From 286535320029f669a2b4b96723a2a1ba313214a1 Mon Sep 17 00:00:00 2001
From: Vincent Batts <vbatts@hashbangbash.com>
Date: Wed, 23 Sep 2015 13:30:00 -0400
Subject: [PATCH] common: add a UTF-8 check helper

---
 tar/common/utf8.go      | 21 +++++++++++++++++++++
 tar/common/utf8_test.go | 34 ++++++++++++++++++++++++++++++++++
 2 files changed, 55 insertions(+)
 create mode 100644 tar/common/utf8.go
 create mode 100644 tar/common/utf8_test.go

diff --git a/tar/common/utf8.go b/tar/common/utf8.go
new file mode 100644
index 0000000..ffb1646
--- /dev/null
+++ b/tar/common/utf8.go
@@ -0,0 +1,21 @@
+package common
+
+// IsValidUtf8String checks for in valid UTF-8 characters
+func IsValidUtf8String(s string) bool {
+	for _, r := range s {
+		if int(r) == 0xfffd {
+			return false
+		}
+	}
+	return true
+}
+
+// IsValidUtf8Btyes checks for in valid UTF-8 characters
+func IsValidUtf8Btyes(b []byte) bool {
+	for _, r := range string(b) {
+		if int(r) == 0xfffd {
+			return false
+		}
+	}
+	return true
+}
diff --git a/tar/common/utf8_test.go b/tar/common/utf8_test.go
new file mode 100644
index 0000000..e546f55
--- /dev/null
+++ b/tar/common/utf8_test.go
@@ -0,0 +1,34 @@
+package common
+
+import "testing"
+
+func TestStringValidation(t *testing.T) {
+	cases := []struct {
+		value  string
+		result bool
+	}{
+		{"aä\uFFFD本☺", false},
+		{"aä本☺", true},
+	}
+
+	for _, c := range cases {
+		if got := IsValidUtf8String(c.value); got != c.result {
+			t.Errorf("string %q - expected %v, got %v", c.value, c.result, got)
+		}
+	}
+}
+func TestBytesValidation(t *testing.T) {
+	cases := []struct {
+		value  []byte
+		result bool
+	}{
+		{[]byte{0xE4}, false},
+		{[]byte("aä本☺"), true},
+	}
+
+	for _, c := range cases {
+		if got := IsValidUtf8Btyes(c.value); got != c.result {
+			t.Errorf("bytes %q - expected %v, got %v", c.value, c.result, got)
+		}
+	}
+}