Skip to content

Commit

Permalink
Support global ords in top_metrics (#64967)
Browse files Browse the repository at this point in the history
Adds support for fetching `keyword` and `ip` fields in `top_metrics`.
This only works if we can get global ordinals for those fields so we
can't support the entire `keyword` family, but this gets us *somewhere*.

Closes #64774
  • Loading branch information
nik9000 authored Nov 16, 2020
1 parent e4a90d5 commit 450fb55
Show file tree
Hide file tree
Showing 10 changed files with 525 additions and 160 deletions.
Original file line number Diff line number Diff line change
Expand Up @@ -171,7 +171,7 @@ public int getBucketSize() {
*/
@FunctionalInterface
public interface ResultBuilder<T> {
T build(long index, SortValue sortValue);
T build(long index, SortValue sortValue) throws IOException;
}

/**
Expand All @@ -180,7 +180,7 @@ public interface ResultBuilder<T> {
* @param builder builds results. See {@link ExtraData} for how to store
* data along side the sort for this to extract.
*/
public final <T extends Comparable<T>> List<T> getValues(long bucket, ResultBuilder<T> builder) {
public final <T extends Comparable<T>> List<T> getValues(long bucket, ResultBuilder<T> builder) throws IOException {
long rootIndex = bucket * bucketSize;
if (rootIndex >= values().size()) {
// We've never seen this bucket.
Expand All @@ -201,7 +201,7 @@ public final <T extends Comparable<T>> List<T> getValues(long bucket, ResultBuil
* Get the values for a bucket if it has been collected. If it hasn't
* then returns an empty array.
*/
public final List<SortValue> getValues(long bucket) {
public final List<SortValue> getValues(long bucket) throws IOException {
return getValues(bucket, (i, sv) -> sv);
}

Expand Down
91 changes: 89 additions & 2 deletions server/src/main/java/org/elasticsearch/search/sort/SortValue.java
Original file line number Diff line number Diff line change
Expand Up @@ -19,6 +19,8 @@

package org.elasticsearch.search.sort;

import org.apache.lucene.util.BytesRef;
import org.elasticsearch.Version;
import org.elasticsearch.common.io.stream.NamedWriteable;
import org.elasticsearch.common.io.stream.NamedWriteableRegistry;
import org.elasticsearch.common.io.stream.StreamInput;
Expand Down Expand Up @@ -48,13 +50,22 @@ public static SortValue from(long l) {
return new LongSortValue(l);
}

/**
* Get a {@linkplain SortValue} for bytes. Callers should be sure that they
* have a {@link BytesRef#deepCopyOf} of any mutable references.
*/
public static SortValue from(BytesRef bytes) {
return new BytesSortValue(bytes);
}

/**
* Get the list of {@linkplain NamedWriteable}s that this class needs.
*/
public static List<NamedWriteableRegistry.Entry> namedWriteables() {
return Arrays.asList(
new NamedWriteableRegistry.Entry(SortValue.class, DoubleSortValue.NAME, DoubleSortValue::new),
new NamedWriteableRegistry.Entry(SortValue.class, LongSortValue.NAME, LongSortValue::new));
new NamedWriteableRegistry.Entry(SortValue.class, LongSortValue.NAME, LongSortValue::new),
new NamedWriteableRegistry.Entry(SortValue.class, BytesSortValue.NAME, BytesSortValue::new));
}

private SortValue() {
Expand Down Expand Up @@ -200,7 +211,7 @@ private static class LongSortValue extends SortValue {
this.key = key;
}

LongSortValue(StreamInput in) throws IOException {
private LongSortValue(StreamInput in) throws IOException {
key = in.readLong();
}

Expand Down Expand Up @@ -259,4 +270,80 @@ public Number numberValue() {
return key;
}
}

private static class BytesSortValue extends SortValue {
public static final String NAME = "bytes";

private final BytesRef key;

BytesSortValue(BytesRef key) {
this.key = key;
}

private BytesSortValue(StreamInput in) throws IOException {
key = in.readBytesRef();
}

@Override
public void writeTo(StreamOutput out) throws IOException {
if (out.getVersion().before(Version.V_7_11_0)) {
throw new IllegalArgumentException(
"versions of Elasticsearch before 7.11.0 can't handle non-numeric sort values and attempted to send to ["
+ out.getVersion()
+ "]"
);
}
out.writeBytesRef(key);
}

