org.apache.nifi.avro

Class NonCachingDatumReader<T>

java.lang.Object

org.apache.avro.generic.GenericDatumReader<T>

org.apache.nifi.avro.NonCachingDatumReader<T>

All Implemented Interfaces:

org.apache.avro.io.DatumReader<T>

public class NonCachingDatumReader<T>
extends org.apache.avro.generic.GenericDatumReader<T>

Override the GenericDatumReader to provide a much more efficient implementation of #readString. The one that is provided by GenericDatumReader performs very poorly in some cases because it uses an IdentityHashMap with the key being the Schema so that it can stash away the "stringClass" but that performs far worse than just calling JsonNode#getProp. I.e., readString(Object, Schema, Decoder) in GenericDatumReader calls #getStringClass, which uses an IdentityHashMap to cache results in order to avoid calling findStringClass(Schema). However, findStringClass(Schema) is much more efficient than using an IdentityHashMap anyway. Additionally, the performance of findStringClass(Schema)} can be improved slightly and made more readable.

Constructor Summary

Constructors

Constructor and Description
NonCachingDatumReader()
NonCachingDatumReader(org.apache.avro.Schema schema)

Method Summary

Modifier and Type	Method and Description
protected Class	findStringClass(org.apache.avro.Schema schema)
protected Object	readString(Object old, org.apache.avro.Schema expected, org.apache.avro.io.Decoder in)

Methods inherited from class org.apache.avro.generic.GenericDatumReader

addToArray, addToMap, convert, createBytes, createEnum, createFixed, createFixed, createString, getData, getExpected, getResolver, getSchema, newArray, newInstanceFromString, newMap, newRecord, peekArray, read, read, readArray, readBytes, readBytes, readEnum, readField, readFixed, readInt, readMap, readMapKey, readRecord, readString, readWithConversion, readWithoutConversion, setExpected, setSchema, skip

Methods inherited from class java.lang.Object

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

Constructor Detail

NonCachingDatumReader

public NonCachingDatumReader()

NonCachingDatumReader

public NonCachingDatumReader(org.apache.avro.Schema schema)

Method Detail

readString

protected Object readString(Object old,
                            org.apache.avro.Schema expected,
                            org.apache.avro.io.Decoder in)
                     throws IOException

Overrides:

readString in class org.apache.avro.generic.GenericDatumReader<T>

Throws:

IOException

findStringClass

protected Class findStringClass(org.apache.avro.Schema schema)

Overrides:

findStringClass in class org.apache.avro.generic.GenericDatumReader<T>