Package org.apache.drill.exec.planner.index

Class FunctionalIndexHelper

java.lang.Object

org.apache.drill.exec.planner.index.FunctionalIndexHelper

public class FunctionalIndexHelper
extends Object

Constructor Summary

Constructors

Constructor	Description
FunctionalIndexHelper()

Method Summary

Modifier and Type	Method	Description
static org.apache.calcite.rex.RexNode	convertConditionForIndexScan(org.apache.calcite.rex.RexNode idxCondition, org.apache.calcite.rel.RelNode origScan, org.apache.calcite.rel.type.RelDataType idxRowType, org.apache.calcite.rex.RexBuilder builder, FunctionalIndexInfo functionInfo)
static org.apache.calcite.rel.type.RelDataType	convertRowTypeForIndexScan(DrillScanRelBase origScan, IndexableExprMarker idxMarker, IndexGroupScan idxScan, FunctionalIndexInfo functionInfo)	For IndexScan in non-covering case, rowType to return contains only row_key('_id') of primary table.
static org.apache.calcite.rel.type.RelDataType	rewriteFunctionalRowType(org.apache.calcite.rel.RelNode origScan, IndexCallContext indexContext, FunctionalIndexInfo functionInfo)
static org.apache.calcite.rel.type.RelDataType	rewriteFunctionalRowType(org.apache.calcite.rel.RelNode origScan, IndexCallContext indexContext, FunctionalIndexInfo functionInfo, Collection<SchemaPath> addedPaths)	if a field in rowType serves only the to-be-replaced column(s), we should replace it with new name "$1", otherwise we should keep this dataTypeField and add a new one for "$1"

Methods inherited from class java.lang.Object

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

Constructor Details

FunctionalIndexHelper

public FunctionalIndexHelper()

Method Details

rewriteFunctionalRowType

public static org.apache.calcite.rel.type.RelDataType rewriteFunctionalRowType(org.apache.calcite.rel.RelNode origScan,
 IndexCallContext indexContext,
 FunctionalIndexInfo functionInfo)

rewriteFunctionalRowType

public static org.apache.calcite.rel.type.RelDataType rewriteFunctionalRowType(org.apache.calcite.rel.RelNode origScan,
 IndexCallContext indexContext,
 FunctionalIndexInfo functionInfo,
 Collection<SchemaPath> addedPaths)

if a field in rowType serves only the to-be-replaced column(s), we should replace it with new name "$1", otherwise we should keep this dataTypeField and add a new one for "$1"

Parameters:

origScan - the original scan whose rowtype is to be rewritten

indexContext - the index plan context

functionInfo - functional index information that may impact rewrite

Returns:

convertRowTypeForIndexScan

public static org.apache.calcite.rel.type.RelDataType convertRowTypeForIndexScan(DrillScanRelBase origScan,
 IndexableExprMarker idxMarker,
 IndexGroupScan idxScan,
 FunctionalIndexInfo functionInfo)

For IndexScan in non-covering case, rowType to return contains only row_key('_id') of primary table. so the rowType for IndexScan should be converted from [Primary_table.row_key, primary_table.indexed_col] to [indexTable.row_key(primary_table.indexed_col), indexTable.invalid input: '<'primary_key.row_key> (Primary_table.row_key)] This will impact the columns of scan, the rowType of ScanRel

Parameters:

origScan -

idxMarker - the IndexableExprMarker that has analyzed original index condition on top of index scan

idxScan -

Returns:

convertConditionForIndexScan

public static org.apache.calcite.rex.RexNode convertConditionForIndexScan(org.apache.calcite.rex.RexNode idxCondition,
 org.apache.calcite.rel.RelNode origScan,
 org.apache.calcite.rel.type.RelDataType idxRowType,
 org.apache.calcite.rex.RexBuilder builder,
 FunctionalIndexInfo functionInfo)