Search in sources :

Example 31 with ScannerFactory

use of datawave.query.tables.ScannerFactory in project datawave by NationalSecurityAgency.

the class RangeStreamTest method testIntersectionOfTwoUnionsAllIndexed.

@Test
public void testIntersectionOfTwoUnionsAllIndexed() throws Exception {
    String originalQuery = "(FOO == 'bag' || FOO == 'bar') && (NUM == 'shell' || NUM == 'tacocat')";
    ASTJexlScript script = JexlASTHelper.parseJexlQuery(originalQuery);
    config.setBeginDate(new Date(0));
    config.setEndDate(new Date(System.currentTimeMillis()));
    Multimap<String, Type<?>> dataTypes = HashMultimap.create();
    dataTypes.putAll("FOO", Sets.newHashSet(new LcNoDiacriticsType()));
    dataTypes.putAll("NUM", Sets.newHashSet(new LcNoDiacriticsType()));
    config.setQueryFieldsDatatypes(dataTypes);
    config.setIndexedFields(dataTypes);
    MockMetadataHelper helper = new MockMetadataHelper();
    helper.setIndexedFields(dataTypes.keySet());
    RangeStream rangeStream = new RangeStream(config, new ScannerFactory(config.getConnector()), helper).setLimitScanners(true);
    rangeStream.streamPlans(script);
    // streamPlans(script) to populate the StreamContext.
    assertEquals(IndexStream.StreamContext.ABSENT, rangeStream.context());
    assertFalse(rangeStream.iterator().hasNext());
}
Also used : LcNoDiacriticsType(datawave.data.type.LcNoDiacriticsType) MockMetadataHelper(datawave.query.util.MockMetadataHelper) NoOpType(datawave.data.type.NoOpType) NumberType(datawave.data.type.NumberType) LcNoDiacriticsType(datawave.data.type.LcNoDiacriticsType) Type(datawave.data.type.Type) ASTJexlScript(org.apache.commons.jexl2.parser.ASTJexlScript) ScannerFactory(datawave.query.tables.ScannerFactory) Test(org.junit.Test)

Example 32 with ScannerFactory

use of datawave.query.tables.ScannerFactory in project datawave by NationalSecurityAgency.

the class RangeStreamTest method testTheSimplestOfQueries.

@Test
public void testTheSimplestOfQueries() throws Exception {
    String originalQuery = "FOO == 'bag'";
    ASTJexlScript script = JexlASTHelper.parseJexlQuery(originalQuery);
    config.setBeginDate(new Date(0));
    config.setEndDate(new Date(System.currentTimeMillis()));
    Multimap<String, Type<?>> dataTypes = HashMultimap.create();
    dataTypes.putAll("FOO", Sets.newHashSet(new LcNoDiacriticsType()));
    dataTypes.putAll("NUM", Sets.newHashSet(new NumberType()));
    config.setQueryFieldsDatatypes(dataTypes);
    config.setIndexedFields(dataTypes);
    MockMetadataHelper helper = new MockMetadataHelper();
    helper.setIndexedFields(dataTypes.keySet());
    Set<Range> expectedRanges = Sets.newHashSet(makeTestRange("20190314", "datatype1\u0000234"), makeTestRange("20190314", "datatype1\u0000345"));
    for (QueryPlan queryPlan : new RangeStream(config, new ScannerFactory(config.getConnector()), helper).streamPlans(script)) {
        for (Range range : queryPlan.getRanges()) {
            assertTrue("Tried to remove unexpected range from expected ranges: " + range.toString(), expectedRanges.remove(range));
        }
    }
    assertTrue("Expected ranges not found in query plan: " + expectedRanges.toString(), expectedRanges.isEmpty());
}
Also used : ASTJexlScript(org.apache.commons.jexl2.parser.ASTJexlScript) RangeFactoryForTests.makeTestRange(datawave.common.test.utils.query.RangeFactoryForTests.makeTestRange) Range(org.apache.accumulo.core.data.Range) QueryPlan(datawave.query.planner.QueryPlan) ScannerFactory(datawave.query.tables.ScannerFactory) LcNoDiacriticsType(datawave.data.type.LcNoDiacriticsType) MockMetadataHelper(datawave.query.util.MockMetadataHelper) NoOpType(datawave.data.type.NoOpType) NumberType(datawave.data.type.NumberType) LcNoDiacriticsType(datawave.data.type.LcNoDiacriticsType) Type(datawave.data.type.Type) NumberType(datawave.data.type.NumberType) Test(org.junit.Test)

Example 33 with ScannerFactory

use of datawave.query.tables.ScannerFactory in project datawave by NationalSecurityAgency.

the class RangeStreamTest method testIntersection_NestedUnionOfHighCardinalityTerm_withSeek.

