lib/promrelabel: optimize action: {drop,keep,labeldrop,labelkeep} with anchored regex prefix

The following commonly used relabeling rules must work faster now:

- action: labeldrop
  regex: "^foo.+$"

- action: labeldrop
  regex: "^bar.*"

lib/promrelabel/config.go

@@ -202,7 +202,7 @@ func parseRelabelConfig(rc *RelabelConfig) (*parsedRelabelConfig, error) {
 	regexOriginalCompiled := defaultOriginalRegexForRelabelConfig
 	var regexOrValues []string
 	if rc.Regex != nil {
-		regex := rc.Regex.S
+		regex := regexutil.RemoveStartEndAnchors(rc.Regex.S)
 		regexOrig := regex
 		if rc.Action != "replace_all" && rc.Action != "labelmap_all" {
 			regex = "^(?:" + regex + ")$"

lib/promrelabel/relabel_test.go

@@ -1,6 +1,7 @@
 package promrelabel
 
 import (
+	"fmt"
 	"testing"
 
 	"github.com/VictoriaMetrics/VictoriaMetrics/lib/prompbmarshal"
@@ -730,16 +731,7 @@ func TestFillLabelReferences(t *testing.T) {
 func TestRegexpMatchStringSuccess(t *testing.T) {
 	f := func(pattern, s string) {
 		t.Helper()
-		rc := &RelabelConfig{
-			Action: "labeldrop",
-			Regex: &MultiLineRegex{
-				S: pattern,
-			},
-		}
-		prc, err := parseRelabelConfig(rc)
-		if err != nil {
-			t.Fatalf("unexpected error in parseRelabelConfig: %s", err)
-		}
+		prc := newTestRegexRelabelConfig(pattern)
 		if !prc.matchString(s) {
 			t.Fatalf("unexpected matchString(%q) result; got false; want true", s)
 		}
@@ -760,16 +752,7 @@ func TestRegexpMatchStringSuccess(t *testing.T) {
 func TestRegexpMatchStringFailure(t *testing.T) {
 	f := func(pattern, s string) {
 		t.Helper()
-		rc := &RelabelConfig{
-			Action: "labeldrop",
-			Regex: &MultiLineRegex{
-				S: pattern,
-			},
-		}
-		prc, err := parseRelabelConfig(rc)
-		if err != nil {
-			t.Fatalf("unexpected error in parseRelabelConfig: %s", err)
-		}
+		prc := newTestRegexRelabelConfig(pattern)
 		if prc.matchString(s) {
 			t.Fatalf("unexpected matchString(%q) result; got true; want false", s)
 		}
@@ -784,3 +767,17 @@ func TestRegexpMatchStringFailure(t *testing.T) {
 	f("foo.+", "foo")
 	f("^foo$", "foobar")
 }
+
+func newTestRegexRelabelConfig(pattern string) *parsedRelabelConfig {
+	rc := &RelabelConfig{
+		Action: "labeldrop",
+		Regex: &MultiLineRegex{
+			S: pattern,
+		},
+	}
+	prc, err := parseRelabelConfig(rc)
+	if err != nil {
+		panic(fmt.Errorf("unexpected error in parseRelabelConfig: %s", err))
+	}
+	return prc
+}

lib/promrelabel/relabel_timing_test.go

@@ -8,19 +8,10 @@ import (
 	"github.com/VictoriaMetrics/VictoriaMetrics/lib/prompbmarshal"
 )
 
-func BenchmarkMatchRegexOrValuesMatchOptimized(b *testing.B) {
-	const pattern = "foo|bar|baz|abc"
-	const s = "foo"
-	rc := &RelabelConfig{
-		Action: "labeldrop",
-		Regex: &MultiLineRegex{
-			S: pattern,
-		},
-	}
-	prc, err := parseRelabelConfig(rc)
-	if err != nil {
-		panic(fmt.Errorf("unexpected error in parseRelabelConfig: %s", err))
-	}
+func BenchmarkMatchRegexPrefixDotPlusMatchOptimized(b *testing.B) {
+	const pattern = "^foo.+$"
+	const s = "foobar"
+	prc := newTestRegexRelabelConfig(pattern)
 	b.ReportAllocs()
 	b.SetBytes(1)
 	b.RunParallel(func(pb *testing.PB) {
@@ -32,19 +23,25 @@ func BenchmarkMatchRegexOrValuesMatchOptimized(b *testing.B) {
 	})
 }
 