@Override
public String getWriteableName() {
return NAME;
}

@Override
public Object getKey() {
return key;
}

@Override
public String format(DocValueFormat format) {
return format.format(key).toString();
}

@Override
protected XContentBuilder rawToXContent(XContentBuilder builder) throws IOException {
return builder.value(key.utf8ToString());
}

@Override
protected int compareToSameType(SortValue obj) {
BytesSortValue other = (BytesSortValue) obj;
return key.compareTo(other.key);
}

@Override
public boolean equals(Object obj) {
if (obj == null || false == getClass().equals(obj.getClass())) {
return false;
}
BytesSortValue other = (BytesSortValue) obj;
return key.equals(other.key);
}

@Override
public int hashCode() {
return key.hashCode();
}

@Override
public String toString() {
return key.toString();
}

@Override
public Number numberValue() {
throw new UnsupportedOperationException();
}
}
}
Original file line number Diff line number Diff line change
Expand Up @@ -73,7 +73,7 @@ private T build(SortOrder order, int bucketSize, double[] values) {
return build(order, format, bucketSize, BucketedSort.NOOP_EXTRA_DATA, values);
}

public final void testNeverCalled() {
public final void testNeverCalled() throws IOException {
SortOrder order = randomFrom(SortOrder.values());
DocValueFormat format = randomFrom(DocValueFormat.RAW, DocValueFormat.BINARY, DocValueFormat.BOOLEAN);
try (T sort = build(order, format, 1, BucketedSort.NOOP_EXTRA_DATA, new double[] {})) {
Expand Down
Original file line number Diff line number Diff line change
Expand Up @@ -19,14 +19,19 @@

package org.elasticsearch.search.sort;

import org.apache.lucene.document.InetAddressPoint;
import org.apache.lucene.util.BytesRef;
import org.elasticsearch.Version;
import org.elasticsearch.common.Strings;
import org.elasticsearch.common.io.stream.NamedWriteableRegistry;
import org.elasticsearch.common.network.InetAddresses;
import org.elasticsearch.common.time.DateFormatter;
import org.elasticsearch.common.xcontent.ToXContentFragment;
import org.elasticsearch.common.xcontent.XContentBuilder;
import org.elasticsearch.index.mapper.DateFieldMapper;
import org.elasticsearch.search.DocValueFormat;
import org.elasticsearch.test.AbstractNamedWriteableTestCase;
import org.elasticsearch.test.VersionUtils;

import java.io.IOException;
import java.time.ZoneId;
Expand All @@ -51,7 +56,12 @@ protected NamedWriteableRegistry getNamedWriteableRegistry() {

@Override
protected SortValue createTestInstance() {
return randomBoolean() ? SortValue.from(randomDouble()) : SortValue.from(randomLong());
switch (between(0, 2)) {
case 0: return SortValue.from(randomDouble());
case 1: return SortValue.from(randomLong());
case 2: return SortValue.from(new BytesRef(randomAlphaOfLength(5)));
default: throw new AssertionError();
}
}

@Override
Expand Down Expand Up @@ -81,11 +91,23 @@ public void testToXContentLong() {
assertThat(toXContent(SortValue.from(1), STRICT_DATE_TIME), equalTo("{\"test\":\"1970-01-01T00:00:00.001Z\"}"));
}

public void testToXContentBytes() {
assertThat(toXContent(SortValue.from(new BytesRef("cat")), DocValueFormat.RAW), equalTo("{\"test\":\"cat\"}"));
assertThat(
toXContent(SortValue.from(new BytesRef(InetAddressPoint.encode(InetAddresses.forString("127.0.0.1")))), DocValueFormat.IP),
equalTo("{\"test\":\"127.0.0.1\"}")
);
}

public void testCompareDifferentTypes() {
assertThat(SortValue.from(1.0), lessThan(SortValue.from(1)));
assertThat(SortValue.from(Double.MAX_VALUE), lessThan(SortValue.from(Long.MIN_VALUE)));
assertThat(SortValue.from(1), greaterThan(SortValue.from(1.0)));
assertThat(SortValue.from(Long.MIN_VALUE), greaterThan(SortValue.from(Double.MAX_VALUE)));
assertThat(SortValue.from(new BytesRef("cat")), lessThan(SortValue.from(1)));
assertThat(SortValue.from(1), greaterThan(SortValue.from(new BytesRef("cat"))));
assertThat(SortValue.from(new BytesRef("cat")), lessThan(SortValue.from(1.0)));
assertThat(SortValue.from(1.0), greaterThan(SortValue.from(new BytesRef("cat"))));
}

