From a174cc029ea3d65216b9153ef87328bc303b6546 Mon Sep 17 00:00:00 2001
From: Alan Protasio <alanprot@gmail.com>
Date: Wed, 17 May 2023 23:39:05 -0700
Subject: [PATCH] Optimizing "grafana generated" regex matchers (#6376)

* Opmizing Group Regex

Signed-off-by: Alan Protasio <alanprot@gmail.com>

* fixing native histogram tests

Signed-off-by: Alan Protasio <alanprot@gmail.com>

---------

Signed-off-by: Alan Protasio <alanprot@gmail.com>
---
 pkg/store/bucket_test.go             | 45 ++++++++++++++++++++--------
 pkg/store/opts.go                    |  9 +++++-
 pkg/store/opts_test.go               | 13 ++++++++
 pkg/store/storepb/testutil/series.go |  4 +--
 4 files changed, 56 insertions(+), 15 deletions(-)

diff --git a/pkg/store/bucket_test.go b/pkg/store/bucket_test.go
index 90ba08feef6..2f1eb8ef938 100644
--- a/pkg/store/bucket_test.go
+++ b/pkg/store/bucket_test.go
@@ -1384,19 +1384,40 @@ func benchBucketSeries(t testutil.TB, sampleType chunkenc.ValueType, skipChunk b
 			seriesCut = expectedSamples / samplesPerSeriesPerBlock
 		}
 
-		bCases = append(bCases, &storetestutil.SeriesCase{
-			Name: fmt.Sprintf("%dof%d", expectedSamples, totalSeries*samplesPerSeries),
-			Req: &storepb.SeriesRequest{
-				MinTime: 0,
-				MaxTime: int64(expectedSamples) - 1,
-				Matchers: []storepb.LabelMatcher{
-					{Type: storepb.LabelMatcher_EQ, Name: "foo", Value: "bar"},
+		matchersCase := []*labels.Matcher{
+			labels.MustNewMatcher(labels.MatchEqual, "foo", "bar"),
+			labels.MustNewMatcher(labels.MatchNotEqual, "foo", "bar"),
+			labels.MustNewMatcher(labels.MatchRegexp, "j", "(0|1)"),
+			labels.MustNewMatcher(labels.MatchRegexp, "j", "0|1"),
+			labels.MustNewMatcher(labels.MatchNotRegexp, "j", "(0|1)"),
+			labels.MustNewMatcher(labels.MatchNotRegexp, "j", "0|1"),
+		}
+
+		for _, lm := range matchersCase {
+			var expectedSeries []*storepb.Series
+			m, err := storepb.PromMatchersToMatchers(lm)
+			testutil.Ok(t, err)
+
+			// seriesCut does not cut chunks properly, but those are assured against for non benchmarks only, where we use 100% case only.
+			for _, s := range series[:seriesCut] {
+				for _, label := range s.Labels {
+					if label.Name == lm.Name && lm.Matches(label.Value) {
+						expectedSeries = append(expectedSeries, s)
+						break
+					}
+				}
+			}
+			bCases = append(bCases, &storetestutil.SeriesCase{
+				Name: fmt.Sprintf("%dof%d[%s]", expectedSamples, totalSeries*samplesPerSeries, lm.String()),
+				Req: &storepb.SeriesRequest{
+					MinTime:    0,
+					MaxTime:    int64(expectedSamples) - 1,
+					Matchers:   m,
+					SkipChunks: skipChunk,
 				},
-				SkipChunks: skipChunk,
-			},
-			// This does not cut chunks properly, but those are assured against for non benchmarks only, where we use 100% case only.
-			ExpectedSeries: series[:seriesCut],
-		})
+				ExpectedSeries: expectedSeries,
+			})
+		}
 	}
 	storetestutil.TestServerSeries(t, st, bCases...)
 
diff --git a/pkg/store/opts.go b/pkg/store/opts.go
index e95a623a706..e7d2dd83669 100644
--- a/pkg/store/opts.go
+++ b/pkg/store/opts.go
@@ -27,7 +27,14 @@ func init() {
 func findSetMatches(pattern string) []string {
 	escaped := false
 	sets := []*strings.Builder{{}}
-	for i := 0; i < len(pattern); i++ {
+	init := 0
+	end := len(pattern)
+	// If the regex is wrapped in a group we can remove the first and last parentheses
+	if pattern[init] == '(' && pattern[end-1] == ')' {
+		init++
+		end--
+	}
+	for i := init; i < end; i++ {
 		if escaped {
 			switch {
 			case isRegexMetaCharacter(pattern[i]):
diff --git a/pkg/store/opts_test.go b/pkg/store/opts_test.go
index 50d629f4cfd..414b1cc569d 100644
--- a/pkg/store/opts_test.go
+++ b/pkg/store/opts_test.go
@@ -24,6 +24,15 @@ func TestFindSetMatches(t *testing.T) {
 				"baz",
 			},
 		},
+		// Simple sets with group wrapper.
+		{
+			pattern: "(foo|bar|baz)",
+			exp: []string{
+				"foo",
+				"bar",
+				"baz",
+			},
+		},
 		// Simple sets containing escaped characters.
 		{
 			pattern: "fo\\.o|bar\\?|\\^baz",
@@ -38,6 +47,10 @@ func TestFindSetMatches(t *testing.T) {
 			pattern: "fo.o|bar?|^baz",
 			exp:     nil,
 		},
+		{
+			pattern: "(fool|bar)|(baz)",
+			exp:     nil,
+		},
 		{
 			pattern: "foo\\|bar\\|baz",
 			exp: []string{
diff --git a/pkg/store/storepb/testutil/series.go b/pkg/store/storepb/testutil/series.go
index aa70498aea9..047f942c2be 100644
--- a/pkg/store/storepb/testutil/series.go
+++ b/pkg/store/storepb/testutil/series.go
@@ -167,7 +167,7 @@ func ReadSeriesFromBlock(t testing.TB, h tsdb.BlockReader, extLabels labels.Labe
 func appendFloatSamples(t testing.TB, app storage.Appender, tsLabel int, opts HeadGenOptions) {
 	ref, err := app.Append(
 		0,
-		labels.FromStrings("foo", "bar", "i", fmt.Sprintf("%07d%s", tsLabel, LabelLongSuffix)),
+		labels.FromStrings("foo", "bar", "i", fmt.Sprintf("%07d%s", tsLabel, LabelLongSuffix), "j", fmt.Sprintf("%v", tsLabel)),
 		int64(tsLabel)*opts.ScrapeInterval.Milliseconds(),
 		opts.Random.Float64(),
 	)
@@ -195,7 +195,7 @@ func appendHistogramSamples(t testing.TB, app storage.Appender, tsLabel int, opt
 
 	ref, err := app.AppendHistogram(
 		0,
-		labels.FromStrings("foo", "bar", "i", fmt.Sprintf("%07d%s", tsLabel, LabelLongSuffix)),
+		labels.FromStrings("foo", "bar", "i", fmt.Sprintf("%07d%s", tsLabel, LabelLongSuffix), "j", fmt.Sprintf("%v", tsLabel)),
 		int64(tsLabel)*opts.ScrapeInterval.Milliseconds(),
 		sample,
 		nil,