use of org.apache.druid.collections.bitmap.ImmutableBitmap in project druid by druid-io.
the class OrFilter method makeMatcher.
@Override
public ValueMatcher makeMatcher(BitmapIndexSelector selector, ColumnSelectorFactory columnSelectorFactory, RowOffsetMatcherFactory rowOffsetMatcherFactory) {
final List<ValueMatcher> matchers = new ArrayList<>();
final List<ImmutableBitmap> bitmaps = new ArrayList<>();
for (Filter filter : filters) {
if (filter.supportsBitmapIndex(selector)) {
bitmaps.add(filter.getBitmapIndex(selector));
} else {
ValueMatcher matcher = filter.makeMatcher(columnSelectorFactory);
matchers.add(matcher);
}
}
if (bitmaps.size() > 0) {
ImmutableBitmap combinedBitmap = selector.getBitmapFactory().union(bitmaps);
ValueMatcher offsetMatcher = rowOffsetMatcherFactory.makeRowOffsetMatcher(combinedBitmap);
matchers.add(0, offsetMatcher);
}
return makeMatcher(matchers.toArray(BooleanFilter.EMPTY_VALUE_MATCHER_ARRAY));
}
use of org.apache.druid.collections.bitmap.ImmutableBitmap in project druid by druid-io.
the class IndexMergerNullHandlingTest method testStringColumnNullHandling.
@Test
public void testStringColumnNullHandling() throws Exception {
List<Map<String, Object>> nonNullFlavors = new ArrayList<>();
nonNullFlavors.add(ImmutableMap.of("d", "a"));
nonNullFlavors.add(ImmutableMap.of("d", ImmutableList.of("a", "b")));
List<Map<String, Object>> nullFlavors = new ArrayList<>();
Map<String, Object> mMissing = ImmutableMap.of();
Map<String, Object> mEmptyList = ImmutableMap.of("d", Collections.emptyList());
Map<String, Object> mNull = new HashMap<>();
mNull.put("d", null);
Map<String, Object> mEmptyString = ImmutableMap.of("d", "");
Map<String, Object> mListOfNull = ImmutableMap.of("d", Collections.singletonList(null));
Map<String, Object> mListOfEmptyString = ImmutableMap.of("d", Collections.singletonList(""));
nullFlavors.add(mMissing);
nullFlavors.add(mEmptyList);
nullFlavors.add(mNull);
nullFlavors.add(mListOfNull);
if (NullHandling.replaceWithDefault()) {
nullFlavors.add(mEmptyString);
nullFlavors.add(mListOfEmptyString);
} else {
nonNullFlavors.add(mEmptyString);
nonNullFlavors.add(mListOfEmptyString);
}
Set<Map<String, Object>> allValues = new HashSet<>();
allValues.addAll(nonNullFlavors);
allValues.addAll(nullFlavors);
for (Set<Map<String, Object>> subset : Sets.powerSet(allValues)) {
if (subset.isEmpty()) {
continue;
}
final List<Map<String, Object>> subsetList = new ArrayList<>(subset);
IncrementalIndex toPersist = IncrementalIndexTest.createIndex(new AggregatorFactory[] {});
for (Map<String, Object> m : subsetList) {
toPersist.add(new MapBasedInputRow(0L, ImmutableList.of("d"), m));
}
final File tempDir = temporaryFolder.newFolder();
try (QueryableIndex index = indexIO.loadIndex(indexMerger.persist(toPersist, tempDir, indexSpec, null))) {
final ColumnHolder columnHolder = index.getColumnHolder("d");
if (nullFlavors.containsAll(subsetList)) {
// all null -> should be missing
Assert.assertNull(subsetList.toString(), columnHolder);
} else {
Assert.assertNotNull(subsetList.toString(), columnHolder);
// The column has multiple values if there are any lists with > 1 element in the input set.
final boolean hasMultipleValues = subsetList.stream().anyMatch(m -> m.get("d") instanceof List && (((List) m.get("d")).size() > 1));
// Compute all unique values, the same way that IndexMerger is expected to do it.
final Set<String> uniqueValues = new HashSet<>();
for (Map<String, Object> m : subsetList) {
final List<String> dValues = normalize(m.get("d"), hasMultipleValues);
uniqueValues.addAll(dValues);
if (nullFlavors.contains(m)) {
uniqueValues.add(null);
}
}
try (final DictionaryEncodedColumn<String> dictionaryColumn = (DictionaryEncodedColumn<String>) columnHolder.getColumn()) {
// Verify unique values against the dictionary.
Assert.assertEquals(subsetList.toString(), uniqueValues.stream().sorted(Comparators.naturalNullsFirst()).collect(Collectors.toList()), IntStream.range(0, dictionaryColumn.getCardinality()).mapToObj(dictionaryColumn::lookupName).collect(Collectors.toList()));
Assert.assertEquals(subsetList.toString(), hasMultipleValues, dictionaryColumn.hasMultipleValues());
Assert.assertEquals(subsetList.toString(), uniqueValues.size(), dictionaryColumn.getCardinality());
// Verify the expected set of rows was indexed, ignoring order.
