org.apache.hadoop.hive.ql.optimizer

Class GenMRProcContext

java.lang.Object

org.apache.hadoop.hive.ql.optimizer.GenMRProcContext

All Implemented Interfaces:

NodeProcessorCtx

public class GenMRProcContext
extends Object
implements NodeProcessorCtx

Processor Context for creating map reduce task. Walk the tree in a DFS manner and process the nodes. Some state is maintained about the current nodes visited so far.

Nested Class Summary

Nested Classes

Modifier and Type	Class and Description
static class	GenMRProcContext.GenMapRedCtx GenMapRedCtx is used to keep track of the current state.
static class	GenMRProcContext.GenMRUnionCtx GenMRUnionCtx.

Constructor Summary

Constructors

Constructor and Description
GenMRProcContext()
GenMRProcContext(HiveConf conf, HashMap<Operator<? extends OperatorDesc>,Task<? extends Serializable>> opTaskMap, ParseContext parseCtx, List<Task<MoveWork>> mvTask, List<Task<? extends Serializable>> rootTasks, LinkedHashMap<Operator<? extends OperatorDesc>,GenMRProcContext.GenMapRedCtx> mapCurrCtx, Set<ReadEntity> inputs, Set<WriteEntity> outputs)

Method Summary

Methods

Modifier and Type	Method and Description
boolean	addRootIfPossible(Task<? extends Serializable> task)
void	addSeenOp(Task task, Operator operator)
HiveConf	getConf()
String	getCurrAliasId()
Task<? extends Serializable>	getCurrTask()
Operator<? extends OperatorDesc>	getCurrTopOp()
UnionOperator	getCurrUnionOp()
DependencyCollectionTask	getDependencyTaskForMultiInsert() Returns dependencyTaskForMultiInsert initializing it if necessary.
Set<ReadEntity>	getInputs() Get the input set.
Map<FileSinkDesc,Task<? extends Serializable>>	getLinkedFileDescTasks()
LinkedHashMap<Operator<? extends OperatorDesc>,GenMRProcContext.GenMapRedCtx>	getMapCurrCtx()
List<Task<MoveWork>>	getMvTask()
HashMap<Operator<? extends OperatorDesc>,Task<? extends Serializable>>	getOpTaskMap()
Set<WriteEntity>	getOutputs() Get the output set.
ParseContext	getParseCtx()
List<Task<? extends Serializable>>	getRootTasks()
List<FileSinkOperator>	getSeenFileSinkOps()
GenMRProcContext.GenMRUnionCtx	getUnionTask(UnionOperator op)
boolean	isSeenOp(Task task, Operator operator)
void	setConf(HiveConf conf)
void	setCurrAliasId(String currAliasId)
void	setCurrTask(Task<? extends Serializable> currTask)
void	setCurrTopOp(Operator<? extends OperatorDesc> currTopOp)
void	setCurrUnionOp(UnionOperator currUnionOp)
void	setLinkedFileDescTasks(Map<FileSinkDesc,Task<? extends Serializable>> linkedFileDescTasks)
void	setMapCurrCtx(LinkedHashMap<Operator<? extends OperatorDesc>,GenMRProcContext.GenMapRedCtx> mapCurrCtx)
void	setMvTask(List<Task<MoveWork>> mvTask)
void	setOpTaskMap(HashMap<Operator<? extends OperatorDesc>,Task<? extends Serializable>> opTaskMap)
void	setParseCtx(ParseContext parseCtx)
void	setRootTasks(List<Task<? extends Serializable>> rootTasks)
void	setSeenFileSinkOps(List<FileSinkOperator> seenFileSinkOps)
void	setUnionTask(UnionOperator op, GenMRProcContext.GenMRUnionCtx uTask)

Methods inherited from class java.lang.Object

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

Constructor Detail

GenMRProcContext

public GenMRProcContext()

GenMRProcContext

public GenMRProcContext(HiveConf conf,
                HashMap<Operator<? extends OperatorDesc>,Task<? extends Serializable>> opTaskMap,
                ParseContext parseCtx,
                List<Task<MoveWork>> mvTask,
                List<Task<? extends Serializable>> rootTasks,
                LinkedHashMap<Operator<? extends OperatorDesc>,GenMRProcContext.GenMapRedCtx> mapCurrCtx,
                Set<ReadEntity> inputs,
                Set<WriteEntity> outputs)

