Search in sources :

Example 6 with JoinFilterSplit

use of org.apache.druid.segment.join.filter.JoinFilterSplit in project druid by druid-io.

the class JoinFilterAnalyzerTest method test_filterPushDown_factToRegionToCountryLeftFilterOnRHSJoinConditionColumnsHelper.

private void test_filterPushDown_factToRegionToCountryLeftFilterOnRHSJoinConditionColumnsHelper(boolean hasLhsExpressionInJoinCondition) {
    Filter expressionFilter = new ExpressionDimFilter("\"rtc.countryIsoCode\" == 'CA'", ExprMacroTable.nil()).toFilter();
    Filter specialSelectorFilter = new SelectorFilter("rtc.countryIsoCode", "CA") {

        @Override
        public boolean supportsRequiredColumnRewrite() {
            return false;
        }
    };
    Filter originalFilter = new AndFilter(ImmutableList.of(new SelectorFilter("r1.regionIsoCode", "ON"), new SelectorFilter("rtc.countryIsoCode", "CA"), specialSelectorFilter, new BoundFilter(new BoundDimFilter("rtc.countryIsoCode", "CA", "CB", false, false, null, null, null)), expressionFilter, new InDimFilter("rtc.countryIsoCode", ImmutableSet.of("CA", "CA2", "CA3"), null, null).toFilter(), new OrFilter(ImmutableList.of(new SelectorFilter("channel", "#fr.wikipedia"), new SelectorFilter("rtc.countryIsoCode", "QQQ"), new BoundFilter(new BoundDimFilter("rtc.countryIsoCode", "YYY", "ZZZ", false, false, null, null, null)))), new OrFilter(ImmutableList.of(new SelectorFilter("namespace", "Main"), new SelectorFilter("rtc.countryIsoCode", "ABCDEF"), new SelectorFilter("rtc.countryName", "Canada"), new BoundFilter(new BoundDimFilter("rtc.countryIsoCode", "XYZXYZ", "XYZXYZ", false, false, null, null, null))))));
    JoinableClause factToRegionClause;
    if (hasLhsExpressionInJoinCondition) {
        factToRegionClause = new JoinableClause(FACT_TO_REGION_PREFIX, new IndexedTableJoinable(regionsTable), JoinType.LEFT, JoinConditionAnalysis.forExpression(StringUtils.format("\"%sregionIsoCode\" == upper(lower(regionIsoCode)) && \"%scountryIsoCode\" == upper(lower(countryIsoCode))", FACT_TO_REGION_PREFIX, FACT_TO_REGION_PREFIX), FACT_TO_REGION_PREFIX, ExprMacroTable.nil()));
    } else {
        factToRegionClause = factToRegion(JoinType.LEFT);
    }
    List<JoinableClause> joinableClauses = ImmutableList.of(factToRegionClause, regionToCountry(JoinType.LEFT));
    JoinFilterPreAnalysis joinFilterPreAnalysis = makeDefaultConfigPreAnalysis(originalFilter, joinableClauses, VirtualColumns.EMPTY);
    HashJoinSegmentStorageAdapter adapter = new HashJoinSegmentStorageAdapter(factSegment.asStorageAdapter(), joinableClauses, joinFilterPreAnalysis);
    String rewrittenCountryIsoCodeColumnName = hasLhsExpressionInJoinCondition ? "JOIN-FILTER-PUSHDOWN-VIRTUAL-COLUMN-1" : "countryIsoCode";
    String rewrittenRegionIsoCodeColumnName = hasLhsExpressionInJoinCondition ? "JOIN-FILTER-PUSHDOWN-VIRTUAL-COLUMN-0" : "regionIsoCode";
    Set<VirtualColumn> expectedVirtualColumns;
    if (hasLhsExpressionInJoinCondition) {
        expectedVirtualColumns = ImmutableSet.of(new ExpressionVirtualColumn(rewrittenRegionIsoCodeColumnName, "(upper [(lower [regionIsoCode])])", ColumnType.STRING, ExprMacroTable.nil()), new ExpressionVirtualColumn(rewrittenCountryIsoCodeColumnName, "(upper [(lower [countryIsoCode])])", ColumnType.STRING, ExprMacroTable.nil()));
    } else {
        expectedVirtualColumns = ImmutableSet.of();
    }