Assert.assertEquals(subsetList.toString(), ImmutableMultiset.copyOf(subsetList.stream().map(m -> normalize(m.get("d"), hasMultipleValues)).distinct().collect(Collectors.toList())), ImmutableMultiset.copyOf(IntStream.range(0, index.getNumRows()).mapToObj(rowNumber -> getRow(dictionaryColumn, rowNumber)).distinct().collect(Collectors.toList())));
// Verify that the bitmap index for null is correct.
final BitmapIndex bitmapIndex = columnHolder.getBitmapIndex();
// Read through the column to find all the rows that should match null.
final List<Integer> expectedNullRows = new ArrayList<>();
for (int i = 0; i < index.getNumRows(); i++) {
final List<String> row = getRow(dictionaryColumn, i);
if (row.isEmpty() || row.stream().anyMatch(NullHandling::isNullOrEquivalent)) {
expectedNullRows.add(i);
}
}
Assert.assertEquals(subsetList.toString(), expectedNullRows.size() > 0, bitmapIndex.hasNulls());
if (expectedNullRows.size() > 0) {
Assert.assertEquals(subsetList.toString(), 0, bitmapIndex.getIndex(null));
final ImmutableBitmap nullBitmap = bitmapIndex.getBitmap(bitmapIndex.getIndex(null));
final List<Integer> actualNullRows = new ArrayList<>();
final IntIterator iterator = nullBitmap.iterator();
while (iterator.hasNext()) {
actualNullRows.add(iterator.next());
}
Assert.assertEquals(subsetList.toString(), expectedNullRows, actualNullRows);
} else {
Assert.assertEquals(-1, bitmapIndex.getIndex(null));
}
}
}
}
}
}
use of org.apache.druid.collections.bitmap.ImmutableBitmap in project druid by druid-io.
the class ColumnSelectorBitmapIndexSelectorTest method setup.
@Before
public void setup() {
bitmapFactory = EasyMock.createMock(BitmapFactory.class);
virtualColumns = EasyMock.createMock(VirtualColumns.class);
index = EasyMock.createMock(ColumnSelector.class);
bitmapIndexSelector = new ColumnSelectorBitmapIndexSelector(bitmapFactory, virtualColumns, index);
EasyMock.expect(virtualColumns.getVirtualColumn(STRING_DICTIONARY_COLUMN_NAME)).andReturn(null).anyTimes();
EasyMock.expect(virtualColumns.getVirtualColumn(NON_STRING_DICTIONARY_COLUMN_NAME)).andReturn(null).anyTimes();
ColumnHolder holder = EasyMock.createMock(ColumnHolder.class);
EasyMock.expect(index.getColumnHolder(STRING_DICTIONARY_COLUMN_NAME)).andReturn(holder).anyTimes();
StringDictionaryEncodedColumn stringColumn = EasyMock.createMock(StringDictionaryEncodedColumn.class);
EasyMock.expect(holder.getCapabilities()).andReturn(ColumnCapabilitiesImpl.createDefault().setType(ColumnType.STRING).setDictionaryEncoded(true).setDictionaryValuesUnique(true).setDictionaryValuesSorted(true).setHasBitmapIndexes(true)).anyTimes();
EasyMock.expect(holder.getColumn()).andReturn(stringColumn).anyTimes();
BitmapIndex someIndex = EasyMock.createMock(BitmapIndex.class);
EasyMock.expect(holder.getBitmapIndex()).andReturn(someIndex).anyTimes();
ImmutableBitmap someBitmap = EasyMock.createMock(ImmutableBitmap.class);
EasyMock.expect(someIndex.getIndex("foo")).andReturn(0).anyTimes();
EasyMock.expect(someIndex.getBitmap(0)).andReturn(someBitmap).anyTimes();
ColumnHolder nonStringHolder = EasyMock.createMock(ColumnHolder.class);
EasyMock.expect(index.getColumnHolder(NON_STRING_DICTIONARY_COLUMN_NAME)).andReturn(nonStringHolder).anyTimes();
EasyMock.expect(nonStringHolder.getCapabilities()).andReturn(ColumnCapabilitiesImpl.createDefault().setType(ColumnType.ofComplex("testBlob")).setDictionaryEncoded(true).setDictionaryValuesUnique(true).setDictionaryValuesSorted(true).setHasBitmapIndexes(true).setFilterable(true)).anyTimes();
EasyMock.replay(bitmapFactory, virtualColumns, index, holder, stringColumn, nonStringHolder, someIndex, someBitmap);
}
use of org.apache.druid.collections.bitmap.ImmutableBitmap in project druid by druid-io.
the class ColumnSelectorBitmapIndexSelectorTest method testNonStringDictionaryDoNotUseIndex.
@Test
public void testNonStringDictionaryDoNotUseIndex() {
BitmapIndex bitmapIndex = bitmapIndexSelector.getBitmapIndex(NON_STRING_DICTIONARY_COLUMN_NAME);
Assert.assertNull(bitmapIndex);
Indexed<String> vals = bitmapIndexSelector.getDimensionValues(NON_STRING_DICTIONARY_COLUMN_NAME);
Assert.assertNull(vals);
ImmutableBitmap valueIndex = bitmapIndexSelector.getBitmapIndex(NON_STRING_DICTIONARY_COLUMN_NAME, "foo");
Assert.assertNull(valueIndex);
EasyMock.verify(bitmapFactory, virtualColumns, index);
}
use of org.apache.druid.collections.bitmap.ImmutableBitmap in project druid by druid-io.
the class VectorSelectorUtilsTest method testConciseImmutableNullVector.
@Test
public void testConciseImmutableNullVector() {
final WrappedConciseBitmap bitmap = new WrappedConciseBitmap();
populate(bitmap, NULLS);
final ImmutableBitmap immutable = new WrappedImmutableConciseBitmap(ImmutableConciseSet.newImmutableFromMutable(bitmap.getBitmap()));
assertNullVector(immutable, NULLS);
final WrappedConciseBitmap bitmap2 = new WrappedConciseBitmap();
populate(bitmap2, NULLS_PATTERN);
final ImmutableBitmap immutable2 = new WrappedImmutableConciseBitmap(ImmutableConciseSet.newImmutableFromMutable(bitmap2.getBitmap()));
assertNullVector(immutable2, NULLS_PATTERN);
}
Aggregations