Skip to content
Draft
Show file tree
Hide file tree
Changes from 1 commit
Commits
File filter

Filter by extension

Filter by extension

Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view

Some generated files are not rendered by default. Learn more about how customized files appear on GitHub.

Original file line number Diff line number Diff line change
Expand Up @@ -40,11 +40,15 @@ $endif$

@Override
$if(BytesRef)$
public BytesRef getBytesRef(int position, BytesRef ignore) {
public BytesRef getBytesRef(int position, BytesRef scratch) {
scratch.bytes = value.bytes;
Copy link
Member Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

I will open a separate PR for this fix.

scratch.offset = value.offset;
scratch.length = value.length;
return scratch;
$else$
public $type$ get$Type$(int position) {
$endif$
return value;
$endif$
}

@Override
Expand Down
402 changes: 201 additions & 201 deletions x-pack/plugin/esql/qa/testFixtures/src/main/resources/data/k8s.csv

Large diffs are not rendered by default.

Original file line number Diff line number Diff line change
Expand Up @@ -18,6 +18,10 @@
}
}
},
"region": {
"type": "keyword",
"time_series_dimension": true
},
"event": {
"type": "keyword"
},
Expand Down
Original file line number Diff line number Diff line change
Expand Up @@ -185,7 +185,7 @@ events:double | pod:keyword | time_bucket:datetime
15.746543778801843 | one | 2024-05-10T00:00:00.000Z
15.396284829721363 | three | 2024-05-10T00:00:00.000Z
15.045454545454547 | three | 2024-05-10T00:10:00.000Z
14.199942045783832 | two | 2024-05-10T00:00:00.000Z
14.19994204578383 | two | 2024-05-10T00:00:00.000Z
11.5 | two | 2024-05-10T00:20:00.000Z
;

Expand Down
Original file line number Diff line number Diff line change
Expand Up @@ -51,7 +51,7 @@ TS k8s

irate_cost:double | time_bucket:datetime
null | 2024-05-10T00:01:00.000Z
7.836832264957264 | 2024-05-10T00:09:00.000Z
7.836832264957265 | 2024-05-10T00:09:00.000Z
3.590324074074074 | 2024-05-10T00:17:00.000Z
2.6708333333333334 | 2024-05-10T00:02:00.000Z
2.2916666666666665 | 2024-05-10T00:08:00.000Z
Expand Down
Original file line number Diff line number Diff line change
Expand Up @@ -223,3 +223,22 @@ TS k8s

;

sum_rate_per_region
required_capability: ts_command_v0
required_capability: pack_dimensions_in_ts
TS k8s
| STATS rate_bytes_in=avg(rate(network.total_bytes_in)) BY cluster, region, time_bucket = bucket(@timestamp,5minute)
| SORT rate_bytes_in DESC, time_bucket, cluster | LIMIT 10;

rate_bytes_in:double | cluster:keyword | region:keyword | time_bucket:datetime
15.025267167998313 | qa | null | 2024-05-10T00:15:00.000Z
13.638384356589611 | qa | null | 2024-05-10T00:05:00.000Z
11.761724575728252 | prod | [eu, us] | 2024-05-10T00:15:00.000Z
7.453275209904956 | qa | null | 2024-05-10T00:10:00.000Z
7.307225056633641 | staging | us | 2024-05-10T00:05:00.000Z
7.203958127639015 | prod | [eu, us] | 2024-05-10T00:10:00.000Z
6.34494062999877 | staging | us | 2024-05-10T00:10:00.000Z
5.700488689624205 | prod | [eu, us] | 2024-05-10T00:20:00.000Z
5.4539153439153445 | prod | [eu, us] | 2024-05-10T00:00:00.000Z
Copy link
Member Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

We return arrays for grouping dimensions.

5.241187469367376 | staging | us | 2024-05-10T00:00:00.000Z
;
Original file line number Diff line number Diff line change
Expand Up @@ -4742,7 +4742,7 @@ FROM k8s
| RENAME network.bytes_in AS language_code
| WHERE language_code < 10
| LOOKUP JOIN languages_lookup ON language_code
| DROP network*, event*
| DROP network*, event*, region
| SORT language_code, @timestamp
;

Expand Down
Original file line number Diff line number Diff line change
Expand Up @@ -1499,7 +1499,14 @@ public enum Cap {
/**
* Support for dots in FUSE attributes
*/
DOTS_IN_FUSE;
DOTS_IN_FUSE,

/**
* Pack dimension values in TS command
*/
PACK_DIMENSIONS_IN_TS

;

private final boolean enabled;