    JoinTestHelper.verifyCursors(adapter.makeCursors(originalFilter, Intervals.ETERNITY, VirtualColumns.EMPTY, Granularities.ALL, false, null), ImmutableList.of("page", FACT_TO_REGION_PREFIX + "regionName", REGION_TO_COUNTRY_PREFIX + "countryName"), ImmutableList.of(new Object[] { "Didier Leclair", "Ontario", "Canada" }));
    JoinFilterSplit expectedFilterSplit = new JoinFilterSplit(new AndFilter(ImmutableList.of(new SelectorFilter(rewrittenRegionIsoCodeColumnName, "ON"), new SelectorFilter(rewrittenCountryIsoCodeColumnName, "CA"), new InDimFilter(rewrittenCountryIsoCodeColumnName, ImmutableSet.of("CA"), null, null).toFilter(), new BoundFilter(new BoundDimFilter(rewrittenCountryIsoCodeColumnName, "CA", "CB", false, false, null, null, null)), new InDimFilter(rewrittenCountryIsoCodeColumnName, ImmutableSet.of("CA", "CA2", "CA3"), null, null).toFilter(), new OrFilter(ImmutableList.of(new SelectorFilter("channel", "#fr.wikipedia"), new SelectorFilter(rewrittenCountryIsoCodeColumnName, "QQQ"), new BoundFilter(new BoundDimFilter(rewrittenCountryIsoCodeColumnName, "YYY", "ZZZ", false, false, null, null, null)))), new OrFilter(ImmutableList.of(new SelectorFilter("namespace", "Main"), new SelectorFilter(rewrittenCountryIsoCodeColumnName, "ABCDEF"), new InDimFilter(rewrittenCountryIsoCodeColumnName, ImmutableSet.of("CA"), null, null).toFilter(), new BoundFilter(new BoundDimFilter(rewrittenCountryIsoCodeColumnName, "XYZXYZ", "XYZXYZ", false, false, null, null, null)))))), new AndFilter(ImmutableList.of(specialSelectorFilter, expressionFilter, new OrFilter(ImmutableList.of(new SelectorFilter("namespace", "Main"), new SelectorFilter("rtc.countryIsoCode", "ABCDEF"), new SelectorFilter("rtc.countryName", "Canada"), new BoundFilter(new BoundDimFilter("rtc.countryIsoCode", "XYZXYZ", "XYZXYZ", false, false, null, null, null)))))), expectedVirtualColumns);
    JoinFilterSplit actualFilterSplit = JoinFilterAnalyzer.splitFilter(joinFilterPreAnalysis);
    Assert.assertEquals(expectedFilterSplit, actualFilterSplit);
}
Also used : BoundDimFilter(org.apache.druid.query.filter.BoundDimFilter) BoundFilter(org.apache.druid.segment.filter.BoundFilter) JoinFilterSplit(org.apache.druid.segment.join.filter.JoinFilterSplit) OrFilter(org.apache.druid.segment.filter.OrFilter) AndFilter(org.apache.druid.segment.filter.AndFilter) SelectorFilter(org.apache.druid.segment.filter.SelectorFilter) ExpressionVirtualColumn(org.apache.druid.segment.virtual.ExpressionVirtualColumn) JoinFilterPreAnalysis(org.apache.druid.segment.join.filter.JoinFilterPreAnalysis) FalseFilter(org.apache.druid.segment.filter.FalseFilter) OrFilter(org.apache.druid.segment.filter.OrFilter) BoundDimFilter(org.apache.druid.query.filter.BoundDimFilter) SelectorFilter(org.apache.druid.segment.filter.SelectorFilter) AndFilter(org.apache.druid.segment.filter.AndFilter) BoundFilter(org.apache.druid.segment.filter.BoundFilter) InDimFilter(org.apache.druid.query.filter.InDimFilter) ExpressionDimFilter(org.apache.druid.query.filter.ExpressionDimFilter) Filter(org.apache.druid.query.filter.Filter) InDimFilter(org.apache.druid.query.filter.InDimFilter) IndexedTableJoinable(org.apache.druid.segment.join.table.IndexedTableJoinable) ExpressionVirtualColumn(org.apache.druid.segment.virtual.ExpressionVirtualColumn) VirtualColumn(org.apache.druid.segment.VirtualColumn) ExpressionDimFilter(org.apache.druid.query.filter.ExpressionDimFilter)

Example 7 with JoinFilterSplit

use of org.apache.druid.segment.join.filter.JoinFilterSplit in project druid by druid-io.

the class JoinFilterAnalyzerTest method test_filterPushDown_factToRegionToCountryLeftFilterOnTwoRHSColumnsSameValue.

