org.apache.jena.hadoop.rdf.io

Class RdfIOConstants

java.lang.Object

org.apache.jena.hadoop.rdf.io.RdfIOConstants

public class RdfIOConstants
extends Object

RDF IO related constants

Field Summary

Fields

Modifier and Type	Field and Description
static long	DEFAULT_OUTPUT_BATCH_SIZE Default batch size for batched output formats
static String	GLOBAL_BNODE_IDENTITY Configuration key used to control behaviour with regards to how blank nodes are handled.
static String	INPUT_IGNORE_BAD_TUPLES Configuration key used to set whether bad tuples are ignored.
static String	OUTPUT_BATCH_SIZE Configuration key used to set the batch size used for RDF output formats that take a batched writing approach.

Method Summary

Methods inherited from class java.lang.Object

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

Field Detail

INPUT_IGNORE_BAD_TUPLES

public static final String INPUT_IGNORE_BAD_TUPLES

Configuration key used to set whether bad tuples are ignored. This is the default behaviour, when explicitly set to false bad tuples will result in IOException being thrown by the relevant record readers.

See Also:

Constant Field Values

OUTPUT_BATCH_SIZE

public static final String OUTPUT_BATCH_SIZE

Configuration key used to set the batch size used for RDF output formats that take a batched writing approach. Default value is given by the constant DEFAULT_OUTPUT_BATCH_SIZE.

See Also:

Constant Field Values

DEFAULT_OUTPUT_BATCH_SIZE

public static final long DEFAULT_OUTPUT_BATCH_SIZE

Default batch size for batched output formats

See Also:

Constant Field Values

GLOBAL_BNODE_IDENTITY

public static final String GLOBAL_BNODE_IDENTITY

Configuration key used to control behaviour with regards to how blank nodes are handled.

The default behaviour is that blank nodes are file scoped which is what the RDF specifications require.

However in the case of a multi-stage pipeline this behaviour can cause blank nodes to diverge over several jobs and introduce spurious blank nodes over time. This is described in JENA-820 and enabling this flag for jobs in your pipeline allow you to work around this problem.

Warning

You should only enable this flag for jobs that take in RDF output originating from previous jobs since our normal blank node allocation policy ensures that blank nodes will be file scoped and unique over all files (barring unfortunate hasing collisions). If you enable this for jobs that take in RDF originating from other sources you may incorrectly conflate blank nodes that are supposed to distinct and separate nodes.

See Also:

Constant Field Values