Expand Down
Original file line number Diff line number Diff line change
@@ -0,0 +1,272 @@
/*
* Copyright Elasticsearch B.V. and/or licensed to Elasticsearch B.V. under one
* or more contributor license agreements. Licensed under the Elastic License
* 2.0; you may not use this file except in compliance with the Elastic License
* 2.0.
*/

package org.elasticsearch.xpack.esql.expression.function.scalar.internal;

import org.apache.lucene.util.BytesRef;
import org.elasticsearch.compute.data.BooleanBlock;
import org.elasticsearch.compute.data.BytesRefBlock;
import org.elasticsearch.compute.data.BytesRefVector;
import org.elasticsearch.compute.data.IntBlock;
import org.elasticsearch.compute.data.LongBlock;
import org.elasticsearch.compute.data.OrdinalBytesRefVector;
import org.elasticsearch.compute.operator.BreakingBytesRefBuilder;
import org.elasticsearch.compute.operator.DriverContext;
import org.elasticsearch.compute.operator.topn.TopNEncoder;

final class InternalPacks {
private static final TopNEncoder ENCODER = TopNEncoder.DEFAULT_UNSORTABLE;
public static final int INITIAL_SIZE_IN_BYTES = 6 * 1024;

static int estimateForBytesBuilder(int positionCount) {
// allocate at least one page for the bytes block builder to avoid copying during resizing
return Math.max(INITIAL_SIZE_IN_BYTES, positionCount);
}

static BytesRefBlock packBytesValues(DriverContext driverContext, BytesRefBlock raw) {
Copy link
Member Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

This is where we encode multi-valued block into a single value.

BytesRefVector vector = raw.asVector();
if (vector != null) {
OrdinalBytesRefVector ordinals = vector.asOrdinals();
if (ordinals != null) {
var encoded = packBytesVector(driverContext, ordinals.getDictionaryVector());
ordinals.getOrdinalsVector().incRef();
return new OrdinalBytesRefVector(ordinals.getOrdinalsVector(), encoded).asBlock();
} else {
return packBytesVector(driverContext, vector).asBlock();
}
}
int positionCount = raw.getPositionCount();
try (
var builder = driverContext.blockFactory().newBytesRefBlockBuilder(estimateForBytesBuilder(positionCount));
var work = new BreakingBytesRefBuilder(driverContext.breaker(), "pack_dimensions", 1024)
) {
BytesRef scratch = new BytesRef();
for (int p = 0; p < positionCount; p++) {
int valueCount = raw.getValueCount(p);
if (valueCount == 0) {
builder.appendNull();
continue;
}
work.clear();
int first = raw.getFirstValueIndex(p);
int end = first + valueCount;
for (int i = first; i < end; i++) {
raw.getBytesRef(i, scratch);
ENCODER.encodeBytesRef(scratch, work);
}
builder.appendBytesRef(work.bytesRefView());
}
return builder.build();
}
}

static BytesRefVector packBytesVector(DriverContext driverContext, BytesRefVector encode) {
int positionCount = encode.getPositionCount();
try (
var builder = driverContext.blockFactory().newBytesRefVectorBuilder(estimateForBytesBuilder(positionCount));
var work = new BreakingBytesRefBuilder(driverContext.breaker(), "pack_values", 1024)
) {
BytesRef scratch = new BytesRef();
for (int p = 0; p < positionCount; p++) {
ENCODER.encodeBytesRef(encode.getBytesRef(p, scratch), work);
builder.appendBytesRef(work.bytesRefView());
work.clear();
}
return builder.build();
}
}

static BytesRefBlock unpackBytesValues(DriverContext driverContext, BytesRefBlock encoded) {
Copy link
Member Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

And decode here.

int positionCount = encoded.getPositionCount();
try (var builder = driverContext.blockFactory().newBytesRefBlockBuilder(estimateForBytesBuilder(positionCount));) {
BytesRef inScratch = new BytesRef();
BytesRef outScratch = new BytesRef();
for (int p = 0; p < positionCount; p++) {
if (encoded.isNull(p)) {
builder.appendNull();
continue;
}
BytesRef row = encoded.getBytesRef(p, inScratch);
var v = ENCODER.decodeBytesRef(row, outScratch);
if (row.length == 0) {
builder.appendBytesRef(v);
} else {
builder.beginPositionEntry();
builder.appendBytesRef(v);
while (row.length > 0) {
v = ENCODER.decodeBytesRef(row, outScratch);
builder.appendBytesRef(v);
}
builder.endPositionEntry();
}
}
return builder.build();
}
}