@Test
public void test_filterPushDown_factToRegionToCountryLeftFilterOnTwoRHSColumnsSameValue() {
    Filter originalFilter = new AndFilter(ImmutableList.of(new SelectorFilter("r1.regionName", "California"), new SelectorFilter("r1.extraField", "California")));
    List<JoinableClause> joinableClauses = ImmutableList.of(factToRegion(JoinType.LEFT), regionToCountry(JoinType.LEFT));
    JoinFilterPreAnalysis joinFilterPreAnalysis = makeDefaultConfigPreAnalysis(originalFilter, joinableClauses, VirtualColumns.EMPTY);
    HashJoinSegmentStorageAdapter adapter = new HashJoinSegmentStorageAdapter(factSegment.asStorageAdapter(), joinableClauses, joinFilterPreAnalysis);
    JoinTestHelper.verifyCursors(adapter.makeCursors(originalFilter, Intervals.ETERNITY, VirtualColumns.EMPTY, Granularities.ALL, false, null), ImmutableList.of("page", FACT_TO_REGION_PREFIX + "regionName", REGION_TO_COUNTRY_PREFIX + "countryName"), ImmutableList.of());
    JoinFilterSplit expectedFilterSplit = new JoinFilterSplit(new AndFilter(ImmutableList.of(new InDimFilter("countryIsoCode", ImmutableSet.of("US"), null, null).toFilter(), new InDimFilter("regionIsoCode", ImmutableSet.of("CA"), null, null).toFilter(), new InDimFilter("countryIsoCode", ImmutableSet.of("MMMM", "AAAA"), null, null).toFilter(), new InDimFilter("regionIsoCode", ImmutableSet.of("MMMM", "AAAA"), null, null).toFilter())), originalFilter, ImmutableSet.of());
    JoinFilterSplit actualFilterSplit = JoinFilterAnalyzer.splitFilter(joinFilterPreAnalysis);
    Assert.assertEquals(expectedFilterSplit, actualFilterSplit);
}
Also used : AndFilter(org.apache.druid.segment.filter.AndFilter) SelectorFilter(org.apache.druid.segment.filter.SelectorFilter) JoinFilterPreAnalysis(org.apache.druid.segment.join.filter.JoinFilterPreAnalysis) FalseFilter(org.apache.druid.segment.filter.FalseFilter) OrFilter(org.apache.druid.segment.filter.OrFilter) BoundDimFilter(org.apache.druid.query.filter.BoundDimFilter) SelectorFilter(org.apache.druid.segment.filter.SelectorFilter) AndFilter(org.apache.druid.segment.filter.AndFilter) BoundFilter(org.apache.druid.segment.filter.BoundFilter) InDimFilter(org.apache.druid.query.filter.InDimFilter) ExpressionDimFilter(org.apache.druid.query.filter.ExpressionDimFilter) Filter(org.apache.druid.query.filter.Filter) JoinFilterSplit(org.apache.druid.segment.join.filter.JoinFilterSplit) InDimFilter(org.apache.druid.query.filter.InDimFilter) Test(org.junit.Test)

Example 8 with JoinFilterSplit

use of org.apache.druid.segment.join.filter.JoinFilterSplit in project druid by druid-io.

the class JoinFilterAnalyzerTest method test_filterPushDown_factToRegionToCountryLeftFilterOnChannelVirtualColumn.

