| /* |
| * Licensed to the Apache Software Foundation (ASF) under one or more |
| * contributor license agreements. See the NOTICE file distributed with |
| * this work for additional information regarding copyright ownership. |
| * The ASF licenses this file to You under the Apache License, Version 2.0 |
| * (the "License"); you may not use this file except in compliance with |
| * the License. You may obtain a copy of the License at |
| * |
| * http://www.apache.org/licenses/LICENSE-2.0 |
| * |
| * Unless required by applicable law or agreed to in writing, software |
| * distributed under the License is distributed on an "AS IS" BASIS, |
| * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. |
| * See the License for the specific language governing permissions and |
| * limitations under the License. |
| */ |
| |
| package org.apache.carbondata.core.index; |
| |
| import java.io.IOException; |
| import java.util.ArrayList; |
| import java.util.Collections; |
| import java.util.HashSet; |
| import java.util.List; |
| import java.util.Set; |
| |
| import org.apache.carbondata.common.annotations.InterfaceAudience; |
| import org.apache.carbondata.core.constants.CarbonCommonConstants; |
| import org.apache.carbondata.core.index.dev.expr.AndIndexExprWrapper; |
| import org.apache.carbondata.core.index.dev.expr.IndexExprWrapper; |
| import org.apache.carbondata.core.index.dev.expr.IndexExprWrapperImpl; |
| import org.apache.carbondata.core.index.dev.expr.OrIndexExprWrapper; |
| import org.apache.carbondata.core.index.status.IndexStatus; |
| import org.apache.carbondata.core.metadata.schema.table.CarbonTable; |
| import org.apache.carbondata.core.scan.expression.ColumnExpression; |
| import org.apache.carbondata.core.scan.expression.Expression; |
| import org.apache.carbondata.core.scan.expression.MatchExpression; |
| import org.apache.carbondata.core.scan.expression.logical.AndExpression; |
| import org.apache.carbondata.core.scan.expression.logical.OrExpression; |
| import org.apache.carbondata.core.scan.filter.intf.ExpressionType; |
| import org.apache.carbondata.core.scan.filter.resolver.FilterResolverIntf; |
| import org.apache.carbondata.core.scan.filter.resolver.resolverinfo.TrueConditionalResolverImpl; |
| |
| /** |
| * This chooser does 2 jobs. |
| * 1. Based on filter expression it converts the available Indexes to Index expression. |
| * For example, there are 2 Indexes available on table1 |
| * Index1 : column1 |
| * Index2 : column2 |
| * Query: select * from table1 where column1 ='a' and column2 =b |
| * For the above query, we create Index expression as AndIndexExpression(Index1, Index2). |
| * So for the above query both the Indexes are included and the output of them will be |
| * applied AND condition to improve the performance |
| * |
| * 2. It chooses the Index out of available Indexes based on simple logic. |
| * Like if there is filter condition on column1 then for |
| * supposing 2 Indexes(1. column1 2. column1+column2) are supporting this column then we choose |
| * the Index which has fewer columns that is the first Index. |
| */ |
| @InterfaceAudience.Internal |
| public class IndexChooser { |
| |
| private CarbonTable carbonTable; |
| private List<TableIndex> cgIndexes; |
| private List<TableIndex> fgIndexes; |
| |
| public IndexChooser(CarbonTable carbonTable) throws IOException { |
| this.carbonTable = carbonTable; |
| // read all indexes for this table and populate CG and FG index list |
| List<TableIndex> visibleIndexes = carbonTable.getAllVisibleIndexes(); |
| cgIndexes = new ArrayList<>(visibleIndexes.size()); |
| fgIndexes = new ArrayList<>(visibleIndexes.size()); |
| for (TableIndex visibleIndex : visibleIndexes) { |
| if (visibleIndex.getIndexSchema().getProperties().get(CarbonCommonConstants.INDEX_STATUS) |
| != null && visibleIndex.getIndexSchema().getProperties() |
| .get(CarbonCommonConstants.INDEX_STATUS).equalsIgnoreCase(IndexStatus.ENABLED.name())) { |
| IndexLevel level = visibleIndex.getIndexFactory().getIndexLevel(); |
| if (level == IndexLevel.CG) { |
| cgIndexes.add(visibleIndex); |
| } else { |
| fgIndexes.add(visibleIndex); |
| } |
| } |
| } |
| } |
| |
| /** |
| * Return a chosen Index based on input filter. See {@link IndexChooser} |
| */ |
| public IndexExprWrapper choose(FilterResolverIntf filter) { |
| if (filter != null) { |
| Expression expression = filter.getFilterExpression(); |
| // First check for FG Indexes if any exist |
| ExpressionTuple tuple = selectIndex(expression, fgIndexes, filter); |
| if (tuple.indexExprWrapper == null) { |
| // Check for CG Index |
| tuple = selectIndex(expression, cgIndexes, filter); |
