From c0f1abb574ae928662146907e0c0fbe6eca979e0 Mon Sep 17 00:00:00 2001
From: Alan Protasio <alanprot@gmail.com>
Date: Wed, 10 May 2023 15:46:14 -0700
Subject: [PATCH] MatchNotRegexp optimization

Signed-off-by: Alan Protasio <alanprot@gmail.com>
---
 tsdb/querier.go            |  8 +++++
 tsdb/querier_bench_test.go |  3 ++
 tsdb/querier_test.go       | 60 ++++++++++++++++++++++++++------------
 3 files changed, 52 insertions(+), 19 deletions(-)

diff --git a/tsdb/querier.go b/tsdb/querier.go
index 70f8f6de8..8b17bb745 100644
--- a/tsdb/querier.go
+++ b/tsdb/querier.go
@@ -371,6 +371,14 @@ func inversePostingsForMatcher(ix IndexReader, m *labels.Matcher) (index.Posting
 	if m.Type == labels.MatchEqual && m.Value == "" {
 		res = vals
 	} else {
+		// Inverse of a MatchNotRegexp is MatchRegexp (double negation).
+		// Fast-path for set matching.
+		if m.Type == labels.MatchNotRegexp {
+			setMatches := findSetMatches(m.GetRegexString())
+			if len(setMatches) > 0 {
+				return ix.Postings(m.Name, setMatches...)
+			}
+		}
 		for _, val := range vals {
 			if !m.Matches(val) {
 				res = append(res, val)
diff --git a/tsdb/querier_bench_test.go b/tsdb/querier_bench_test.go
index 89758a1d3..6e5243ef1 100644
--- a/tsdb/querier_bench_test.go
+++ b/tsdb/querier_bench_test.go
@@ -114,6 +114,7 @@ func benchmarkPostingsForMatchers(b *testing.B, ir IndexReader) {
 	iCharSet := labels.MustNewMatcher(labels.MatchRegexp, "i", "1[0-9]")
 	iAlternate := labels.MustNewMatcher(labels.MatchRegexp, "i", "(1|2|3|4|5|6|20|55)")
 	iXYZ := labels.MustNewMatcher(labels.MatchRegexp, "i", "X|Y|Z")
+	iNotXYZ := labels.MustNewMatcher(labels.MatchNotRegexp, "i", "X|Y|Z")
 	cases := []struct {
 		name     string
 		matchers []*labels.Matcher
@@ -131,6 +132,7 @@ func benchmarkPostingsForMatchers(b *testing.B, ir IndexReader) {
 		{`j=~"X.+"`, []*labels.Matcher{jXplus}},
 		{`i=~"(1|2|3|4|5|6|20|55)"`, []*labels.Matcher{iAlternate}},
 		{`i=~"X|Y|Z"`, []*labels.Matcher{iXYZ}},
+		{`i!~"X|Y|Z"`, []*labels.Matcher{iNotXYZ}},
 		{`i=~".*"`, []*labels.Matcher{iStar}},
 		{`i=~"1.*"`, []*labels.Matcher{i1Star}},
 		{`i=~".*1"`, []*labels.Matcher{iStar1}},
@@ -146,6 +148,7 @@ func benchmarkPostingsForMatchers(b *testing.B, ir IndexReader) {
 		{`n="1",i!="",j=~"X.+"`, []*labels.Matcher{n1, iNotEmpty, jXplus}},
 		{`n="1",i!="",j=~"XXX|YYY"`, []*labels.Matcher{n1, iNotEmpty, jXXXYYY}},
 		{`n="1",i=~"X|Y|Z",j="foo"`, []*labels.Matcher{n1, iXYZ, jFoo}},
+		{`n="1",i!~"X|Y|Z",j="foo"`, []*labels.Matcher{n1, iNotXYZ, jFoo}},
 		{`n="1",i=~".+",j="foo"`, []*labels.Matcher{n1, iPlus, jFoo}},
 		{`n="1",i=~"1.+",j="foo"`, []*labels.Matcher{n1, i1Plus, jFoo}},
 		{`n="1",i=~".*1.*",j="foo"`, []*labels.Matcher{n1, iStar1Star, jFoo}},
diff --git a/tsdb/querier_test.go b/tsdb/querier_test.go
index 5bb99a453..9e17f5b8d 100644
--- a/tsdb/querier_test.go
+++ b/tsdb/querier_test.go
@@ -1801,6 +1801,19 @@ func TestPostingsForMatchers(t *testing.T) {
 				labels.FromStrings("n", "2.5"),
 			},
 		},
+		{
+			matchers: []*labels.Matcher{labels.MustNewMatcher(labels.MatchNotRegexp, "n", "1")},
+			exp: []labels.Labels{
+				labels.FromStrings("n", "2"),
+				labels.FromStrings("n", "2.5"),
+			},
+		},
+		{
+			matchers: []*labels.Matcher{labels.MustNewMatcher(labels.MatchNotRegexp, "n", "1|2.5")},
+			exp: []labels.Labels{
+				labels.FromStrings("n", "2"),
+			},
+		},
 		{
 			matchers: []*labels.Matcher{labels.MustNewMatcher(labels.MatchEqual, "n", "1"), labels.MustNewMatcher(labels.MatchNotRegexp, "i", "^a$")},
 			exp: []labels.Labels{
@@ -1890,27 +1903,36 @@ func TestPostingsForMatchers(t *testing.T) {
 	require.NoError(t, err)
 
 	for _, c := range cases {
-		exp := map[string]struct{}{}
-		for _, l := range c.exp {
-			exp[l.String()] = struct{}{}
-		}
-		p, err := PostingsForMatchers(ir, c.matchers...)
-		require.NoError(t, err)
-
-		var builder labels.ScratchBuilder
-		for p.Next() {
-			require.NoError(t, ir.Series(p.At(), &builder, &[]chunks.Meta{}))
-			lbls := builder.Labels()
-			if _, ok := exp[lbls.String()]; !ok {
-				t.Errorf("Evaluating %v, unexpected result %s", c.matchers, lbls.String())
-			} else {
-				delete(exp, lbls.String())
+		name := ""
+		for i, matcher := range c.matchers {
+			if i > 0 {
+				name += ","
 			}
+			name += matcher.String()
 		}
-		require.NoError(t, p.Err())
-		if len(exp) != 0 {
-			t.Errorf("Evaluating %v, missing results %+v", c.matchers, exp)
-		}
+		t.Run(name, func(t *testing.T) {
+			exp := map[string]struct{}{}
+			for _, l := range c.exp {
+				exp[l.String()] = struct{}{}
+			}
+			p, err := PostingsForMatchers(ir, c.matchers...)
+			require.NoError(t, err)
+
+			var builder labels.ScratchBuilder
+			for p.Next() {
+				require.NoError(t, ir.Series(p.At(), &builder, &[]chunks.Meta{}))
+				lbls := builder.Labels()
+				if _, ok := exp[lbls.String()]; !ok {
+					t.Errorf("Evaluating %v, unexpected result %s", c.matchers, lbls.String())
+				} else {
+					delete(exp, lbls.String())
+				}
+			}
+			require.NoError(t, p.Err())
+			if len(exp) != 0 {
+				t.Errorf("Evaluating %v, missing results %+v", c.matchers, exp)
+			}
+		})
 	}
 }