@Test
public void test_filterPushDown_factToRegionToCountryLeftFilterOnChannelVirtualColumn() {
    List<JoinableClause> joinableClauses = ImmutableList.of(factToRegion(JoinType.LEFT), regionToCountry(JoinType.LEFT));
    Filter originalFilter = new AndFilter(ImmutableList.of(new SelectorFilter("v1", "virtual-column-#en.wikipedia")));
    VirtualColumns virtualColumns = VirtualColumns.create(ImmutableList.of(new ExpressionVirtualColumn("v1", "concat('virtual-column-', \"channel\")", ColumnType.STRING, TestExprMacroTable.INSTANCE)));
    JoinFilterPreAnalysis joinFilterPreAnalysis = makeDefaultConfigPreAnalysis(originalFilter, joinableClauses, virtualColumns);
    HashJoinSegmentStorageAdapter adapter = new HashJoinSegmentStorageAdapter(factSegment.asStorageAdapter(), joinableClauses, joinFilterPreAnalysis);
    JoinTestHelper.verifyCursors(adapter.makeCursors(originalFilter, Intervals.ETERNITY, virtualColumns, Granularities.ALL, false, null), ImmutableList.of("page", FACT_TO_REGION_PREFIX + "regionName", REGION_TO_COUNTRY_PREFIX + "countryName"), ImmutableList.of(new Object[] { "Talk:Oswald Tilghman", null, null }, new Object[] { "Peremptory norm", "New South Wales", "Australia" }, new Object[] { "President of India", "California", "United States" }, new Object[] { "Glasgow", "Kingston upon Hull", "United Kingdom" }, new Object[] { "Otjiwarongo Airport", "California", "United States" }, new Object[] { "Sarah Michelle Gellar", "Ontario", "Canada" }, new Object[] { "DirecTV", "North Carolina", "United States" }, new Object[] { "Carlo Curti", "California", "United States" }, new Object[] { "Giusy Ferreri discography", "Provincia di Varese", "Italy" }, new Object[] { "Roma-Bangkok", "Provincia di Varese", "Italy" }, new Object[] { "Old Anatolian Turkish", "Virginia", "United States" }, new Object[] { "Cream Soda", "Ainigriv", "States United" }, new Object[] { "Orange Soda", null, null }, new Object[] { "History of Fourems", "Fourems Province", "Fourems" }));
    JoinFilterSplit expectedFilterSplit = new JoinFilterSplit(new SelectorFilter("v1", "virtual-column-#en.wikipedia"), null, ImmutableSet.of());
    JoinFilterSplit actualFilterSplit = JoinFilterAnalyzer.splitFilter(joinFilterPreAnalysis);
    Assert.assertEquals(expectedFilterSplit, actualFilterSplit);
}
Also used : AndFilter(org.apache.druid.segment.filter.AndFilter) SelectorFilter(org.apache.druid.segment.filter.SelectorFilter) ExpressionVirtualColumn(org.apache.druid.segment.virtual.ExpressionVirtualColumn) JoinFilterPreAnalysis(org.apache.druid.segment.join.filter.JoinFilterPreAnalysis) FalseFilter(org.apache.druid.segment.filter.FalseFilter) OrFilter(org.apache.druid.segment.filter.OrFilter) BoundDimFilter(org.apache.druid.query.filter.BoundDimFilter) SelectorFilter(org.apache.druid.segment.filter.SelectorFilter) AndFilter(org.apache.druid.segment.filter.AndFilter) BoundFilter(org.apache.druid.segment.filter.BoundFilter) InDimFilter(org.apache.druid.query.filter.InDimFilter) ExpressionDimFilter(org.apache.druid.query.filter.ExpressionDimFilter) Filter(org.apache.druid.query.filter.Filter) JoinFilterSplit(org.apache.druid.segment.join.filter.JoinFilterSplit) VirtualColumns(org.apache.druid.segment.VirtualColumns) Test(org.junit.Test)

Example 9 with JoinFilterSplit

use of org.apache.druid.segment.join.filter.JoinFilterSplit in project druid by druid-io.

the class JoinFilterAnalyzerTest method test_filterPushDown_factToRegionExprToCountryLeftFilterOnCountryName.