| } |
| if (tuple.indexExprWrapper != null) { |
| return tuple.indexExprWrapper; |
| } |
| } |
| // Return the default Index if no other Index exists. |
| return new IndexExprWrapperImpl( |
| IndexStoreManager.getInstance().getDefaultIndex(carbonTable), filter); |
| } |
| |
| /** |
| * Return a chosen FG Index based on input filter. See {@link IndexChooser} |
| */ |
| public IndexExprWrapper chooseFGIndex(FilterResolverIntf resolverIntf) { |
| return chooseIndex(IndexLevel.FG, resolverIntf); |
| } |
| |
| /** |
| * Return a chosen CG Index based on input filter. See {@link IndexChooser} |
| */ |
| public IndexExprWrapper chooseCGIndex(FilterResolverIntf resolverIntf) { |
| return chooseIndex(IndexLevel.CG, resolverIntf); |
| } |
| |
| IndexExprWrapper chooseIndex(IndexLevel level, FilterResolverIntf resolverIntf) { |
| if (resolverIntf != null) { |
| Expression expression = resolverIntf.getFilterExpression(); |
| List<TableIndex> indexes = level == IndexLevel.CG ? cgIndexes : fgIndexes; |
| if (indexes.size() > 0) { |
| ExpressionTuple tuple = selectIndex(expression, indexes, resolverIntf); |
| if (tuple.indexExprWrapper != null) { |
| return tuple.indexExprWrapper; |
| } |
| } |
| } |
| return null; |
| } |
| |
| /** |
| * Returns default blocklet index |
| * @param carbonTable |
| * @param resolverIntf |
| * @return |
| */ |
| public static IndexExprWrapper getDefaultIndex(CarbonTable carbonTable, |
| FilterResolverIntf resolverIntf) { |
| // Return the default index if no other index exists. |
| return new IndexExprWrapperImpl( |
| IndexStoreManager.getInstance().getDefaultIndex(carbonTable), resolverIntf); |
| } |
| |
| private ExpressionTuple selectIndex(Expression expression, List<TableIndex> allIndex, |
| FilterResolverIntf filterResolverIntf) { |
| switch (expression.getFilterExpressionType()) { |
| case AND: |
| if (expression instanceof AndExpression) { |
| AndExpression andExpression = (AndExpression) expression; |
| ExpressionTuple left = selectIndex(andExpression.getLeft(), allIndex, |
| filterResolverIntf.getLeft()); |
| ExpressionTuple right = selectIndex(andExpression.getRight(), allIndex, |
| filterResolverIntf.getRight()); |
| Set<ExpressionType> filterExpressionTypes = new HashSet<>(); |
| // If both left and right has Index then we can either merge both Indexes to single |
| // Index if possible. Otherwise apply AND expression. |
| if (left.indexExprWrapper != null && right.indexExprWrapper != null) { |
| filterExpressionTypes.addAll(left.filterExpressionTypes); |
| filterExpressionTypes.addAll(right.filterExpressionTypes); |
| List<ColumnExpression> columnExpressions = new ArrayList<>(); |
| columnExpressions.addAll(left.columnExpressions); |
| columnExpressions.addAll(right.columnExpressions); |
| // Check if we can merge them to single Index. |
| TableIndex index = |
| chooseIndex(allIndex, columnExpressions, filterExpressionTypes); |
| TrueConditionalResolverImpl resolver = new TrueConditionalResolverImpl( |
| new AndExpression(left.expression, right.expression), false, |
| true); |
| if (index != null) { |
| ExpressionTuple tuple = new ExpressionTuple(); |
| tuple.columnExpressions = columnExpressions; |
| tuple.indexExprWrapper = new IndexExprWrapperImpl(index, resolver); |
| tuple.expression = resolver.getFilterExpression(); |
| return tuple; |
| } else { |
| // Apply AND expression. |
| ExpressionTuple tuple = new ExpressionTuple(); |
| tuple.columnExpressions = columnExpressions; |
| tuple.indexExprWrapper = new AndIndexExprWrapper(left.indexExprWrapper, |
| right.indexExprWrapper, resolver); |
| tuple.expression = resolver.getFilterExpression(); |
| return tuple; |
| } |
| } else if (left.indexExprWrapper != null) { |
| return left; |
| } else if (right.indexExprWrapper != null) { |
| return right; |
| } else { |
| return left; |
| } |
| } |
| break; |
| case OR: |
| if (expression instanceof OrExpression) { |
| OrExpression orExpression = (OrExpression) expression; |
| ExpressionTuple left = selectIndex(orExpression.getLeft(), allIndex, |
| filterResolverIntf.getLeft()); |
| ExpressionTuple right = selectIndex(orExpression.getRight(), allIndex, |
| filterResolverIntf.getRight()); |
| // If both left and right has Index then we can either merge both Indexes to single |
| // Index if possible. Otherwise apply OR expression. |
| if (left.indexExprWrapper != null && right.indexExprWrapper != null) { |
| TrueConditionalResolverImpl resolver = new TrueConditionalResolverImpl( |
| new OrExpression(left.expression, right.expression), false, |
| true); |
| List<ColumnExpression> columnExpressions = new ArrayList<>(); |
