@Override
public void hashRow(DimensionSelector dimSelector, Hasher hasher) {
final IndexedInts row = dimSelector.getRow();
final int size = row.size();
// nothing to add to hasher if size == 0, only handle size == 1 and size != 0 cases.
if (size == 1) {
final String value = dimSelector.lookupName(row.get(0));
hasher.putUnencodedChars(nullToSpecial(value));
} else if (size != 0) {
final String[] values = new String[size];
for (int i = 0; i < size; ++i) {
final String value = dimSelector.lookupName(row.get(i));
values[i] = nullToSpecial(value);
}
// Values need to be sorted to ensure consistent multi-value ordering across different
// segments
Arrays.sort(values);
for (int i = 0; i < size; ++i) {
if (i != 0) {
hasher.putChar(CARDINALITY_AGG_SEPARATOR);
}
hasher.putUnencodedChars(values[i]);
}
}
}
@Override
public void hashValues(DimensionSelector dimSelector, HyperLogLogCollector collector) {
for (IntIterator rowIt = dimSelector.getRow().iterator(); rowIt.hasNext(); ) {
int index = rowIt.nextInt();
final String value = dimSelector.lookupName(index);
collector.add(
CardinalityAggregator.hashFn.hashUnencodedChars(nullToSpecial(value)).asBytes());
}
}
@Test
public void testResetSanity() throws IOException {
IncrementalIndex index = indexCreator.createIndex();
DateTime t = DateTime.now();
Interval interval = new Interval(t.minusMinutes(1), t.plusMinutes(1));
index.add(
new MapBasedInputRow(
t.minus(1).getMillis(),
Lists.newArrayList("billy"),
ImmutableMap.<String, Object>of("billy", "hi")));
index.add(
new MapBasedInputRow(
t.minus(1).getMillis(),
Lists.newArrayList("sally"),
ImmutableMap.<String, Object>of("sally", "bo")));
IncrementalIndexStorageAdapter adapter = new IncrementalIndexStorageAdapter(index);
for (boolean descending : Arrays.asList(false, true)) {
Sequence<Cursor> cursorSequence =
adapter.makeCursors(
new SelectorFilter("sally", "bo"), interval, QueryGranularity.NONE, descending);
Cursor cursor =
Sequences.toList(Sequences.limit(cursorSequence, 1), Lists.<Cursor>newArrayList()).get(0);
DimensionSelector dimSelector;
dimSelector = cursor.makeDimensionSelector(new DefaultDimensionSpec("sally", "sally"));
Assert.assertEquals("bo", dimSelector.lookupName(dimSelector.getRow().get(0)));
index.add(
new MapBasedInputRow(
t.minus(1).getMillis(),
Lists.newArrayList("sally"),
ImmutableMap.<String, Object>of("sally", "ah")));
// Cursor reset should not be affected by out of order values
cursor.reset();
dimSelector = cursor.makeDimensionSelector(new DefaultDimensionSpec("sally", "sally"));
Assert.assertEquals("bo", dimSelector.lookupName(dimSelector.getRow().get(0)));
}
}
public BaseArrayProvider(
DimensionSelector dimSelector, TopNQuery query, Capabilities capabilities) {
this.dimSelector = dimSelector;
this.query = query;
this.capabilities = capabilities;
previousStop = null;
ignoreAfterThreshold = false;
ignoreFirstN = 0;
keepOnlyN = dimSelector.getValueCardinality();
}
@Override
public DimensionSelector decorate(final DimensionSelector selector) {
if (selector == null) {
return selector;
}
int count = 0;
final Map<Integer, Integer> forwardMapping = new HashMap<>();
for (int i = 0; i < selector.getValueCardinality(); i++) {
if (compiledRegex.matcher(Strings.nullToEmpty(selector.lookupName(i))).matches()) {
forwardMapping.put(i, count++);
}
}
final int[] reverseMapping = new int[forwardMapping.size()];
for (Map.Entry<Integer, Integer> e : forwardMapping.entrySet()) {
reverseMapping[e.getValue().intValue()] = e.getKey().intValue();
}
return BaseFilteredDimensionSpec.decorate(selector, forwardMapping, reverseMapping);
}
public BaseArrayProvider(
DimensionSelector dimSelector, TopNQuery query, Capabilities capabilities) {
this.dimSelector = dimSelector;
this.query = query;
this.capabilities = capabilities;
previousStop = null;
ignoreAfterThreshold = false;
ignoreFirstN = 0;
keepOnlyN = dimSelector.getValueCardinality();
if (keepOnlyN < 0) {
throw new UnsupportedOperationException("Cannot operate on a dimension with no dictionary");
}
}
protected Pair<Integer, Integer> computeStartEnd(int cardinality) {
int startIndex = ignoreFirstN;
if (previousStop != null) {
int lookupId = dimSelector.lookupId(previousStop) + 1;
if (lookupId < 0) {
lookupId *= -1;
}
if (lookupId > ignoreFirstN + keepOnlyN) {
startIndex = ignoreFirstN + keepOnlyN;
} else {
startIndex = Math.max(lookupId, startIndex);
}
}
int endIndex = Math.min(ignoreFirstN + keepOnlyN, cardinality);
if (ignoreAfterThreshold && query.getDimensionsFilter() == null) {
endIndex = Math.min(endIndex, startIndex + query.getThreshold());
}
return Pair.of(startIndex, endIndex);
}