@Test
public void test_filterPushDown_factToRegionExprToCountryLeftFilterOnCountryName() {
    Filter originalFilter = new SelectorFilter("rtc.countryName", "United States");
    JoinableClause regionExprToCountry = new JoinableClause(REGION_TO_COUNTRY_PREFIX, new IndexedTableJoinable(countriesTable), JoinType.LEFT, JoinConditionAnalysis.forExpression(StringUtils.format("reverse(\"%scountryIsoCode\") == \"%scountryIsoCode\"", FACT_TO_REGION_PREFIX, REGION_TO_COUNTRY_PREFIX), REGION_TO_COUNTRY_PREFIX, ExprMacroTable.nil()));
    List<JoinableClause> joinableClauses = ImmutableList.of(factToRegion(JoinType.LEFT), regionExprToCountry);
    JoinFilterPreAnalysis joinFilterPreAnalysis = makeDefaultConfigPreAnalysis(originalFilter, joinableClauses, VirtualColumns.EMPTY);
    HashJoinSegmentStorageAdapter adapter = new HashJoinSegmentStorageAdapter(factSegment.asStorageAdapter(), joinableClauses, joinFilterPreAnalysis);
    JoinTestHelper.verifyCursors(adapter.makeCursors(originalFilter, Intervals.ETERNITY, VirtualColumns.EMPTY, Granularities.ALL, false, null), ImmutableList.of("page", FACT_TO_REGION_PREFIX + "regionName", REGION_TO_COUNTRY_PREFIX + "countryName"), ImmutableList.of(new Object[] { "Cream Soda", "Ainigriv", "United States" }));
    JoinFilterSplit expectedFilterSplit = new JoinFilterSplit(null, new SelectorFilter("rtc.countryName", "United States"), ImmutableSet.of());
    JoinFilterSplit actualFilterSplit = JoinFilterAnalyzer.splitFilter(joinFilterPreAnalysis);
    Assert.assertEquals(expectedFilterSplit, actualFilterSplit);
}
Also used : SelectorFilter(org.apache.druid.segment.filter.SelectorFilter) JoinFilterPreAnalysis(org.apache.druid.segment.join.filter.JoinFilterPreAnalysis) FalseFilter(org.apache.druid.segment.filter.FalseFilter) OrFilter(org.apache.druid.segment.filter.OrFilter) BoundDimFilter(org.apache.druid.query.filter.BoundDimFilter) SelectorFilter(org.apache.druid.segment.filter.SelectorFilter) AndFilter(org.apache.druid.segment.filter.AndFilter) BoundFilter(org.apache.druid.segment.filter.BoundFilter) InDimFilter(org.apache.druid.query.filter.InDimFilter) ExpressionDimFilter(org.apache.druid.query.filter.ExpressionDimFilter) Filter(org.apache.druid.query.filter.Filter) JoinFilterSplit(org.apache.druid.segment.join.filter.JoinFilterSplit) IndexedTableJoinable(org.apache.druid.segment.join.table.IndexedTableJoinable) Test(org.junit.Test)

Example 10 with JoinFilterSplit

use of org.apache.druid.segment.join.filter.JoinFilterSplit in project druid by druid-io.

the class JoinFilterAnalyzerTest method test_filterPushDown_factToCountryInnerUsingCountryNumberFilterOnChannelAndCountryName.

