use of datawave.query.tables.ScannerFactory in project datawave by NationalSecurityAgency.
the class RangeStreamTest method testIntersection_OfTwoNestedUnions_LeftLowCardTerms_withSeek.
// (A || B) && (C || D)
@Test
public void testIntersection_OfTwoNestedUnions_LeftLowCardTerms_withSeek() throws Exception {
String originalQuery = "(FOO == 'low_card' || 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("20190312_1", "datatype1\u0000a.b.c");
Range range3 = makeTestRange("20190314_22", "datatype1\u0000a.b.c");
Range range4 = makeTestRange("20190315_33", "datatype1\u0000a.b.c");
Range range5 = makeTestRange("20190315_49", "datatype1\u0000a.b.c");
Range range6 = makeTestRange("20190317_1", "datatype1\u0000a.b.c");
Set<Range> expectedRanges = Sets.newHashSet(range1, range2, range3, range4, range5, range6);
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());
}
use of datawave.query.tables.ScannerFactory in project datawave by NationalSecurityAgency.
the class RangeStreamTest method testNonExistentFieldInOr.
@Test
public void testNonExistentFieldInOr() throws Exception {
String originalQuery = "FOO == 'bag' || CANDY_TYPE == 'candy corn'";
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());
RangeStream rangeStream = new RangeStream(config, new ScannerFactory(config.getConnector()), helper).setLimitScanners(true);
rangeStream.streamPlans(script);
// streamPlans(script) to populate the StreamContext.
assertEquals(IndexStream.StreamContext.PRESENT, rangeStream.context());
assertTrue(rangeStream.iterator().hasNext());
}
use of datawave.query.tables.ScannerFactory in project datawave by NationalSecurityAgency.
the class RangeStreamTest method testShardAndDaysHints.
public void testShardAndDaysHints(String originalQuery) throws Exception {
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();
for (String shard : Lists.newArrayList("20190314_0", "20190314_1", "20190314_9")) {
expectedRanges.add(makeTestRange(shard, "datatype1\u0000345"));
expectedRanges.add(makeTestRange(shard, "datatype1\u0000456"));
expectedRanges.add(makeTestRange(shard, "datatype1\u0000567"));
}
for (String shard : Lists.newArrayList("20190314_10", "20190314_100")) {
expectedRanges.add(makeTestRange(shard, "datatype2\u0000345"));
expectedRanges.add(makeTestRange(shard, "datatype2\u0000456"));
expectedRanges.add(makeTestRange(shard, "datatype2\u0000567"));
}
for (QueryPlan queryPlan : new RangeStream(config, new ScannerFactory(config.getConnector()), helper).streamPlans(script)) {
for (Range range : queryPlan.getRanges()) {
assertTrue("Tried to remove unexpected range " + range.toString() + " from expected ranges: " + expectedRanges.toString(), expectedRanges.remove(range));
}
}
assertTrue(expectedRanges.size() + " expected ranges not found in query plan: " + expectedRanges.toString(), expectedRanges.isEmpty());
}
use of datawave.query.tables.ScannerFactory in project datawave by NationalSecurityAgency.
the class RangeStreamTest method testIntersectionOfTwoUnions.
/**
* Interesting case when the right side terms are indexed and the left side terms are not indexed.
*
* @throws Exception
*/
@Test
public void testIntersectionOfTwoUnions() throws Exception {
String originalQuery = "(FOO == 'bag' || FOO == 'bar') && (TACO == 'shell' || TACO == '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()));
config.setQueryFieldsDatatypes(dataTypes);
config.setIndexedFields(dataTypes);
MockMetadataHelper helper = new MockMetadataHelper();
helper.setIndexedFields(dataTypes.keySet());
helper.addFields(ImmutableSet.of("TACO"));
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);
CloseableIterable<QueryPlan> queryPlans = rangeStream.streamPlans(script);
// streamPlans(script) to populate the StreamContext.
assertEquals(IndexStream.StreamContext.PRESENT, rangeStream.context());
for (QueryPlan queryPlan : queryPlans) {
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());
}
use of datawave.query.tables.ScannerFactory in project datawave by NationalSecurityAgency.
the class RangeStreamTest method testAndTwoFieldsIndexed.
@Test
public void testAndTwoFieldsIndexed() throws Exception {
String originalQuery = "(FOO == 'bag' && FOO == '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());
}
Aggregations