIterator)

Provides Read-Your-Own-Writes like functionality by creating a new Iterator that will use WBWIRocksIterator as a delta and baseIterator as a base Updating write batch with the current key of the iterator is not safe. We strongly recommend users not to do it. It will invalidate the current key() and value() of the iterator. This invalidation happens even before the write batch update finishes. The state may recover after Next() is called.

Parameters:

columnFamilyHandle - The column family to iterate over

baseIterator - The base iterator, e.g. RocksDB.newIterator()

Returns:

An iterator which shows a view comprised of both the database point-in-time from baseIterator and modifications made in this write batch.

newIteratorWithBase

public RocksIterator newIteratorWithBase(ColumnFamilyHandle columnFamilyHandle,
                                         RocksIterator baseIterator,
                                         ReadOptions readOptions)

Provides Read-Your-Own-Writes like functionality by creating a new Iterator that will use WBWIRocksIterator as a delta and baseIterator as a base Updating write batch with the current key of the iterator is not safe. We strongly recommend users not to do it. It will invalidate the current key() and value() of the iterator. This invalidation happens even before the write batch update finishes. The state may recover after Next() is called.

Parameters:

columnFamilyHandle - The column family to iterate over

baseIterator - The base iterator, e.g. RocksDB.newIterator()

readOptions - the read options, or null

Returns:

An iterator which shows a view comprised of both the database point-in-time from baseIterator and modifications made in this write batch.

newIteratorWithBase

public RocksIterator newIteratorWithBase(RocksIterator baseIterator)

Provides Read-Your-Own-Writes like functionality by creating a new Iterator that will use WBWIRocksIterator as a delta and baseIterator as a base. Operates on the default column family.

Parameters:

baseIterator - The base iterator, e.g. RocksDB.newIterator()

Returns:

An iterator which shows a view comprised of both the database point-in-timefrom baseIterator and modifications made in this write batch.

newIteratorWithBase

public RocksIterator newIteratorWithBase(RocksIterator baseIterator,
                                         ReadOptions readOptions)

Provides Read-Your-Own-Writes like functionality by creating a new Iterator that will use WBWIRocksIterator as a delta and baseIterator as a base. Operates on the default column family.

Parameters:

baseIterator - The base iterator, e.g. RocksDB.newIterator()

readOptions - the read options, or null

Returns:

An iterator which shows a view comprised of both the database point-in-timefrom baseIterator and modifications made in this write batch.

getFromBatch

public byte[] getFromBatch(ColumnFamilyHandle columnFamilyHandle,
                           DBOptions options,
                           byte[] key)
                    throws RocksDBException

Similar to RocksDB.get(ColumnFamilyHandle, byte[]) but will only read the key from this batch.

Parameters:

columnFamilyHandle - The column family to retrieve the value from

options - The database options to use

key - The key to read the value for

Returns:

a byte array storing the value associated with the input key if any. null if it does not find the specified key.

Throws:

RocksDBException - if the batch does not have enough data to resolve Merge operations, MergeInProgress status may be returned.

getFromBatch

public byte[] getFromBatch(DBOptions options,
                           byte[] key)
                    throws RocksDBException

Similar to RocksDB.get(byte[]) but will only read the key from this batch.

Parameters:

options - The database options to use

key - The key to read the value for

Returns:

a byte array storing the value associated with the input key if any. null if it does not find the specified key.

Throws:

RocksDBException - if the batch does not have enough data to resolve Merge operations, MergeInProgress status may be returned.

getFromBatchAndDB

public byte[] getFromBatchAndDB(RocksDB db,
                                ColumnFamilyHandle columnFamilyHandle,
                                ReadOptions options,
                                byte[] key)
                         throws RocksDBException

Similar to RocksDB.get(ColumnFamilyHandle, byte[]) but will also read writes from this batch. This function will query both this batch and the DB and then merge the results using the DB's merge operator (if the batch contains any merge requests). Setting ReadOptions.setSnapshot(Snapshot) will affect what is read from the DB but will NOT change which keys are read from the batch (the keys in this batch do not yet belong to any snapshot and will be fetched regardless).

Parameters:

db - The Rocks database

columnFamilyHandle - The column family to retrieve the value from

options - The read options to use

key - The key to read the value for

Returns:

a byte array storing the value associated with the input key if any. null if it does not find the specified key.

Throws:

RocksDBException - if the value for the key cannot be read

getFromBatchAndDB

public byte[] getFromBatchAndDB(RocksDB db,
                                ReadOptions options,
                                byte[] key)
                         throws RocksDBException

Similar to RocksDB.get(byte[]) but will also read writes from this batch. This function will query both this batch and the DB and then merge the results using the DB's merge operator (if the batch contains any merge requests). Setting ReadOptions.setSnapshot(Snapshot) will affect what is read from the DB but will NOT change which keys are read from the batch (the keys in this batch do not yet belong to any snapshot and will be fetched regardless).

Parameters:

db - The Rocks database

options - The read options to use

key - The key to read the value for

Returns:

a byte array storing the value associated with the input key if any. null if it does not find the specified key.

Throws:

RocksDBException - if the value for the key cannot be read

disposeInternal

protected final void disposeInternal(long handle)

Specified by:

disposeInternal in class RocksObject