org.apache.hadoop.hive.ql.parse

Class ColumnStatsAutoGatherContext

java.lang.Object

org.apache.hadoop.hive.ql.parse.ColumnStatsAutoGatherContext

public class ColumnStatsAutoGatherContext
extends Object

ColumnStatsAutoGatherContext: This is passed to the compiler when set hive.stats.autogather=true during the INSERT, INSERT OVERWRITE, or CTAS commands.

Field Summary

Fields

Modifier and Type	Field and Description
BaseSemanticAnalyzer.AnalyzeRewriteContext	analyzeRewrite

Constructor Summary

Constructors

Constructor and Description
ColumnStatsAutoGatherContext(SemanticAnalyzer sa, HiveConf conf, Operator<? extends OperatorDesc> op, Table tbl, Map<String,String> partSpec, boolean isInsertInto, Context ctx)

Method Summary

Modifier and Type	Method and Description
static boolean	canRunAutogatherStats(Operator curr)
BaseSemanticAnalyzer.AnalyzeRewriteContext	getAnalyzeRewrite()
String	getCompleteName()
List<LoadFileDesc>	getLoadFileWork()
void	insertAnalyzePipeline() Generate the statement of analyze table [tablename] compute statistics for columns In non-partitioned table case, it will generate TS-SEL-GBY-RS-GBY-SEL-FS operator In static-partitioned table case, it will generate TS-FIL(partitionKey)-SEL-GBY(partitionKey)-RS-GBY-SEL-FS operator In dynamic-partitioned table case, it will generate TS-SEL-GBY(partitionKey)-RS-GBY-SEL-FS operator However, we do not need to specify the partition-spec because (1) the data is going to be inserted to that specific partition (2) we can compose the static/dynamic partition using a select operator in replaceSelectOperatorProcess.
void	insertTableValuesAnalyzePipeline() Generate the statement of SELECT compute_stats(col1) compute_stats(col2),..., similar to the one generated from ANALYZE TABLE t1 COMPUTE STATISTICS FOR COLUMNS, but t1 is replaced by a TABLE(VALUES(cast(null as int),cast(null as string))) AS t1(col1,col2).
boolean	isInsertInto()
void	setAnalyzeRewrite(BaseSemanticAnalyzer.AnalyzeRewriteContext analyzeRewrite)

Methods inherited from class java.lang.Object

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

Field Detail

analyzeRewrite

public BaseSemanticAnalyzer.AnalyzeRewriteContext analyzeRewrite

Constructor Detail

ColumnStatsAutoGatherContext

public ColumnStatsAutoGatherContext(SemanticAnalyzer sa,
                                    HiveConf conf,
                                    Operator<? extends OperatorDesc> op,
                                    Table tbl,
                                    Map<String,String> partSpec,
                                    boolean isInsertInto,
                                    Context ctx)
                             throws SemanticException

Throws:

SemanticException

Method Detail

getLoadFileWork

public List<LoadFileDesc> getLoadFileWork()

getAnalyzeRewrite

public BaseSemanticAnalyzer.AnalyzeRewriteContext getAnalyzeRewrite()

setAnalyzeRewrite

public void setAnalyzeRewrite(BaseSemanticAnalyzer.AnalyzeRewriteContext analyzeRewrite)

insertAnalyzePipeline

public void insertAnalyzePipeline()
                           throws SemanticException

Generate the statement of analyze table [tablename] compute statistics for columns In non-partitioned table case, it will generate TS-SEL-GBY-RS-GBY-SEL-FS operator In static-partitioned table case, it will generate TS-FIL(partitionKey)-SEL-GBY(partitionKey)-RS-GBY-SEL-FS operator In dynamic-partitioned table case, it will generate TS-SEL-GBY(partitionKey)-RS-GBY-SEL-FS operator However, we do not need to specify the partition-spec because (1) the data is going to be inserted to that specific partition (2) we can compose the static/dynamic partition using a select operator in replaceSelectOperatorProcess.

Throws:

SemanticException

insertTableValuesAnalyzePipeline

public void insertTableValuesAnalyzePipeline()
                                      throws SemanticException

Generate the statement of SELECT compute_stats(col1) compute_stats(col2),..., similar to the one generated from ANALYZE TABLE t1 COMPUTE STATISTICS FOR COLUMNS, but t1 is replaced by a TABLE(VALUES(cast(null as int),cast(null as string))) AS t1(col1,col2). We use TABLE-VALUES statement for computing stats for CTAS statement because in those cases the table has not been created yet. Once the plan for the SELECT statement is generated, we connect it to the existing CTAS plan as we do for INSERT or INSERT OVERWRITE.

Throws:

SemanticException

getCompleteName

public String getCompleteName()

isInsertInto

public boolean isInsertInto()

canRunAutogatherStats

public static boolean canRunAutogatherStats(Operator curr)