static BytesRefBlock packLongValues(DriverContext driverContext, LongBlock raw) {
int positionCount = raw.getPositionCount();
try (
var builder = driverContext.blockFactory().newBytesRefBlockBuilder(estimateForBytesBuilder(positionCount));
var work = new BreakingBytesRefBuilder(driverContext.breaker(), "pack_values", 32)
) {
for (int p = 0; p < positionCount; p++) {
int valueCount = raw.getValueCount(p);
if (valueCount == 0) {
builder.appendNull();
continue;
}
work.clear();
int first = raw.getFirstValueIndex(p);
if (valueCount == 1) {
ENCODER.encodeLong(raw.getLong(first), work);
} else {
int end = first + valueCount;
for (int i = first; i < end; i++) {
ENCODER.encodeLong(raw.getLong(i), work);
}
}
builder.appendBytesRef(work.bytesRefView());
}
return builder.build();
}
}

static LongBlock unpackLongValues(DriverContext driverContext, BytesRefBlock encoded) {
int positionCount = encoded.getPositionCount();
try (var builder = driverContext.blockFactory().newLongBlockBuilder(positionCount)) {
BytesRef inScratch = new BytesRef();
for (int p = 0; p < positionCount; p++) {
if (encoded.isNull(p)) {
builder.appendNull();
continue;
}
BytesRef row = encoded.getBytesRef(p, inScratch);
var v = ENCODER.decodeLong(row);
if (row.length == 0) {
builder.appendLong(v);
} else {
builder.beginPositionEntry();
builder.appendLong(v);
while (row.length > 0) {
builder.appendLong(ENCODER.decodeLong(row));
}
builder.endPositionEntry();
}
}
return builder.build();
}
}

static BytesRefBlock packIntValues(DriverContext driverContext, IntBlock raw) {
int positionCount = raw.getPositionCount();
try (
var builder = driverContext.blockFactory().newBytesRefBlockBuilder(estimateForBytesBuilder(positionCount));
var work = new BreakingBytesRefBuilder(driverContext.breaker(), "pack_values", 32)
) {
for (int p = 0; p < positionCount; p++) {
int valueCount = raw.getValueCount(p);
if (valueCount == 0) {
builder.appendNull();
continue;
}
work.clear();
int first = raw.getFirstValueIndex(p);
if (valueCount == 1) {
ENCODER.encodeInt(raw.getInt(first), work);
} else {
int end = first + valueCount;
for (int i = first; i < end; i++) {
ENCODER.encodeInt(raw.getInt(i), work);
}
}
builder.appendBytesRef(work.bytesRefView());
}
return builder.build();
}
}

static IntBlock unpackIntValues(DriverContext driverContext, BytesRefBlock encoded) {
int positionCount = encoded.getPositionCount();
try (IntBlock.Builder builder = driverContext.blockFactory().newIntBlockBuilder(positionCount)) {
BytesRef inScratch = new BytesRef();
for (int p = 0; p < positionCount; p++) {
if (encoded.isNull(p)) {
builder.appendNull();
continue;
}
BytesRef row = encoded.getBytesRef(p, inScratch);
int v = ENCODER.decodeInt(row);
if (row.length == 0) {
builder.appendInt(v);
} else {
builder.beginPositionEntry();
builder.appendInt(v);
while (row.length > 0) {
builder.appendInt(ENCODER.decodeInt(row));
}
builder.endPositionEntry();
}
}
return builder.build();
}
}

static BytesRefBlock packBooleanValues(DriverContext driverContext, BooleanBlock raw) {
int positionCount = raw.getPositionCount();
try (
var builder = driverContext.blockFactory().newBytesRefBlockBuilder(positionCount);
var work = new BreakingBytesRefBuilder(driverContext.breaker(), "pack_values", 32)
) {
for (int p = 0; p < positionCount; p++) {
work.clear();
int valueCount = raw.getValueCount(p);
if (valueCount == 0) {
builder.appendNull();
continue;
}
int first = raw.getFirstValueIndex(p);
if (valueCount == 1) {
ENCODER.encodeBoolean(raw.getBoolean(first), work);
} else {
int end = first + valueCount;
for (int i = first; i < end; i++) {
ENCODER.encodeBoolean(raw.getBoolean(i), work);
}
}
builder.appendBytesRef(work.bytesRefView());
}
return builder.build();
}
}

static BooleanBlock unpackBooleanValues(DriverContext driverContext, BytesRefBlock encoded) {
int positionCount = encoded.getPositionCount();
try (var builder = driverContext.blockFactory().newBooleanBlockBuilder(positionCount)) {
BytesRef inScratch = new BytesRef();
for (int p = 0; p < positionCount; p++) {
if (encoded.isNull(p)) {
builder.appendNull();
continue;
}
BytesRef row = encoded.getBytesRef(p, inScratch);
boolean v = ENCODER.decodeBoolean(row);
if (row.length == 0) {
builder.appendBoolean(v);
} else {
builder.beginPositionEntry();
builder.appendBoolean(v);
while (row.length > 0) {
builder.appendBoolean(ENCODER.decodeBoolean(row));
}
builder.endPositionEntry();
}
}
return builder.build();
}
}
}
Loading