Package org.apache.pinot.segment.local.segment.index.readers

Class OnHeapStringDictionary

java.lang.Object

org.apache.pinot.segment.local.segment.index.readers.BaseImmutableDictionary

org.apache.pinot.segment.local.segment.index.readers.OnHeapStringDictionary

All Implemented Interfaces:

Closeable, AutoCloseable, Dictionary

public class OnHeapStringDictionary
extends BaseImmutableDictionary

Implementation of String dictionary that cache all values on-heap.

This is useful for String columns that:

• Has low cardinality string dictionary where memory footprint on-heap is acceptably small
• Is heavily queried

This helps avoid creation of String from byte[], which is expensive as well as creates garbage.

Field Summary

Fields inherited from interface org.apache.pinot.segment.spi.index.reader.Dictionary

NULL_VALUE_INDEX

Constructor Summary

Constructors

Constructor	Description
OnHeapStringDictionary(PinotDataBuffer dataBuffer, int length, int numBytesPerValue, byte paddingByte)

Method Summary

Modifier and Type	Method	Description
String	get(int dictId)
BigDecimal	getBigDecimalValue(int dictId)
byte[]	getBytesValue(int dictId)
double	getDoubleValue(int dictId)
float	getFloatValue(int dictId)
int	getIntValue(int dictId)
long	getLongValue(int dictId)
String	getStringValue(int dictId)
FieldSpec.DataType	getValueType()
int	indexOf(String stringValue)
int	insertionIndexOf(String stringValue)	WARNING: With non-zero padding byte, binary search result might not reflect the real insertion index for the value.

Methods inherited from class org.apache.pinot.segment.local.segment.index.readers.BaseImmutableDictionary

binarySearch, binarySearch, binarySearch, binarySearch, binarySearch, binarySearch, binarySearch, close, compare, getBigDecimal, getBuffer, getBytes, getDictIdsInRange, getDouble, getFloat, getInt, getLong, getMaxVal, getMinVal, getPaddedString, getSortedValues, getUnpaddedBytes, getUnpaddedString, isSorted, length, normalizeIndex, padString

Methods inherited from class java.lang.Object

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

Methods inherited from interface org.apache.pinot.segment.spi.index.reader.Dictionary

getByteArrayValue, getInternal, indexOf, indexOf, indexOf, indexOf, indexOf, indexOf, readBigDecimalValues, readBytesValues, readDoubleValues, readFloatValues, readIntValues, readLongValues, readStringValues

Constructor Detail

OnHeapStringDictionary

public OnHeapStringDictionary(PinotDataBuffer dataBuffer,
                              int length,
                              int numBytesPerValue,
                              byte paddingByte)

Method Detail

getValueType

public FieldSpec.DataType getValueType()

indexOf

public int indexOf(String stringValue)

Specified by:

indexOf in interface Dictionary

Overrides:

indexOf in class BaseImmutableDictionary

insertionIndexOf

public int insertionIndexOf(String stringValue)

WARNING: With non-zero padding byte, binary search result might not reflect the real insertion index for the value. E.g. with padding byte 'b', if unpadded value "aa" is in the dictionary, and stored as "aab", then unpadded value "a" will be mis-positioned after value "aa"; unpadded value "aab" will return positive value even if value "aab" is not in the dictionary. TODO: Clean up the segments with legacy non-zero padding byte, and remove the support for non-zero padding byte

get

public String get(int dictId)

getIntValue

public int getIntValue(int dictId)

getLongValue

public long getLongValue(int dictId)

getFloatValue

public float getFloatValue(int dictId)

getDoubleValue

public double getDoubleValue(int dictId)

getBigDecimalValue

public BigDecimal getBigDecimalValue(int dictId)

getStringValue

public String getStringValue(int dictId)

getBytesValue

public byte[] getBytesValue(int dictId)