TestSanitizeString

Author: ungerik

strings.go

@@ -5,14 +5,15 @@ import (
 	"unicode"
 )
 
-// SanitizeString returns valid UTF-8 without any control code characters.
+// SanitizeString returns valid UTF-8 only with printable characters.
 func SanitizeString(s string) string {
-	return strings.Map(removeControlCodes, strings.ToValidUTF8(s, ""))
-}
-
-func removeControlCodes(r rune) rune {
-	if unicode.IsControl(r) {
-		return -1
-	}
-	return r
+	return strings.Map(
+		func(r rune) rune {
+			if r == '�' || !unicode.IsPrint(r) {
+				return -1
+			}
+			return r
+		},
+		s,
+	)
 }

strings_test.go

@@ -0,0 +1,22 @@
+package sqldb
+
+import "testing"
+
+func TestSanitizeString(t *testing.T) {
+	tests := map[string]string{
+		"":                                 "",
+		"Hello World!":                     "Hello World!",
+		"\u0000,\u0009,\u007f":             ",,",
+		"\a,\b,\v":                         ",,",
+		"\xc3\x22":                         "\"",
+		"\xbd\xb2\x3d\xbc\x20\xe2\x8c\x98": "= ⌘",
+		string([]byte{71, 101, 115, 99, 104, 195, 164, 102, 116, 115, 118, 111, 114, 102, 195}): "Geschäftsvorf",
+	}
+	for str, want := range tests {
+		t.Run(str, func(t *testing.T) {
+			if got := SanitizeString(str); got != want {
+				t.Errorf("SanitizeString(%q) = %q, want %q", str, got, want)
+			}
+		})
+	}
+}