Module org.apache.arrow.adapter.avro

Package org.apache.arrow.adapter.avro

Class ArrowToAvroUtils

java.lang.Object

org.apache.arrow.adapter.avro.ArrowToAvroUtils

public class ArrowToAvroUtils
extends Object

Field Summary

Fields

Modifier and Type	Field	Description
static final String	GENERIC_RECORD_TYPE_NAME

Constructor Summary

Constructors

Constructor	Description
ArrowToAvroUtils()

Method Summary

Modifier and Type	Method	Description
static org.apache.avro.Schema	createAvroSchema(List<Field> arrowFields)	Overload provided for convenience, sets name = GENERIC_RECORD_TYPE_NAME.
static org.apache.avro.Schema	createAvroSchema(List<Field> arrowFields, String typeName)	Overload provided for convenience, sets namespace = null.
static org.apache.avro.Schema	createAvroSchema(List<Field> arrowFields, String typeName, String namespace)	Create an Avro record schema for a given list of Arrow fields.
static CompositeAvroProducer	createCompositeProducer(List<FieldVector> vectors)	Create a composite Avro producer for a set of field vectors (typically the root set of a VSR).

Methods inherited from class java.lang.Object

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

Field Details

GENERIC_RECORD_TYPE_NAME

public static final String GENERIC_RECORD_TYPE_NAME

See Also:

• Constant Field Values

Constructor Details

ArrowToAvroUtils

public ArrowToAvroUtils()

Method Details

createAvroSchema

public static org.apache.avro.Schema createAvroSchema(List<Field> arrowFields,
 String typeName,
 String namespace)

Create an Avro record schema for a given list of Arrow fields.

This method currently performs following type mapping for Avro data types to corresponding Arrow data types.

Arrow type	Avro encoding
ArrowType.Null	NULL
ArrowType.Bool	BOOLEAN
ArrowType.Int(64 bit, unsigned 32 bit)	LONG
ArrowType.Int(signed 32 bit, < 32 bit)	INT
ArrowType.FloatingPoint(double)	DOUBLE
ArrowType.FloatingPoint(single, half)	FLOAT
ArrowType.Utf8	STRING
ArrowType.LargeUtf8	STRING
ArrowType.Binary	BYTES
ArrowType.LargeBinary	BYTES
ArrowType.FixedSizeBinary	FIXED
ArrowType.Decimal	decimal (FIXED)
ArrowType.Date	date (INT)
ArrowType.Time (SEC | MILLI)	time-millis (INT)
ArrowType.Time (MICRO | NANO)	time-micros (LONG)
ArrowType.Timestamp (NANOSECONDS, TZ != NULL)	time-nanos (LONG)
ArrowType.Timestamp (MICROSECONDS, TZ != NULL)	time-micros (LONG)
ArrowType.Timestamp (MILLISECONDS | SECONDS, TZ != NULL)	time-millis (LONG)
ArrowType.Timestamp (NANOSECONDS, TZ == NULL)	local-time-nanos (LONG)
ArrowType.Timestamp (MICROSECONDS, TZ == NULL)	local-time-micros (LONG)
ArrowType.Timestamp (MILLISECONDS | SECONDS, TZ == NULL)	local-time-millis (LONG)
ArrowType.Duration	duration (FIXED)
ArrowType.Interval	duration (FIXED)
ArrowType.Struct	record
ArrowType.List	array
ArrowType.LargeList	array
ArrowType.FixedSizeList	array
ArrowType.Map	map
ArrowType.Union	union

Nullable fields are represented as a union of [base-type | null]. Special treatment is given to nullability of unions - a union is considered nullable if any of its child fields are nullable. The schema for a nullable union will always contain a null type as its first member, with none of the child types being nullable.

List fields must contain precisely one child field, which may be nullable. Map fields are represented as a list of structs, where the struct fields are "key" and "value". The key field must always be of type STRING (Utf8) and cannot be nullable. The value can be of any type and may be nullable. Record types must contain at least one child field and cannot contain multiple fields with the same name

Parameters:

arrowFields - The arrow fields used to generate the Avro schema

typeName - Name of the top level Avro record type

namespace - Namespace of the top level Avro record type

Returns:

An Avro record schema for the given list of fields, with the specified name and namespace

createAvroSchema

public static org.apache.avro.Schema createAvroSchema(List<Field> arrowFields,
 String typeName)

Overload provided for convenience, sets namespace = null.

createAvroSchema

public static org.apache.avro.Schema createAvroSchema(List<Field> arrowFields)

Overload provided for convenience, sets name = GENERIC_RECORD_TYPE_NAME.

createCompositeProducer

public static CompositeAvroProducer createCompositeProducer(List<FieldVector> vectors)

Create a composite Avro producer for a set of field vectors (typically the root set of a VSR).

Parameters:

vectors - The vectors that will be used to produce Avro data

Returns:

The resulting composite Avro producer