Package org.apache.beam.sdk.transforms

Class Create<T>

java.lang.Object

org.apache.beam.sdk.transforms.Create<T>

Type Parameters:

T - the type of the elements of the resulting PCollection

public class Create<T>
extends Object

Create<T> takes a collection of elements of type T known when the pipeline is constructed and returns a PCollection<T> containing the elements.

Example of use:

 Pipeline p = ...;

 PCollection<Integer> pc = p.apply(Create.of(3, 4, 5).withCoder(BigEndianIntegerCoder.of()));

 Map<String, Integer> map = ...;
 PCollection<KV<String, Integer>> pt =
     p.apply(Create.of(map)
      .withCoder(KvCoder.of(StringUtf8Coder.of(),
                            BigEndianIntegerCoder.of())));
 

Create can automatically determine the Coder to use if all elements have the same run-time class, and a default coder is registered for that class. See CoderRegistry for details on how defaults are determined.

If a coder can not be inferred, Create.Values.withCoder(org.apache.beam.sdk.coders.Coder<T>) must be called explicitly to set the encoding of the resulting PCollection.

A good use for Create is when a PCollection needs to be created without dependencies on files or other external entities. This is especially useful during testing.

Caveat: Create only supports small in-memory datasets.

Nested Class Summary

Nested Classes

Modifier and Type	Class	Description
static class	Create.OfValueProvider<T>	Implementation of ofProvider(org.apache.beam.sdk.options.ValueProvider<T>, org.apache.beam.sdk.coders.Coder<T>).
static class	Create.TimestampedValues<T>	A PTransform that creates a PCollection whose elements have associated timestamps.
static class	Create.Values<T>	A PTransform that creates a PCollection from a set of in-memory objects.
static class	Create.WindowedValues<T>	A PTransform that creates a PCollection whose elements have associated windowing metadata.

Constructor Summary

Constructors

Constructor	Description
Create()

Method Summary

Modifier and Type	Method	Description
static <T> Create.Values<T>	empty(Coder<T> coder)	Returns a new Create.Values transform that produces an empty PCollection.
static Create.Values<Row>	empty(Schema schema)	Returns a new Create.Values transform that produces an empty PCollection of rows.
static <T> Create.Values<T>	empty(TypeDescriptor<T> type)	Returns a new Create.Values transform that produces an empty PCollection.
static <T> Create.Values<T>	of(@Nullable T elem, @Nullable T... elems)	Returns a new Create.Values transform that produces a PCollection containing the specified elements.
static <T> Create.Values<T>	of(Iterable<T> elems)	Returns a new Create.Values transform that produces a PCollection containing elements of the provided Iterable.
static <K, V> Create.Values<KV<K,V>>	of(Map<K,V> elems)	Returns a new Create.Values transform that produces a PCollection of KVs corresponding to the keys and values of the specified Map.
static <T> Create.OfValueProvider<T>	ofProvider(ValueProvider<T> provider, Coder<T> coder)	Returns an Create.OfValueProvider transform that produces a PCollection of a single element provided by the given ValueProvider.
static <T> Create.TimestampedValues<T>	timestamped(Iterable<TimestampedValue<T>> elems)	Returns a new Create.TimestampedValues transform that produces a PCollection containing the elements of the provided Iterable with the specified timestamps.
static <T> Create.TimestampedValues<T>	timestamped(Iterable<T> values, Iterable<Long> timestamps)	Returns a new root transform that produces a PCollection containing the specified elements with the specified timestamps.
static <T> Create.TimestampedValues<T>	timestamped(TimestampedValue<T> elem, TimestampedValue<T>... elems)	Returns a new Create.TimestampedValues transform that produces a PCollection containing the specified elements with the specified timestamps.
static <T> Create.WindowedValues<T>	windowedValues(Iterable<WindowedValue<T>> elems)	Returns a new Create.WindowedValues transform that produces a PCollection containing the elements of the provided Iterable with the specified windowing metadata.
static <T> Create.WindowedValues<T>	windowedValues(WindowedValue<T> elem, WindowedValue<T>... elems)	Returns a new Create.WindowedValues transform that produces a PCollection containing the specified elements with the specified windowing metadata.

Methods inherited from class java.lang.Object

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

Constructor Details

Create

public Create()

Method Details

of

public static <T> Create.Values<T> of(Iterable<T> elems)

Returns a new Create.Values transform that produces a PCollection containing elements of the provided Iterable.

The argument should not be modified after this is called.

The elements of the output PCollection will have a timestamp of negative infinity, see timestamped(java.lang.Iterable<org.apache.beam.sdk.values.TimestampedValue<T>>) for a way of creating a PCollection with timestamped elements.

By default, Create.Values can automatically determine the Coder to use if all elements have the same non-parameterized run-time class, and a default coder is registered for that class. See CoderRegistry for details on how defaults are determined. Otherwise, use Create.Values.withCoder(org.apache.beam.sdk.coders.Coder<T>) to set the coder explicitly.

of

@SafeVarargs
public static <T> Create.Values<T> of(@Nullable T elem,
 @Nullable T... elems)

Returns a new Create.Values transform that produces a PCollection containing the specified elements.

