org.apache.hadoop.hive.ql.udf.ptf

Class TableFunctionResolver

java.lang.Object

org.apache.hadoop.hive.ql.udf.ptf.TableFunctionResolver

Direct Known Subclasses:

MatchPath.MatchPathResolver, Noop.NoopResolver, NoopWithMap.NoopWithMapResolver, WindowingTableFunction.WindowingTableFunctionResolver

public abstract class TableFunctionResolver
extends Object

Based on Hive GenericUDAFResolver. Break up the responsibility of the old AbstractTableFunction class into a Resolver and Evaluator. The Resolver is responsible for:

1. setting up the tableFunctionEvaluator
2. Setting up the The raw and output ObjectInspectors of the Evaluator.
3. The Evaluator also holds onto the TableFunctionDef. This provides information about the arguments to the function, the shape of the Input partition and the Partitioning details.

The Resolver for a function is obtained from the FunctionRegistry. The Resolver is initialized by the following 4 step process:

1. The initialize method is called; which is passed the PTFDesc and the TableFunctionDef.
2. The resolver is then asked to setup the Raw ObjectInspector. This is only required if the Function reshapes the raw input.
3. Once the Resolver has had a chance to compute the shape of the Raw Input that is fed to the partitioning machinery; the translator sets up the partitioning details on the tableFuncDef.
4. finally the resolver is asked to setup the output ObjectInspector.

Constructor Summary

Constructors

Constructor and Description
TableFunctionResolver()

Method Summary

Methods

Modifier and Type	Method and Description
boolean	carryForwardNames()
protected abstract TableFunctionEvaluator	createEvaluator(PTFDesc ptfDesc, PartitionedTableFunctionDef tDef)
TableFunctionEvaluator	getEvaluator()
abstract List<String>	getOutputColumnNames()
PTFDesc	getPtfDesc()
List<String>	getRawInputColumnNames()
List<String>	getReferencedColumns() Provide referenced columns names to be used in partition function
void	initialize(HiveConf cfg, PTFDesc ptfDesc, PartitionedTableFunctionDef tDef)
void	initialize(PTFDesc ptfDesc, PartitionedTableFunctionDef tDef, TableFunctionEvaluator evaluator)
abstract void	initializeOutputOI() This method is invoked during runtime(during deserialization of theQueryDef).
void	initializeRawInputOI()
protected void	setOutputOI(StructObjectInspector outputOI)
protected void	setRawInputOI(StructObjectInspector rawInputOI)
abstract void	setupOutputOI()
void	setupRawInputOI()
abstract boolean	transformsRawInput()

Methods inherited from class java.lang.Object

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

Constructor Detail

TableFunctionResolver

public TableFunctionResolver()

Method Detail

initialize

public void initialize(HiveConf cfg,
              PTFDesc ptfDesc,
              PartitionedTableFunctionDef tDef)
                throws SemanticException

Throws:

SemanticException

initialize

public void initialize(PTFDesc ptfDesc,
              PartitionedTableFunctionDef tDef,
              TableFunctionEvaluator evaluator)
                throws HiveException

Throws:

HiveException

getEvaluator

public TableFunctionEvaluator getEvaluator()

setupOutputOI

public abstract void setupOutputOI()
                            throws SemanticException

Throws:

SemanticException

getOutputColumnNames

public abstract List<String> getOutputColumnNames()
                                           throws SemanticException

Throws:

SemanticException

initializeOutputOI

public abstract void initializeOutputOI()
                                 throws HiveException

This method is invoked during runtime(during deserialization of theQueryDef). At this point the TableFunction can assume that the Expression Nodes exist for all the Def (ArgDef, ColumnDef, WindowDef..). It is the responsibility of the TableFunction to construct the evaluators and setup the OI.

Parameters:

tblFuncDef -

ptfDesc -

Throws:

HiveException

setupRawInputOI

public void setupRawInputOI()
                     throws SemanticException

Throws:

SemanticException

getRawInputColumnNames

public List<String> getRawInputColumnNames()
                                    throws SemanticException

Throws:

SemanticException

initializeRawInputOI

public void initializeRawInputOI()
                          throws HiveException

Throws:

HiveException

setRawInputOI

protected void setRawInputOI(StructObjectInspector rawInputOI)

setOutputOI

protected void setOutputOI(StructObjectInspector outputOI)

getPtfDesc

public PTFDesc getPtfDesc()

carryForwardNames

public boolean carryForwardNames()

transformsRawInput

public abstract boolean transformsRawInput()

createEvaluator

protected abstract TableFunctionEvaluator createEvaluator(PTFDesc ptfDesc,
                                     PartitionedTableFunctionDef tDef)

getReferencedColumns

public List<String> getReferencedColumns()
                                  throws SemanticException

Provide referenced columns names to be used in partition function

Returns:

null for unknown (will get all columns from table including virtual columns)

Throws:

SemanticException