-func BenchmarkMatchRegexOrValuesMismatchOptimized(b *testing.B) {
-	const pattern = "foo|bar|baz|abc"
-	const s = "qwert"
-	rc := &RelabelConfig{
-		Action: "labeldrop",
-		Regex: &MultiLineRegex{
-			S: pattern,
-		},
+func BenchmarkMatchRegexPrefixDotPlusMatchUnoptimized(b *testing.B) {
+	const pattern = "^foo.+$"
+	const s = "foobar"
+	re := regexp.MustCompile(pattern)
+	b.ReportAllocs()
+	b.SetBytes(1)
+	b.RunParallel(func(pb *testing.PB) {
+		for pb.Next() {
+			if !re.MatchString(s) {
+				panic(fmt.Errorf("unexpected string mismatch for pattern=%q, s=%q", pattern, s))
 			}
-	prc, err := parseRelabelConfig(rc)
-	if err != nil {
-		panic(fmt.Errorf("unexpected error in parseRelabelConfig: %s", err))
 		}
+	})
+}
+
+func BenchmarkMatchRegexPrefixDotPlusMismatchOptimized(b *testing.B) {
+	const pattern = "^foo.+$"
+	const s = "xfoobar"
+	prc := newTestRegexRelabelConfig(pattern)
 	b.ReportAllocs()
 	b.SetBytes(1)
 	b.RunParallel(func(pb *testing.PB) {
@@ -56,8 +53,98 @@ func BenchmarkMatchRegexOrValuesMismatchOptimized(b *testing.B) {
 	})
 }
 
-func BenchmarkMatchRegexOrValuesMatchUnoptimized(b *testing.B) {
-	const pattern = "foo|bar|baz|abc"
+func BenchmarkMatchRegexPrefixDotPlusMismatchUnoptimized(b *testing.B) {
+	const pattern = "^foo.+$"
+	const s = "xfoobar"
+	re := regexp.MustCompile(pattern)
+	b.ReportAllocs()
+	b.SetBytes(1)
+	b.RunParallel(func(pb *testing.PB) {
+		for pb.Next() {
+			if re.MatchString(s) {
+				panic(fmt.Errorf("unexpected string match for pattern=%q, s=%q", pattern, s))
+			}
+		}
+	})
+}
+
+func BenchmarkMatchRegexPrefixDotStarMatchOptimized(b *testing.B) {
+	const pattern = "^foo.*$"
+	const s = "foobar"
+	prc := newTestRegexRelabelConfig(pattern)
+	b.ReportAllocs()
+	b.SetBytes(1)
+	b.RunParallel(func(pb *testing.PB) {
+		for pb.Next() {
+			if !prc.matchString(s) {
+				panic(fmt.Errorf("unexpected string mismatch for pattern=%q, s=%q", pattern, s))
+			}
+		}
+	})
+}
+
+func BenchmarkMatchRegexPrefixDotStarMatchUnoptimized(b *testing.B) {
+	const pattern = "^foo.*$"
+	const s = "foobar"
+	re := regexp.MustCompile(pattern)
+	b.ReportAllocs()
+	b.SetBytes(1)
+	b.RunParallel(func(pb *testing.PB) {
+		for pb.Next() {
+			if !re.MatchString(s) {
+				panic(fmt.Errorf("unexpected string mismatch for pattern=%q, s=%q", pattern, s))
+			}
+		}
+	})
+}
+
+func BenchmarkMatchRegexPrefixDotStarMismatchOptimized(b *testing.B) {
+	const pattern = "^foo.*$"
+	const s = "xfoobar"
+	prc := newTestRegexRelabelConfig(pattern)
+	b.ReportAllocs()
+	b.SetBytes(1)
+	b.RunParallel(func(pb *testing.PB) {
+		for pb.Next() {
+			if prc.matchString(s) {
+				panic(fmt.Errorf("unexpected string match for pattern=%q, s=%q", pattern, s))
+			}
+		}
+	})
+}
+
+func BenchmarkMatchRegexPrefixDotStarMismatchUnoptimized(b *testing.B) {
+	const pattern = "^foo.*$"
+	const s = "xfoobar"
+	re := regexp.MustCompile(pattern)
+	b.ReportAllocs()
+	b.SetBytes(1)
+	b.RunParallel(func(pb *testing.PB) {
+		for pb.Next() {
+			if re.MatchString(s) {
+				panic(fmt.Errorf("unexpected string match for pattern=%q, s=%q", pattern, s))
+			}
+		}
+	})
+}
+
+func BenchmarkMatchRegexSingleValueMatchOptimized(b *testing.B) {
+	const pattern = "^foo$"
+	const s = "foo"
+	prc := newTestRegexRelabelConfig(pattern)
+	b.ReportAllocs()
+	b.SetBytes(1)
+	b.RunParallel(func(pb *testing.PB) {
+		for pb.Next() {
+			if !prc.matchString(s) {
+				panic(fmt.Errorf("unexpected string mismatch for pattern=%q, s=%q", pattern, s))
+			}
+		}
+	})
+}
+
+func BenchmarkMatchRegexSingleValueMatchUnoptimized(b *testing.B) {
+	const pattern = "^foo$"
 	const s = "foo"
 	re := regexp.MustCompile(pattern)
 	b.ReportAllocs()
@@ -71,8 +158,83 @@ func BenchmarkMatchRegexOrValuesMatchUnoptimized(b *testing.B) {
 	})
 }
 
