public class PCollectionTuple extends Object implements PInput, POutput
PCollectionTuple is an immutable tuple of
heterogeneously-typed PCollections, "keyed" by
TupleTags. A PCollectionTuple can be used as the input or
output of a
PTransform taking
or producing multiple PCollection inputs or outputs that can be of
different types, for instance a
ParDo with side
outputs.
A PCollectionTuple can be created and accessed like follows:
PCollection<String> pc1 = ...;
PCollection<Integer> pc2 = ...;
PCollection<Iterable<String>> pc3 = ...;
// Create TupleTags for each of the PCollections to put in the
// PCollectionTuple (the type of the TupleTag enables tracking the
// static type of each of the PCollections in the PCollectionTuple):
TupleTag<String> tag1 = new TupleTag<>();
TupleTag<Integer> tag2 = new TupleTag<>();
TupleTag<Iterable<String>> tag3 = new TupleTag<>();
// Create a PCollectionTuple with three PCollections:
PCollectionTuple pcs =
PCollectionTuple.of(tag1, pc1)
.and(tag2, pc2)
.and(tag3, pc3);
// Create an empty PCollectionTuple:
Pipeline p = ...;
PCollectionTuple pcs2 = PCollectionTuple.empty(p);
// Get PCollections out of a PCollectionTuple, using the same tags
// that were used to put them in:
PCollection<Integer> pcX = pcs.get(tag2);
PCollection<String> pcY = pcs.get(tag1);
PCollection<Iterable<String>> pcZ = pcs.get(tag3);
// Get a map of all PCollections in a PCollectionTuple:
Map<TupleTag<?>, PCollection<?>> allPcs = pcs.getAll();
| Modifier and Type | Method and Description |
|---|---|
<T> PCollectionTuple |
and(TupleTag<T> tag,
PCollection<T> pc)
Returns a new
PCollectionTuple that has each PCollection and
TupleTag of this PCollectionTuple plus the given PCollection
associated with the given TupleTag. |
<OutputT extends POutput> |
apply(PTransform<PCollectionTuple,OutputT> t)
Like
apply(String, PTransform) but defaulting to the name
of the PTransform. |
<OutputT extends POutput> |
apply(String name,
PTransform<PCollectionTuple,OutputT> t)
Applies the given
PTransform to this input PCollectionTuple,
using name to identify this specific application of the transform. |
static PCollectionTuple |
empty(Pipeline pipeline)
Returns an empty
PCollectionTuple that is part of the given Pipeline. |
Collection<? extends PValue> |
expand()
|
void |
finishSpecifying()
After building, finalizes this
PInput to make it ready for
being used as an input to a PTransform. |
void |
finishSpecifyingOutput()
As part of applying the producing
PTransform, finalizes this
output to make it ready for being used as an input and for running. |
<T> PCollection<T> |
get(TupleTag<T> tag)
|
Map<TupleTag<?>,PCollection<?>> |
getAll()
Returns an immutable Map from
TupleTag to corresponding
PCollection, for all the members of this PCollectionTuple. |
Pipeline |
getPipeline()
|
<T> boolean |
has(TupleTag<T> tag)
Returns whether this
PCollectionTuple contains a PCollection with
the given tag. |
static <T> PCollectionTuple |
of(TupleTag<T> tag,
PCollection<T> pc)
|
static PCollectionTuple |
ofPrimitiveOutputsInternal(Pipeline pipeline,
TupleTagList outputTags,
org.apache.beam.sdk.util.WindowingStrategy<?,?> windowingStrategy,
PCollection.IsBounded isBounded)
Returns a
PCollectionTuple with each of the given tags mapping to a new
output PCollection. |
void |
recordAsOutput(AppliedPTransform<?,?,?> transform)
Records that this
POutput is an output of the given
PTransform. |
public static PCollectionTuple empty(Pipeline pipeline)
PCollectionTuple that is part of the given Pipeline.
A PCollectionTuple containing additional elements can be created by calling
and(org.apache.beam.sdk.values.TupleTag<T>, org.apache.beam.sdk.values.PCollection<T>) on the result.
public static <T> PCollectionTuple of(TupleTag<T> tag, PCollection<T> pc)
PCollectionTuple containing the given
PCollection keyed by the given TupleTag.
A PCollectionTuple containing additional elements can be created by calling
and(org.apache.beam.sdk.values.TupleTag<T>, org.apache.beam.sdk.values.PCollection<T>) on the result.
public <T> PCollectionTuple and(TupleTag<T> tag, PCollection<T> pc)
PCollectionTuple that has each PCollection and
TupleTag of this PCollectionTuple plus the given PCollection
associated with the given TupleTag.
The given TupleTag should not already be mapped to a
PCollection in this PCollectionTuple.
Each PCollection in the resulting PCollectionTuple must be
part of the same Pipeline.
public <T> boolean has(TupleTag<T> tag)
PCollectionTuple contains a PCollection with
the given tag.public <T> PCollection<T> get(TupleTag<T> tag)
PCollection associated with the given TupleTag
in this PCollectionTuple. Throws IllegalArgumentException if there is no
such PCollection, i.e., !has(tag).public Map<TupleTag<?>,PCollection<?>> getAll()
TupleTag to corresponding
PCollection, for all the members of this PCollectionTuple.public <OutputT extends POutput> OutputT apply(PTransform<PCollectionTuple,OutputT> t)
apply(String, PTransform) but defaulting to the name
of the PTransform.PTransformpublic <OutputT extends POutput> OutputT apply(String name, PTransform<PCollectionTuple,OutputT> t)
PTransform to this input PCollectionTuple,
using name to identify this specific application of the transform.
This name is used in various places, including the monitoring UI, logging,
and to stably identify this application node in the job graph.PTransformpublic static PCollectionTuple ofPrimitiveOutputsInternal(Pipeline pipeline, TupleTagList outputTags, org.apache.beam.sdk.util.WindowingStrategy<?,?> windowingStrategy, PCollection.IsBounded isBounded)
PCollectionTuple with each of the given tags mapping to a new
output PCollection.
For use by primitive transformations only.
public Pipeline getPipeline()
PInputgetPipeline in interface PInputgetPipeline in interface POutputpublic Collection<? extends PValue> expand()
PInputPInput into a list of its component output
PValues.
PValue expands to itself.PValues (such as
PCollectionTuple or PCollectionList)
expands to its component PValue PValues.Not intended to be invoked directly by user code.
public void recordAsOutput(AppliedPTransform<?,?,?> transform)
POutputPOutput is an output of the given
PTransform.
For a compound POutput, it is advised to call
this method on each component POutput.
This is not intended to be invoked by user code, but
is automatically invoked as part of applying the
producing PTransform.
recordAsOutput in interface POutputpublic void finishSpecifying()
PInputAfter building, finalizes this PInput to make it ready for
being used as an input to a PTransform.
Automatically invoked whenever apply() is invoked on
this PInput, so users do not normally call this explicitly.
finishSpecifying in interface PInputpublic void finishSpecifyingOutput()
POutputPTransform, finalizes this
output to make it ready for being used as an input and for running.
This includes ensuring that all PCollections
have Coders specified or defaulted.
Automatically invoked whenever this POutput is used
as a PInput to another PTransform, or if never
used as a PInput, when Pipeline.run()
is called, so users do not normally call this explicitly.
finishSpecifyingOutput in interface POutput