Parameters:

conf - hive configuration

opTaskMap - reducer to task mapping

seenOps - operator already visited

parseCtx - current parse context

rootTasks - root tasks for the plan

mvTask - the final move task

mapCurrCtx - operator to task mappings

inputs - the set of input tables/partitions generated by the walk

outputs - the set of destinations generated by the walk

Method Detail

getOpTaskMap

public HashMap<Operator<? extends OperatorDesc>,Task<? extends Serializable>> getOpTaskMap()

Returns:

reducer to task mapping

setOpTaskMap

public void setOpTaskMap(HashMap<Operator<? extends OperatorDesc>,Task<? extends Serializable>> opTaskMap)

Parameters:

opTaskMap - reducer to task mapping

isSeenOp

public boolean isSeenOp(Task task,
               Operator operator)

addSeenOp

public void addSeenOp(Task task,
             Operator operator)

getSeenFileSinkOps

public List<FileSinkOperator> getSeenFileSinkOps()

Returns:

file operators already visited

setSeenFileSinkOps

public void setSeenFileSinkOps(List<FileSinkOperator> seenFileSinkOps)

Parameters:

seenFileSinkOps - file sink operators already visited

getParseCtx

public ParseContext getParseCtx()

Returns:

current parse context

setParseCtx

public void setParseCtx(ParseContext parseCtx)

Parameters:

parseCtx - current parse context

getMvTask

public List<Task<MoveWork>> getMvTask()

Returns:

the final move task

setMvTask

public void setMvTask(List<Task<MoveWork>> mvTask)

Parameters:

mvTask - the final move task

getRootTasks

public List<Task<? extends Serializable>> getRootTasks()

Returns:

root tasks for the plan

setRootTasks

public void setRootTasks(List<Task<? extends Serializable>> rootTasks)

Parameters:

rootTasks - root tasks for the plan

addRootIfPossible

public boolean addRootIfPossible(Task<? extends Serializable> task)

getMapCurrCtx

public LinkedHashMap<Operator<? extends OperatorDesc>,GenMRProcContext.GenMapRedCtx> getMapCurrCtx()

Returns:

operator to task mappings

setMapCurrCtx

public void setMapCurrCtx(LinkedHashMap<Operator<? extends OperatorDesc>,GenMRProcContext.GenMapRedCtx> mapCurrCtx)

Parameters:

mapCurrCtx - operator to task mappings

getCurrTask

public Task<? extends Serializable> getCurrTask()

Returns:

current task

setCurrTask

public void setCurrTask(Task<? extends Serializable> currTask)

Parameters:

currTask - current task

getCurrTopOp

public Operator<? extends OperatorDesc> getCurrTopOp()

Returns:

current top operator

setCurrTopOp

public void setCurrTopOp(Operator<? extends OperatorDesc> currTopOp)

Parameters:

currTopOp - current top operator

getCurrUnionOp

public UnionOperator getCurrUnionOp()

setCurrUnionOp

public void setCurrUnionOp(UnionOperator currUnionOp)

Parameters:

currUnionOp - current union operator

getCurrAliasId

public String getCurrAliasId()

Returns:

current top alias

setCurrAliasId

public void setCurrAliasId(String currAliasId)

Parameters:

currAliasId - current top alias

getUnionTask

public GenMRProcContext.GenMRUnionCtx getUnionTask(UnionOperator op)

setUnionTask

public void setUnionTask(UnionOperator op,
                GenMRProcContext.GenMRUnionCtx uTask)

getInputs

public Set<ReadEntity> getInputs()

Get the input set.

getOutputs

public Set<WriteEntity> getOutputs()

Get the output set.

getConf

public HiveConf getConf()

Returns:

the conf

setConf

public void setConf(HiveConf conf)

Parameters:

conf - the conf to set

getDependencyTaskForMultiInsert

public DependencyCollectionTask getDependencyTaskForMultiInsert()

Returns dependencyTaskForMultiInsert initializing it if necessary. dependencyTaskForMultiInsert serves as a mutual dependency for the final move tasks in a multi-insert query.

Returns:

getLinkedFileDescTasks

public Map<FileSinkDesc,Task<? extends Serializable>> getLinkedFileDescTasks()

setLinkedFileDescTasks

public void setLinkedFileDescTasks(Map<FileSinkDesc,Task<? extends Serializable>> linkedFileDescTasks)