io.cdap.cdap.etl.api.relational

Interface RelationalTransform

All Known Subinterfaces:

LinearRelationalTransform

public interface RelationalTransform

An analytical or transformation plugin can implement this interface to provide an alternative declarative way of doing transformations. If relational transform call is successfull, the transform will be passed to the underlying engine and plugin will not be called to transform each row. If relational transform is not successful (no engines are available, engines don't have capabilities or can't perform the requested transformation), fall back to per-row transformation will be performed.

Method Summary

Modifier and Type	Method and Description
default boolean	canUseEngine(Engine engine) This call allows to prefilter engines before doing transform(io.cdap.cdap.etl.api.relational.RelationalTranformContext).
default boolean	requireUnionInputs() This will union multiple input data together if required
boolean	transform(RelationalTranformContext context) This call will be done for every suitable engine until after supported tranformation is declared or there is no more engines left.

Method Detail

canUseEngine

default boolean canUseEngine(Engine engine)

This call allows to prefilter engines before doing transform(io.cdap.cdap.etl.api.relational.RelationalTranformContext).

Parameters:

engine - engine to check

Returns:

if this engine can be used for transform requested

transform

boolean transform(RelationalTranformContext context)

This call will be done for every suitable engine until after supported tranformation is declared or there is no more engines left.

A transformation is supported if plugin returned true, all registered output relations are valid and engine can perform the transformation requested.

If no engine can perform a transformation, a fall back to regular by-row transformation is performed.

Parameters:

context - tranformation context with engine, input and output parameters

Returns:

if plugin could perform a transformation. If true is returned, plugin must set all outputs to relations. Plugin may still set any output to invalid relation and return true. In this case transformation would be considered unsupported. Even if plugin returned true and set all outputs to valid relations, engine may deny performing a transformation.

See Also:

LinearRelationalTransform} for simpler interface for plugins that take 1 input and produce 1 output.

requireUnionInputs

default boolean requireUnionInputs()

This will union multiple input data together if required

Returns:

false