org.apache.cassandra.index.sai.disk.v1.bitpack.AbstractBlockPackedReader Maven / Gradle / Ivy
Go to download
Show more of this group Show more artifacts with this name
Show all versions of cassandra-all Show documentation
Show all versions of cassandra-all Show documentation
The Apache Cassandra Project develops a highly scalable second-generation distributed database, bringing together Dynamo's fully distributed design and Bigtable's ColumnFamily-based data model.
/*
* Licensed to the Apache Software Foundation (ASF) under one
* or more contributor license agreements. See the NOTICE file
* distributed with this work for additional information
* regarding copyright ownership. The ASF licenses this file
* to you under the Apache License, Version 2.0 (the
* "License"); you may not use this file except in compliance
* with the License. You may obtain a copy of the License at
*
* http://www.apache.org/licenses/LICENSE-2.0
*
* Unless required by applicable law or agreed to in writing, software
* distributed under the License is distributed on an "AS IS" BASIS,
* WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
* See the License for the specific language governing permissions and
* limitations under the License.
*/
package org.apache.cassandra.index.sai.disk.v1.bitpack;
import javax.annotation.concurrent.NotThreadSafe;
import org.apache.cassandra.index.sai.disk.io.SeekingRandomAccessInput;
import org.apache.cassandra.index.sai.disk.v1.LongArray;
import org.apache.lucene.store.IndexInput;
import org.apache.lucene.util.LongValues;
import org.apache.lucene.util.packed.DirectReader;
@NotThreadSafe
public abstract class AbstractBlockPackedReader implements LongArray
{
private final int blockShift;
private final int blockMask;
private final long valueCount;
private final byte[] blockBitsPerValue;
private final SeekingRandomAccessInput input;
private long previousValue = Long.MIN_VALUE;
private long lastIndex; // the last index visited by token -> row ID searches
AbstractBlockPackedReader(IndexInput indexInput, byte[] blockBitsPerValue, int blockShift, int blockMask, long valueCount)
{
this.blockShift = blockShift;
this.blockMask = blockMask;
this.valueCount = valueCount;
this.input = new SeekingRandomAccessInput(indexInput);
this.blockBitsPerValue = blockBitsPerValue;
}
protected abstract long blockOffsetAt(int block);
@Override
public long get(final long valueIndex)
{
if (valueIndex < 0 || valueIndex >= valueCount)
{
throw new IndexOutOfBoundsException(String.format("Index should be between [0, %d), but was %d.", valueCount, valueIndex));
}
int blockIndex = (int) (valueIndex >>> blockShift);
int inBlockIndex = (int) (valueIndex & blockMask);
byte bitsPerValue = blockBitsPerValue[blockIndex];
final LongValues subReader = bitsPerValue == 0 ? LongValues.ZEROES
: DirectReader.getInstance(input, bitsPerValue, blockOffsetAt(blockIndex));
return delta(blockIndex, inBlockIndex) + subReader.get(inBlockIndex);
}
@Override
public long length()
{
return valueCount;
}
@Override
public long indexOf(long value)
{
// If we are searching backwards, we need to reset the lastIndex. This is not normal since we normally move
// forwards when searching for tokens. We only (may) search backwards in vector searchs where we need the
// primary key ranges presented as row IDs.
if (value < previousValue)
lastIndex = 0;
// already out of range
if (lastIndex >= valueCount)
return -1;
previousValue = value;
int blockIndex = binarySearchBlockMinValues(value);
// We need to check next block's min value on an exact match.
boolean exactMatch = blockIndex >= 0;
if (blockIndex < 0)
{
// A non-exact match, which is the negative index of the first value greater than the target.
// For example, searching for 4 against min values [3,3,5,7] produces -2, which we convert to 2.
blockIndex = -blockIndex;
}
if (blockIndex > 0)
{
// Start at the previous block, because there could be duplicate values in the previous block.
// For example, with block 1: [1,2,3,3] & block 2: [3,3,5,7], binary search for 3 would find
// block 2, but we need to start from block 1 and search both.
// In case non-exact match, we need to pivot left as target is less than next block's min.
blockIndex--;
}
// Find the global (not block-specific) index of the target token, which is equivalent to its row ID:
lastIndex = findBlockRowID(value, blockIndex, exactMatch);
return lastIndex >= valueCount ? -1 : lastIndex;
}
/**
*
* @return a positive block index for an exact match, or a negative one for a non-exact match
*/
private int binarySearchBlockMinValues(long targetValue)
{
int high = Math.toIntExact(blockBitsPerValue.length) - 1;
// Assume here that we'll never move backward through the blocks:
int low = Math.toIntExact(lastIndex >> blockShift);
// Short-circuit the search if the target is in current block:
if (low + 1 <= high)
{
long cmp = Long.compare(targetValue, delta(low + 1, 0));
if (cmp == 0)
{
// We have an exact match, so return the index of the next block, which means we'll start
// searching from the current one and also inspect the first value of the next block.
return low + 1;
}
else if (cmp < 0)
{
// We're in the same block. Indicate a non-exact match, and this value will be both
// negated and then decremented to wind up at the current value of "low" here.
return -low - 1;
}
// The target is greater than the next block's min value, so advance to that
// block before starting the usual search...
low++;
}
while (low <= high)
{
int mid = low + ((high - low) >> 1);
long midVal = delta(mid, 0);
if (midVal < targetValue)
{
low = mid + 1;
}
else if (midVal > targetValue)
{
high = mid - 1;
}
else
{
// target found, but we need to check for duplicates
if (mid > 0 && delta(mid - 1, 0) == targetValue)
{
// there are duplicates, pivot left
high = mid - 1;
}
else
{
// no duplicates
return mid;
}
}
}
return -low; // no exact match found
}
private long findBlockRowID(long targetValue, long blockIdx, boolean exactMatch)
{
// Calculate the global offset for the selected block:
long offset = blockIdx << blockShift;
// Resume from previous index if it's larger than offset
long low = Math.max(lastIndex, offset);
// The high is either the last local index in the block, or something smaller if the block isn't full:
long high = Math.min(offset + blockMask + (exactMatch ? 1 : 0), valueCount - 1);
return binarySearchBlock(targetValue, low, high);
}
/**
* binary search target value between low and high.
*
* @return index if exact match is found, or *positive* insertion point if no exact match is found.
*/
private long binarySearchBlock(long target, long low, long high)
{
while (low <= high)
{
long mid = low + ((high - low) >> 1);
long midVal = get(mid);
if (midVal < target)
{
low = mid + 1;
// future rowId cannot be smaller than mid as long as next token not smaller than current token.
lastIndex = mid;
}
else if (midVal > target)
{
high = mid - 1;
}
else
{
// target found, but we need to check for duplicates
if (mid > 0 && get(mid - 1) == target)
{
// there are duplicates, pivot left
high = mid - 1;
}
else
{
// exact match and no duplicates
return mid;
}
}
}
// target not found
return low;
}
abstract long delta(int block, int idx);
}