+func BenchmarkMatchRegexSingleValueMismatchOptimized(b *testing.B) {
+	const pattern = "^foo$"
+	const s = "bar"
+	prc := newTestRegexRelabelConfig(pattern)
+	b.ReportAllocs()
+	b.SetBytes(1)
+	b.RunParallel(func(pb *testing.PB) {
+		for pb.Next() {
+			if prc.matchString(s) {
+				panic(fmt.Errorf("unexpected string match for pattern=%q, s=%q", pattern, s))
+			}
+		}
+	})
+}
+
+func BenchmarkMatchRegexSingleValueMismatchUnoptimized(b *testing.B) {
+	const pattern = "^foo$"
+	const s = "bar"
+	re := regexp.MustCompile(pattern)
+	b.ReportAllocs()
+	b.SetBytes(1)
+	b.RunParallel(func(pb *testing.PB) {
+		for pb.Next() {
+			if re.MatchString(s) {
+				panic(fmt.Errorf("unexpected string match for pattern=%q, s=%q", pattern, s))
+			}
+		}
+	})
+}
+
+func BenchmarkMatchRegexOrValuesMatchOptimized(b *testing.B) {
+	const pattern = "^(foo|bar|baz|abc)$"
+	const s = "foo"
+	prc := newTestRegexRelabelConfig(pattern)
+	b.ReportAllocs()
+	b.SetBytes(1)
+	b.RunParallel(func(pb *testing.PB) {
+		for pb.Next() {
+			if !prc.matchString(s) {
+				panic(fmt.Errorf("unexpected string mismatch for pattern=%q, s=%q", pattern, s))
+			}
+		}
+	})
+}
+
+func BenchmarkMatchRegexOrValuesMatchUnoptimized(b *testing.B) {
+	const pattern = "^(foo|bar|baz|abc)$"
+	const s = "foo"
+	re := regexp.MustCompile(pattern)
+	b.ReportAllocs()
+	b.SetBytes(1)
+	b.RunParallel(func(pb *testing.PB) {
+		for pb.Next() {
+			if !re.MatchString(s) {
+				panic(fmt.Errorf("unexpected string mismatch for pattern=%q, s=%q", pattern, s))
+			}
+		}
+	})
+}
+
+func BenchmarkMatchRegexOrValuesMismatchOptimized(b *testing.B) {
+	const pattern = "^(foo|bar|baz|abc)"
+	const s = "qwert"
+	prc := newTestRegexRelabelConfig(pattern)
+	b.ReportAllocs()
+	b.SetBytes(1)
+	b.RunParallel(func(pb *testing.PB) {
+		for pb.Next() {
+			if prc.matchString(s) {
+				panic(fmt.Errorf("unexpected string match for pattern=%q, s=%q", pattern, s))
+			}
+		}
+	})
+}
+
 func BenchmarkMatchRegexOrValuesMismatchUnoptimized(b *testing.B) {
-	const pattern = "foo|bar|baz|abc"
+	const pattern = "^(foo|bar|baz|abc)$"
 	const s = "qwert"
 	re := regexp.MustCompile(pattern)
 	b.ReportAllocs()

lib/regexutil/regexutil.go

@@ -3,16 +3,31 @@ package regexutil
 import (
 	"regexp/syntax"
 	"sort"
+	"strings"
 
 	"github.com/VictoriaMetrics/VictoriaMetrics/lib/logger"
 )
 
+// RemoveStartEndAnchors removes '^' at the start of expr and '$' at the end of the expr.
+func RemoveStartEndAnchors(expr string) string {
+	for strings.HasPrefix(expr, "^") {
+		expr = expr[1:]
+	}
+	for strings.HasSuffix(expr, "$") {
+		expr = expr[:len(expr)-1]
+	}
+	return expr
+}
+
 // GetOrValues returns "or" values from the given regexp expr.
 //
-// E.g. it returns ["foo", "bar"] for "foo|bar" regexp.
-// It returns an empty list if it is impossible to extract "or" values from the regexp.
+// It ignores start and end anchors ('^') and ('$') at the start and the end of expr.
+// It returns ["foo", "bar"] for "foo|bar" regexp.
+// It returns ["foo"] for "foo" regexp.
 // It returns [""] for "" regexp.
+// It returns an empty list if it is impossible to extract "or" values from the regexp.
 func GetOrValues(expr string) []string {
+	expr = RemoveStartEndAnchors(expr)
 	sre, err := syntax.Parse(expr, syntax.Perl)
 	if err != nil {
 		logger.Panicf("BUG: unexpected error when parsing verified expr=%q: %s", expr, err)