public void testCompareDoubles() {
Expand All @@ -102,6 +124,25 @@ public void testCompareLongs() {
assertThat(SortValue.from(r), greaterThan(SortValue.from(r - 1)));
}

public void testBytes() {
String r = randomAlphaOfLength(5);
assertThat(SortValue.from(new BytesRef(r)), equalTo(SortValue.from(new BytesRef(r))));
assertThat(SortValue.from(new BytesRef(r)), lessThan(SortValue.from(new BytesRef(r + "with_suffix"))));
assertThat(SortValue.from(new BytesRef(r)), greaterThan(SortValue.from(new BytesRef(new byte[] {}))));
}

public void testSerializeBytesToOldVersion() {
SortValue value = SortValue.from(new BytesRef("can't send me!"));
Version version = VersionUtils.randomVersionBetween(random(), Version.V_7_0_0, Version.V_7_10_1);
Exception e = expectThrows(IllegalArgumentException.class, () -> copyInstance(value, version));
assertThat(
e.getMessage(),
equalTo(
"versions of Elasticsearch before 7.11.0 can't handle non-numeric sort values and attempted to send to [" + version + "]"
)
);
}

public String toXContent(SortValue sortValue, DocValueFormat format) {
return Strings.toString(new ToXContentFragment() {
@Override
Expand Down
Original file line number Diff line number Diff line change
Expand Up @@ -114,7 +114,8 @@ public List<AggregationSpec> getAggregations() {
TopMetricsAggregationBuilder.NAME,
TopMetricsAggregationBuilder::new,
usage.track(AnalyticsStatsAction.Item.TOP_METRICS, checkLicense(TopMetricsAggregationBuilder.PARSER)))
.addResultReader(InternalTopMetrics::new),
.addResultReader(InternalTopMetrics::new)
.setAggregatorRegistrar(TopMetricsAggregationBuilder::registerAggregators),
new AggregationSpec(
TTestAggregationBuilder.NAME,
TTestAggregationBuilder::new,
Expand Down
Original file line number Diff line number Diff line change
Expand Up @@ -18,7 +18,10 @@
import org.elasticsearch.search.aggregations.AggregatorFactories.Builder;
import org.elasticsearch.search.aggregations.AggregatorFactory;
import org.elasticsearch.search.aggregations.support.AggregationContext;
import org.elasticsearch.search.aggregations.support.CoreValuesSourceType;
import org.elasticsearch.search.aggregations.support.MultiValuesSourceFieldConfig;
import org.elasticsearch.search.aggregations.support.ValuesSourceRegistry;
import org.elasticsearch.search.aggregations.support.ValuesSourceRegistry.RegistryKey;
import org.elasticsearch.search.sort.SortBuilder;

import java.io.IOException;
Expand All @@ -34,6 +37,28 @@ public class TopMetricsAggregationBuilder extends AbstractAggregationBuilder<Top
public static final String NAME = "top_metrics";
public static final ParseField METRIC_FIELD = new ParseField("metrics");

static final RegistryKey<TopMetricsAggregator.MetricValuesSupplier> REGISTRY_KEY = new RegistryKey<>(
TopMetricsAggregationBuilder.NAME,
TopMetricsAggregator.MetricValuesSupplier.class
);

public static void registerAggregators(ValuesSourceRegistry.Builder registry) {
registry.registerUsage(NAME);
registry.register(REGISTRY_KEY, List.of(CoreValuesSourceType.NUMERIC), TopMetricsAggregator::buildNumericMetricValues, false);
registry.register(
REGISTRY_KEY,
List.of(CoreValuesSourceType.BOOLEAN, CoreValuesSourceType.DATE),
TopMetricsAggregator.LongMetricValues::new,
false
);
registry.register(
REGISTRY_KEY,
List.of(CoreValuesSourceType.BYTES, CoreValuesSourceType.IP),
TopMetricsAggregator.GlobalOrdsValues::new,
false
);
}

/**
* Default to returning only a single top metric.
*/
Expand Down
Loading

0 comments on commit 450fb55

Please sign in to comment.