Fix FastRegexMatcher matching multibyte runes with . (#14059)

When `zeroOrOneCharacterStringMatcher` wach checking the input string,
it assumed that if there are more than one bytes, then there are more
than one runes, but that's not necessarily true.

Signed-off-by: Oleg Zaytsev <[email protected]>

Author: colega

model/labels/regexp.go

@@ -16,6 +16,7 @@ package labels
 import (
 	"slices"
 	"strings"
+	"unicode/utf8"
 
 	"github.com/grafana/regexp"
 	"github.com/grafana/regexp/syntax"
@@ -827,8 +828,7 @@ type zeroOrOneCharacterStringMatcher struct {
 }
 
 func (m *zeroOrOneCharacterStringMatcher) Matches(s string) bool {
-	// Zero or one.
-	if len(s) > 1 {
+	if moreThanOneRune(s) {
 		return false
 	}
 
@@ -840,6 +840,27 @@ func (m *zeroOrOneCharacterStringMatcher) Matches(s string) bool {
 	return s[0] != '\n'
 }
 
+// moreThanOneRune returns true if there are more than one runes in the string.
+// It doesn't check whether the string is valid UTF-8.
+// The return value should be always equal to utf8.RuneCountInString(s) > 1,
+// but the function is optimized for the common case where the string prefix is ASCII.
+func moreThanOneRune(s string) bool {
+	// If len(s) is exactly one or zero, there can't be more than one rune.
+	// Exit through this path quickly.
+	if len(s) <= 1 {
+		return false
+	}
+
+	// There's one or more bytes:
+	// If first byte is ASCII then there are multiple runes if there are more bytes after that.
+	if s[0] < utf8.RuneSelf {
+		return len(s) > 1
+	}
+
+	// Less common case: first is a multibyte rune.
+	return utf8.RuneCountInString(s) > 1
+}
+
 // trueMatcher is a stringMatcher which matches any string (always returns true).
 type trueMatcher struct{}
 

model/labels/regexp_test.go

@@ -84,7 +84,7 @@ var (
 		"foo", " foo bar", "bar", "buzz\nbar", "bar foo", "bfoo", "\n", "\nfoo", "foo\n", "hello foo world", "hello foo\n world", "",
 		"FOO", "Foo", "OO", "Oo", "\nfoo\n", strings.Repeat("f", 20), "prometheus", "prometheus_api_v1", "prometheus_api_v1_foo",
 		"10.0.1.20", "10.0.2.10", "10.0.3.30", "10.0.4.40",
-		"foofoo0", "foofoo",
+		"foofoo0", "foofoo", "😀foo0",
 
 		// Values matching / not matching the test regexps on long alternations.
 		"zQPbMkNO", "zQPbMkNo", "jyyfj00j0061", "jyyfj00j006", "jyyfj00j00612", "NNSPdvMi", "NNSPdvMiXXX", "NNSPdvMixxx", "nnSPdvMi", "nnSPdvMiXXX",