cascading.operation
Interface Operation<Context>

All Superinterfaces:
DeclaresResults
All Known Subinterfaces:
Aggregator<Context>, Assertion<PC>, Buffer<Context>, Filter<Context>, Function<Context>, GroupAssertion<C>, PlannedOperation<Context>, ValueAssertion<C>
All Known Implementing Classes:
AggregateBy.CompositeFunction, And, AssertEquals, AssertEqualsAll, AssertExpression, AssertGroupBase, AssertGroupSizeEquals, AssertGroupSizeLessThan, AssertGroupSizeMoreThan, AssertMatches, AssertMatchesAll, AssertNotEquals, AssertNotNull, AssertNull, AssertSizeEquals, AssertSizeLessThan, AssertSizeMoreThan, Average, AverageBy.AverageFinal, BaseAssertion, BaseOperation, Count, Counter, DateFormatter, DateOperation, DateParser, Debug, ExpressionFilter, ExpressionFunction, ExpressionOperation, ExtentBase, ExtremaBase, ExtremaValueBase, FieldFormatter, FieldJoiner, FilterNotNull, FilterNull, First, FirstNBuffer, Identity, Insert, Last, Limit, Logic, Max, MaxValue, Min, MinValue, NoOp, Not, Or, RegexFilter, RegexGenerator, RegexMatcher, RegexOperation, RegexParser, RegexReplace, RegexSplitGenerator, RegexSplitter, Sample, ScriptFilter, ScriptFunction, ScriptOperation, ScriptTupleFunction, SetValue, Status, Sum, UnGroup, Unique.FilterPartialDuplicates, Xor

public interface Operation<Context>
extends DeclaresResults

Interface Operation is the base interface for all functions applied to Tuple streams.

Specifically Function, Filter, Aggregator, Buffer, and Assertion.

Use BaseOperation for a convenient way to create new Operation types.

See Also:
BaseOperation, Function, Filter, Aggregator, Buffer, Assertion

Field Summary
static int ANY
          Field ANY denotes that a given Operation will take any number of argument values
 
Method Summary
 void cleanup(FlowProcess flowProcess, OperationCall<Context> operationCall)
          The cleanup method is called immediately after the current Operation instance is taken out of play processing Tuples.
 void flush(FlowProcess flowProcess, OperationCall<Context> operationCall)
          The flush method is called when an Operation that is caching values must empty the cache.
 Fields getFieldDeclaration()
          Returns the fields created by this Operation instance.
 int getNumArgs()
          The minimum number of arguments this Operation expects from the calling Each or Every Operator.
 boolean isSafe()
          Returns true if this Operation instance can safely execute on the same 'record' multiple times, false otherwise.
 void prepare(FlowProcess flowProcess, OperationCall<Context> operationCall)
          The prepare method is called immediately before the current Operation instance is put into play processing Tuples.
 

Field Detail

ANY

static final int ANY
Field ANY denotes that a given Operation will take any number of argument values

See Also:
Constant Field Values
Method Detail

prepare

void prepare(FlowProcess flowProcess,
             OperationCall<Context> operationCall)
The prepare method is called immediately before the current Operation instance is put into play processing Tuples. This method should initialize any resources that can be shutdown or released in the cleanup(cascading.flow.FlowProcess, OperationCall) method.

Any resources created should be stored in the Context, not as instance fields on the class.

This method may be called more than once during the life of this instance. But it will never be called multiple times without a cleanup invocation immediately before subsequent invocations.

If the Flow this Operation instance belongs will execute on a remote cluster, this method will be called cluster side, not client side.

Parameters:
flowProcess -
operationCall -

flush

void flush(FlowProcess flowProcess,
           OperationCall<Context> operationCall)
The flush method is called when an Operation that is caching values must empty the cache. It is called before cleanup(cascading.flow.FlowProcess, OperationCall) is invoked.

It is safe to cast the OperationCall to a FunctionCall, or equivalent, and get its FunctionCall.getOutputCollector().

Parameters:
flowProcess -
operationCall -

cleanup

void cleanup(FlowProcess flowProcess,
             OperationCall<Context> operationCall)
The cleanup method is called immediately after the current Operation instance is taken out of play processing Tuples. This method should shutdown any resources created or initialized during the prepare(cascading.flow.FlowProcess, OperationCall) method.

This method may be called more than once during the life of this instance. But it will never be called multiple times without a prepare invocation before.

If the Flow this Operation instance belongs will execute on a remote cluster, this method will be called cluster side, not client side.

Parameters:
flowProcess -
operationCall -

getFieldDeclaration

Fields getFieldDeclaration()
Returns the fields created by this Operation instance. If this instance is a Filter, it should always return Fields.ALL.

Specified by:
getFieldDeclaration in interface DeclaresResults
Returns:
a Fields instance

getNumArgs

int getNumArgs()
The minimum number of arguments this Operation expects from the calling Each or Every Operator.

Operations should be willing to receive more arguments than expected, but should ignore them if they are unused, instead of failing.

Returns:
an int

isSafe

boolean isSafe()
Returns true if this Operation instance can safely execute on the same 'record' multiple times, false otherwise.

That is, this Operation is safe if it has no side-effects, or if it does, they are idempotent.

If seeing the same 'record' more than once can cause errors (internally or externally), this method must return false.

Returns:
a boolean


Copyright © 2007-2015 Concurrent, Inc. All Rights Reserved.