add UTF8-safe truncation utility function

ircutils/unicode.go

@@ -0,0 +1,25 @@
+// Copyright (c) 2021 Shivaram Lingamneni
+// Released under the MIT License
+
+package ircutils
+
+import (
+	"unicode/utf8"
+)
+
+// truncate a message, taking care not to make valid UTF8 into invalid UTF8
+func TruncateUTF8Safe(message string, byteLimit int) (result string) {
+	if len(message) <= byteLimit {
+		return message
+	}
+	message = message[:byteLimit]
+	for i := 0; i < (utf8.UTFMax - 1); i++ {
+		r, n := utf8.DecodeLastRuneInString(message)
+		if r == utf8.RuneError && n <= 1 {
+			message = message[:len(message)-1]
+		} else {
+			break
+		}
+	}
+	return message
+}

ircutils/unicode_test.go

@@ -0,0 +1,33 @@
+// Copyright (c) 2021 Shivaram Lingamneni
+// Released under the MIT License
+
+package ircutils
+
+import (
+	"fmt"
+	"reflect"
+	"testing"
+)
+
+func assertEqual(found, expected interface{}) {
+	if !reflect.DeepEqual(found, expected) {
+		panic(fmt.Sprintf("expected %#v, found %#v", expected, found))
+	}
+}
+func TestTruncateUTF8(t *testing.T) {
+	assertEqual(TruncateUTF8Safe("fffff", 512), "fffff")
+	assertEqual(TruncateUTF8Safe("fffff", 5), "fffff")
+	assertEqual(TruncateUTF8Safe("ffffff", 5), "fffff")
+	assertEqual(TruncateUTF8Safe("ffffffffff", 5), "fffff")
+
+	assertEqual(TruncateUTF8Safe("12345🐬", 9), "12345🐬")
+	assertEqual(TruncateUTF8Safe("12345🐬", 8), "12345")
+	assertEqual(TruncateUTF8Safe("12345🐬", 7), "12345")
+	assertEqual(TruncateUTF8Safe("12345🐬", 6), "12345")
+	assertEqual(TruncateUTF8Safe("12345", 5), "12345")
+
+	assertEqual(TruncateUTF8Safe("\xff\xff\xff\xff\xff\xff", 512), "\xff\xff\xff\xff\xff\xff")
+	assertEqual(TruncateUTF8Safe("\xff\xff\xff\xff\xff\xff", 6), "\xff\xff\xff\xff\xff\xff")
+	// shouldn't truncate the whole string
+	assertEqual(TruncateUTF8Safe("\xff\xff\xff\xff\xff\xff", 5), "\xff\xff")
+}