| columnExpressions.addAll(left.columnExpressions); |
| columnExpressions.addAll(right.columnExpressions); |
| ExpressionTuple tuple = new ExpressionTuple(); |
| tuple.columnExpressions = columnExpressions; |
| tuple.indexExprWrapper = new OrIndexExprWrapper(left.indexExprWrapper, |
| right.indexExprWrapper, resolver); |
| tuple.expression = resolver.getFilterExpression(); |
| return tuple; |
| } else { |
| left.indexExprWrapper = null; |
| return left; |
| } |
| } |
| break; |
| default: |
| ExpressionTuple tuple = new ExpressionTuple(); |
| extractColumnExpression(expression, tuple.columnExpressions); |
| Set<ExpressionType> filterExpressionTypes = new HashSet<>(); |
| filterExpressionTypes.add(expression.getFilterExpressionType()); |
| TrueConditionalResolverImpl resolver = new TrueConditionalResolverImpl( |
| filterResolverIntf.getFilterExpression(), false, |
| true); |
| TableIndex index = |
| chooseIndex(allIndex, tuple.columnExpressions, filterExpressionTypes); |
| if (index != null) { |
| tuple.indexExprWrapper = new IndexExprWrapperImpl(index, resolver); |
| tuple.filterExpressionTypes.addAll(filterExpressionTypes); |
| tuple.expression = filterResolverIntf.getFilterExpression(); |
| } |
| return tuple; |
| } |
| return new ExpressionTuple(); |
| } |
| |
| private void extractColumnExpression(Expression expression, |
| List<ColumnExpression> columnExpressions) { |
| if (expression instanceof ColumnExpression) { |
| columnExpressions.add((ColumnExpression) expression); |
| } else if (expression instanceof MatchExpression) { |
| // this is a special case for lucene |
| // build a fake ColumnExpression to filter Indexes which contain target column |
| // a Lucene query string is alike "column:query term" |
| String[] queryItems = expression.getString().split(":", 2); |
| if (queryItems.length == 2) { |
| columnExpressions.add(new ColumnExpression(queryItems[0], null)); |
| } |
| } else if (expression != null) { |
| List<Expression> children = expression.getChildren(); |
| if (children != null && children.size() > 0) { |
| for (Expression exp : children) { |
| if (exp != null && exp.getFilterExpressionType() != ExpressionType.UNKNOWN) { |
| extractColumnExpression(exp, columnExpressions); |
| } |
| } |
| } |
| } |
| } |
| |
| private TableIndex chooseIndex(List<TableIndex> allIndex, |
| List<ColumnExpression> columnExpressions, Set<ExpressionType> expressionTypes) { |
| List<IndexTuple> tuples = new ArrayList<>(); |
| for (TableIndex index : allIndex) { |
| if (null != index.getIndexFactory().getMeta() && contains( |
| index.getIndexFactory().getMeta(), columnExpressions, expressionTypes)) { |
| tuples.add( |
| new IndexTuple(index.getIndexFactory().getMeta().getIndexedColumns().size(), |
| index)); |
| } |
| } |
| if (tuples.size() > 0) { |
| Collections.sort(tuples); |
| return tuples.get(0).index; |
| } |
| return null; |
| } |
| |
| /** |
| * Return true if the input Index contains the column that needed in |
| * specified expression |
| */ |
| private boolean contains(IndexMeta mapMeta, List<ColumnExpression> columnExpressions, |
| Set<ExpressionType> expressionTypes) { |
| if (mapMeta.getIndexedColumns().size() == 0 || columnExpressions.size() == 0) { |
| return false; |
| } |
| boolean contains = true; |
| for (ColumnExpression expression : columnExpressions) { |
| if (!mapMeta.getIndexedColumnNames().contains(expression.getColumnName()) || |
| !mapMeta.getOptimizedOperation().containsAll(expressionTypes)) { |
| contains = false; |
| break; |
| } |
| } |
| return contains; |
| } |
| |
| private static class ExpressionTuple { |
| |
| IndexExprWrapper indexExprWrapper; |
| |
| List<ColumnExpression> columnExpressions = new ArrayList<>(); |
| |
| Set<ExpressionType> filterExpressionTypes = new HashSet<>(); |
| |
| Expression expression; |
| |
| } |
| |
| private static class IndexTuple implements Comparable<IndexTuple> { |
| |
| int order; |
| |
| TableIndex index; |
| |
| public IndexTuple(int order, TableIndex index) { |
| this.order = order; |
| this.index = index; |
| } |
| |
| @Override |
| public int compareTo(IndexTuple o) { |
| return order - o.order; |
| } |
| |
| @Override |
| public boolean equals(Object o) { |
| if (this == o) return true; |
| if (o == null || getClass() != o.getClass()) return false; |
| |
| IndexTuple that = (IndexTuple) o; |
| |
| if (order != that.order) return false; |
| return index != null ? index.equals(that.index) : that.index == null; |
| } |
| |
| @Override |
| public int hashCode() { |
| int result = order; |
| result = 31 * result + (index != null ? index.hashCode() : 0); |
| return result; |
| } |
| } |
| |
| } |