link: sanitizeSymbol: rewrite with no regexp use

Instead of using a regular expression in sanitizeSymbol, write a small
for loop to replace any repetitions of unknown characters with a "_".

This removes the last use of regexp from this package.

Signed-off-by: Kir Kolyshkin <kolyshkin@gmail.com>

link/uprobe.go

@@ -6,7 +6,7 @@ import (
 	"fmt"
 	"os"
 	"path/filepath"
-	"regexp"
+	"strings"
 	"sync"
 
 	"github.com/cilium/ebpf"
@@ -16,10 +16,6 @@ import (
 var (
 	uprobeEventsPath = filepath.Join(tracefsPath, "uprobe_events")
 
-	// rgxEventSymbol is used to strip invalid characters from the [k,u]probe symbol
-	// as they are not allowed to be used as the EVENT token in tracefs.
-	rgxEventSymbol = regexp.MustCompile("[^a-zA-Z0-9]+")
-
 	uprobeRetprobeBit = struct {
 		once  sync.Once
 		value uint64
@@ -316,8 +312,28 @@ func tracefsUprobe(args probeArgs) (*perfEvent, error) {
 }
 
 // sanitizedSymbol replaces every invalid character for the tracefs api with an underscore.
-func sanitizedSymbol(symbol string) string {
-	return rgxEventSymbol.ReplaceAllString(symbol, "_")
+// It is equivalent to calling regexp.MustCompile("[^a-zA-Z0-9]+").ReplaceAllString("_").
+func sanitizedSymbol(s string) string {
+	var b strings.Builder
+	b.Grow(len(s))
+	var skip bool
+	for _, c := range []byte(s) {
+		switch {
+		case c >= 'a' && c <= 'z',
+			c >= 'A' && c <= 'Z',
+			c >= '0' && c <= '9':
+			skip = false
+			b.WriteByte(c)
+
+		default:
+			if !skip {
+				b.WriteByte('_')
+				skip = true
+			}
+		}
+	}
+
+	return b.String()
 }
 
 // uprobeToken creates the PATH:OFFSET(REF_CTR_OFFSET) token for the tracefs api.

link/uprobe_test.go

@@ -280,9 +280,11 @@ func TestUprobeSanitizedSymbol(t *testing.T) {
 		expected string
 	}{
 		{"readline", "readline"},
-		{"main.Func", "main_Func"},
+		{"main.Func123", "main_Func123"},
 		{"a.....a", "a_a"},
 		{"./;'{}[]a", "_a"},
+		{"***xx**xx###", "_xx_xx_"},
+		{`@P#r$i%v^3*+t)i&k++--`, "_P_r_i_v_3_t_i_k_"},
 	}
 
 	for i, tt := range tests {