@Test
public void test_filterPushDown_factToCountryInnerUsingCountryNumberFilterOnChannelAndCountryName() {
    List<JoinableClause> joinableClauses = ImmutableList.of(factToCountryOnNumber(JoinType.INNER));
    Filter originalFilter = new AndFilter(ImmutableList.of(new SelectorFilter("channel", "#en.wikipedia"), new SelectorFilter(FACT_TO_COUNTRY_ON_NUMBER_PREFIX + "countryName", "Australia")));
    JoinFilterPreAnalysis joinFilterPreAnalysis = makeDefaultConfigPreAnalysis(originalFilter, joinableClauses, VirtualColumns.EMPTY);
    HashJoinSegmentStorageAdapter adapter = new HashJoinSegmentStorageAdapter(factSegment.asStorageAdapter(), joinableClauses, joinFilterPreAnalysis);
    // In non-SQL-compatible mode, we get an extra row, since the 'null' countryNumber for "Talk:Oswald Tilghman"
    // is interpreted as 0 (a.k.a. Australia).
    JoinTestHelper.verifyCursors(adapter.makeCursors(originalFilter, Intervals.ETERNITY, VirtualColumns.EMPTY, Granularities.ALL, false, null), ImmutableList.of("page", "countryIsoCode", FACT_TO_COUNTRY_ON_NUMBER_PREFIX + "countryIsoCode", FACT_TO_COUNTRY_ON_NUMBER_PREFIX + "countryName", FACT_TO_COUNTRY_ON_NUMBER_PREFIX + "countryNumber"), NullHandling.sqlCompatible() ? ImmutableList.of(new Object[] { "Peremptory norm", "AU", "AU", "Australia", 0L }) : ImmutableList.of(new Object[] { "Talk:Oswald Tilghman", null, "AU", "Australia", 0L }, new Object[] { "Peremptory norm", "AU", "AU", "Australia", 0L }));
    JoinFilterSplit expectedFilterSplit = new JoinFilterSplit(new AndFilter(ImmutableList.of(new SelectorFilter("channel", "#en.wikipedia"), new InDimFilter("countryNumber", ImmutableSet.of("0"), null, null).toFilter())), new SelectorFilter(FACT_TO_COUNTRY_ON_NUMBER_PREFIX + "countryName", "Australia"), ImmutableSet.of());
    JoinFilterSplit actualFilterSplit = JoinFilterAnalyzer.splitFilter(joinFilterPreAnalysis);
    Assert.assertEquals(expectedFilterSplit, actualFilterSplit);
}
Also used : AndFilter(org.apache.druid.segment.filter.AndFilter) SelectorFilter(org.apache.druid.segment.filter.SelectorFilter) JoinFilterPreAnalysis(org.apache.druid.segment.join.filter.JoinFilterPreAnalysis) FalseFilter(org.apache.druid.segment.filter.FalseFilter) OrFilter(org.apache.druid.segment.filter.OrFilter) BoundDimFilter(org.apache.druid.query.filter.BoundDimFilter) SelectorFilter(org.apache.druid.segment.filter.SelectorFilter) AndFilter(org.apache.druid.segment.filter.AndFilter) BoundFilter(org.apache.druid.segment.filter.BoundFilter) InDimFilter(org.apache.druid.query.filter.InDimFilter) ExpressionDimFilter(org.apache.druid.query.filter.ExpressionDimFilter) Filter(org.apache.druid.query.filter.Filter) JoinFilterSplit(org.apache.druid.segment.join.filter.JoinFilterSplit) InDimFilter(org.apache.druid.query.filter.InDimFilter) Test(org.junit.Test)

Aggregations

Filter (org.apache.druid.query.filter.Filter)35 JoinFilterPreAnalysis (org.apache.druid.segment.join.filter.JoinFilterPreAnalysis)35 JoinFilterSplit (org.apache.druid.segment.join.filter.JoinFilterSplit)35 BoundDimFilter (org.apache.druid.query.filter.BoundDimFilter)34 ExpressionDimFilter (org.apache.druid.query.filter.ExpressionDimFilter)34 InDimFilter (org.apache.druid.query.filter.InDimFilter)34 AndFilter (org.apache.druid.segment.filter.AndFilter)34 BoundFilter (org.apache.druid.segment.filter.BoundFilter)34 FalseFilter (org.apache.druid.segment.filter.FalseFilter)34 OrFilter (org.apache.druid.segment.filter.OrFilter)34 SelectorFilter (org.apache.druid.segment.filter.SelectorFilter)34 Test (org.junit.Test)33 IndexedTableJoinable (org.apache.druid.segment.join.table.IndexedTableJoinable)8 ExpressionVirtualColumn (org.apache.druid.segment.virtual.ExpressionVirtualColumn)6 VirtualColumns (org.apache.druid.segment.VirtualColumns)3 JoinFilterPreAnalysisKey (org.apache.druid.segment.join.filter.JoinFilterPreAnalysisKey)3 JoinableClauses (org.apache.druid.segment.join.filter.JoinableClauses)3 JoinFilterRewriteConfig (org.apache.druid.segment.join.filter.rewrite.JoinFilterRewriteConfig)3 VirtualColumn (org.apache.druid.segment.VirtualColumn)2 Lists (com.google.common.collect.Lists)1