// (A && (B || C))
@Test
public void testIntersection_NestedUnionOfHighCardinalityTerm_withSeek() throws Exception {
    String originalQuery = "(FOO == 'lowest_card' && (FOO == 'high_card' || FOO == 'highest_card'))";
    ASTJexlScript script = JexlASTHelper.parseJexlQuery(originalQuery);
    SimpleDateFormat sdf = new SimpleDateFormat("yyyyMMdd");
    config.setBeginDate(sdf.parse("20190310"));
    config.setEndDate(sdf.parse("20190320"));
    config.setDatatypeFilter(Sets.newHashSet("datatype1", "datatype2"));
    Multimap<String, Type<?>> dataTypes = HashMultimap.create();
    dataTypes.putAll("FOO", Sets.newHashSet(new LcNoDiacriticsType()));
    dataTypes.putAll("LAUGH", Sets.newHashSet(new LcNoDiacriticsType()));
    config.setQueryFieldsDatatypes(dataTypes);
    config.setIndexedFields(dataTypes);
    MockMetadataHelper helper = new MockMetadataHelper();
    helper.setIndexedFields(dataTypes.keySet());
    // Create expected ranges verbosely, so it is obvious which shards contribute to the results.
    Range range1 = makeTestRange("20190310_1", "datatype1\u0000a.b.c");
    Range range2 = makeTestRange("20190314_22", "datatype1\u0000a.b.c");
    Range range3 = makeTestRange("20190315_49", "datatype1\u0000a.b.c");
    Set<Range> expectedRanges = Sets.newHashSet(range1, range2, range3);
    RangeStream rangeStream = new RangeStream(config, new ScannerFactory(config.getConnector(), 1), helper);
    rangeStream.setLimitScanners(true);
    CloseableIterable<QueryPlan> queryPlans = rangeStream.streamPlans(script);
    assertEquals(IndexStream.StreamContext.PRESENT, rangeStream.context());
    for (QueryPlan queryPlan : queryPlans) {
        Iterable<Range> ranges = queryPlan.getRanges();
        for (Range range : ranges) {
            assertTrue("Tried to remove unexpected range " + range.toString() + "\nfrom expected ranges: " + expectedRanges.toString(), expectedRanges.remove(range));
        }
    }
    assertTrue("Expected ranges not found in query plan: " + expectedRanges.toString(), expectedRanges.isEmpty());
}
Also used : ASTJexlScript(org.apache.commons.jexl2.parser.ASTJexlScript) RangeFactoryForTests.makeTestRange(datawave.common.test.utils.query.RangeFactoryForTests.makeTestRange) Range(org.apache.accumulo.core.data.Range) ScannerFactory(datawave.query.tables.ScannerFactory) QueryPlan(datawave.query.planner.QueryPlan) LcNoDiacriticsType(datawave.data.type.LcNoDiacriticsType) MockMetadataHelper(datawave.query.util.MockMetadataHelper) NoOpType(datawave.data.type.NoOpType) NumberType(datawave.data.type.NumberType) LcNoDiacriticsType(datawave.data.type.LcNoDiacriticsType) Type(datawave.data.type.Type) SimpleDateFormat(java.text.SimpleDateFormat) Test(org.junit.Test)

Example 34 with ScannerFactory

use of datawave.query.tables.ScannerFactory in project datawave by NationalSecurityAgency.

the class RangeStreamTest method testOrNoFieldIndexed.

@Test
public void testOrNoFieldIndexed() throws Exception {
    String originalQuery = "(TACO == 'bag' || TACO == 'ba')";
    ASTJexlScript script = JexlASTHelper.parseJexlQuery(originalQuery);
    config.setBeginDate(new Date(0));
    config.setEndDate(new Date(System.currentTimeMillis()));
    Multimap<String, Type<?>> dataTypes = HashMultimap.create();
    dataTypes.putAll("FOO", Sets.newHashSet(new LcNoDiacriticsType()));
    dataTypes.putAll("NUM", Sets.newHashSet(new NumberType()));
    config.setQueryFieldsDatatypes(dataTypes);
    config.setIndexedFields(dataTypes);
    MockMetadataHelper helper = new MockMetadataHelper();
    helper.setIndexedFields(dataTypes.keySet());
    assertFalse(new RangeStream(config, new ScannerFactory(config.getConnector()), helper).streamPlans(script).iterator().hasNext());
}
Also used : LcNoDiacriticsType(datawave.data.type.LcNoDiacriticsType) MockMetadataHelper(datawave.query.util.MockMetadataHelper) NoOpType(datawave.data.type.NoOpType) NumberType(datawave.data.type.NumberType) LcNoDiacriticsType(datawave.data.type.LcNoDiacriticsType) Type(datawave.data.type.Type) NumberType(datawave.data.type.NumberType) ASTJexlScript(org.apache.commons.jexl2.parser.ASTJexlScript) ScannerFactory(datawave.query.tables.ScannerFactory) Test(org.junit.Test)