The elements will have a timestamp of negative infinity, see timestamped(java.lang.Iterable<org.apache.beam.sdk.values.TimestampedValue<T>>) for a way of creating a PCollection with timestamped elements.

The arguments should not be modified after this is called.

By default, Create.Values can automatically determine the Coder to use if all elements have the same non-parameterized run-time class, and a default coder is registered for that class. See CoderRegistry for details on how defaults are determined. Otherwise, use Create.Values.withCoder(org.apache.beam.sdk.coders.Coder<T>) to set the coder explicitly.

empty

public static Create.Values<Row> empty(Schema schema)

Returns a new Create.Values transform that produces an empty PCollection of rows.

empty

public static <T> Create.Values<T> empty(Coder<T> coder)

Returns a new Create.Values transform that produces an empty PCollection.

The elements will have a timestamp of negative infinity, see timestamped(java.lang.Iterable<org.apache.beam.sdk.values.TimestampedValue<T>>) for a way of creating a PCollection with timestamped elements.

Since there are no elements, the Coder cannot be automatically determined. Instead, the Coder is provided via the coder argument.

empty

public static <T> Create.Values<T> empty(TypeDescriptor<T> type)

Returns a new Create.Values transform that produces an empty PCollection.

The elements will have a timestamp of negative infinity, see timestamped(java.lang.Iterable<org.apache.beam.sdk.values.TimestampedValue<T>>) for a way of creating a PCollection with timestamped elements.

Since there are no elements, the Coder cannot be automatically determined. Instead, the Coder is determined from given TypeDescriptor<T>. Note that a default coder must be registered for the class described in the TypeDescriptor<T>.

of

public static <K,
V> Create.Values<KV<K,V>> of(Map<K,V> elems)

Returns a new Create.Values transform that produces a PCollection of KVs corresponding to the keys and values of the specified Map.

The elements will have a timestamp of negative infinity, see timestamped(java.lang.Iterable<org.apache.beam.sdk.values.TimestampedValue<T>>) for a way of creating a PCollection with timestamped elements.

By default, Create.Values can automatically determine the Coder to use if all elements have the same non-parameterized run-time class, and a default coder is registered for that class. See CoderRegistry for details on how defaults are determined. Otherwise, use Create.Values.withCoder(org.apache.beam.sdk.coders.Coder<T>) to set the coder explicitly.

ofProvider

public static <T> Create.OfValueProvider<T> ofProvider(ValueProvider<T> provider,
 Coder<T> coder)

Returns an Create.OfValueProvider transform that produces a PCollection of a single element provided by the given ValueProvider.

timestamped

public static <T> Create.TimestampedValues<T> timestamped(Iterable<TimestampedValue<T>> elems)

Returns a new Create.TimestampedValues transform that produces a PCollection containing the elements of the provided Iterable with the specified timestamps.

The argument should not be modified after this is called.

By default, Create.TimestampedValues can automatically determine the Coder to use if all elements have the same non-parameterized run-time class, and a default coder is registered for that class. See CoderRegistry for details on how defaults are determined. Otherwise, use Create.TimestampedValues.withCoder(org.apache.beam.sdk.coders.Coder<T>) to set the coder explicitly.

timestamped

@SafeVarargs
public static <T> Create.TimestampedValues<T> timestamped(TimestampedValue<T> elem,
 TimestampedValue<T>... elems)

Returns a new Create.TimestampedValues transform that produces a PCollection containing the specified elements with the specified timestamps.

The arguments should not be modified after this is called.

windowedValues

public static <T> Create.WindowedValues<T> windowedValues(Iterable<WindowedValue<T>> elems)

Returns a new Create.WindowedValues transform that produces a PCollection containing the elements of the provided Iterable with the specified windowing metadata.

The argument should not be modified after this is called.

By default, Create.WindowedValues can automatically determine the Coder to use if all elements have the same non-parameterized run-time class, and a default coder is registered for that class. See CoderRegistry for details on how defaults are determined. Otherwise, use Create.WindowedValues.withCoder(org.apache.beam.sdk.coders.Coder<T>) to set the coder explicitly.

Likewise, the window coder can be inferred if the window type is registered with the CoderRegistry. Otherwise, use Create.WindowedValues.withWindowCoder(Coder) to set the window coder explicitly.

windowedValues

@SafeVarargs
public static <T> Create.WindowedValues<T> windowedValues(WindowedValue<T> elem,
 WindowedValue<T>... elems)

Returns a new Create.WindowedValues transform that produces a PCollection containing the specified elements with the specified windowing metadata.

The arguments should not be modified after this is called.

timestamped

public static <T> Create.TimestampedValues<T> timestamped(Iterable<T> values,
 Iterable<Long> timestamps)

Returns a new root transform that produces a PCollection containing the specified elements with the specified timestamps.

The arguments should not be modified after this is called.

By default, Create.TimestampedValues can automatically determine the Coder to use if all elements have the same non-parameterized run-time class, and a default coder is registered for that class. See CoderRegistry for details on how defaults are determined. Otherwise, use Create.TimestampedValues.withCoder(org.apache.beam.sdk.coders.Coder<T>) to set the coder explicitly.

Throws:

IllegalArgumentException - if there are a different number of values and timestamps