Example 35 with ScannerFactory

use of datawave.query.tables.ScannerFactory in project datawave by NationalSecurityAgency.

the class RangeStreamTest method testNegatedIndexWithResultsStandaloneNot.

@Test
public void testNegatedIndexWithResultsStandaloneNot() throws Exception {
    String originalQuery = "FOO == 'bag' && !(FOO == 'bar')";
    ASTJexlScript script = JexlASTHelper.parseJexlQuery(originalQuery);
    config.setBeginDate(new Date(0));
    config.setEndDate(new Date(System.currentTimeMillis()));
    Multimap<String, Type<?>> dataTypes = HashMultimap.create();
    dataTypes.putAll("FOO", Sets.newHashSet(new LcNoDiacriticsType()));
    config.setQueryFieldsDatatypes(dataTypes);
    config.setIndexedFields(dataTypes);
    MockMetadataHelper helper = new MockMetadataHelper();
    helper.setIndexedFields(dataTypes.keySet());
    Range range1 = makeTestRange("20190314", "datatype1\u0000234");
    Range range2 = makeTestRange("20190314", "datatype1\u0000345");
    Set<Range> expectedRanges = Sets.newHashSet(range1, range2);
    RangeStream rangeStream = new RangeStream(config, new ScannerFactory(config.getConnector()), helper).setLimitScanners(true);
    for (QueryPlan queryPlan : rangeStream.streamPlans(script)) {
        for (Range range : queryPlan.getRanges()) {
            assertTrue("Tried to remove unexpected range " + range.toString() + " from expected ranges: " + expectedRanges.toString(), expectedRanges.remove(range));
        }
    }
    assertTrue("Expected ranges not found in query plan: " + expectedRanges.toString(), expectedRanges.isEmpty());
}
Also used : LcNoDiacriticsType(datawave.data.type.LcNoDiacriticsType) MockMetadataHelper(datawave.query.util.MockMetadataHelper) NoOpType(datawave.data.type.NoOpType) NumberType(datawave.data.type.NumberType) LcNoDiacriticsType(datawave.data.type.LcNoDiacriticsType) Type(datawave.data.type.Type) ASTJexlScript(org.apache.commons.jexl2.parser.ASTJexlScript) RangeFactoryForTests.makeTestRange(datawave.common.test.utils.query.RangeFactoryForTests.makeTestRange) Range(org.apache.accumulo.core.data.Range) ScannerFactory(datawave.query.tables.ScannerFactory) QueryPlan(datawave.query.planner.QueryPlan) Test(org.junit.Test)

Aggregations

ScannerFactory (datawave.query.tables.ScannerFactory)40 Type (datawave.data.type.Type)36 ASTJexlScript (org.apache.commons.jexl2.parser.ASTJexlScript)35 LcNoDiacriticsType (datawave.data.type.LcNoDiacriticsType)34 MockMetadataHelper (datawave.query.util.MockMetadataHelper)34 NoOpType (datawave.data.type.NoOpType)33 NumberType (datawave.data.type.NumberType)33 Test (org.junit.Test)32 Range (org.apache.accumulo.core.data.Range)26 RangeFactoryForTests.makeTestRange (datawave.common.test.utils.query.RangeFactoryForTests.makeTestRange)25 QueryPlan (datawave.query.planner.QueryPlan)25 SimpleDateFormat (java.text.SimpleDateFormat)9 ShardQueryConfiguration (datawave.query.config.ShardQueryConfiguration)2 MetadataHelper (datawave.query.util.MetadataHelper)2 Stopwatch (com.google.common.base.Stopwatch)1 ContentQueryConfiguration (datawave.query.config.ContentQueryConfiguration)1 EdgeQueryConfiguration (datawave.query.config.EdgeQueryConfiguration)1 QueryValues (datawave.query.discovery.FindLiteralsAndPatternsVisitor.QueryValues)1 RangeStream (datawave.query.index.lookup.RangeStream)1 LuceneToJexlQueryParser (datawave.query.language.parser.jexl.LuceneToJexlQueryParser)1