From b3a794b7b04ecc4cbec7becdd4cd457fcc1456db Mon Sep 17 00:00:00 2001
From: Florian Thienel <ftl@thecodingflow.com>
Date: Fri, 24 Sep 2021 21:03:59 +0200
Subject: [PATCH] add utility functions to sanitize the text content

---
 sds/text.go      | 34 +++++++++++++++++++++
 sds/text_test.go | 77 ++++++++++++++++++++++++++++++++++++++++++++++++
 2 files changed, 111 insertions(+)
 create mode 100644 sds/text_test.go

diff --git a/sds/text.go b/sds/text.go
index 8b135c9..23153d8 100644
--- a/sds/text.go
+++ b/sds/text.go
@@ -2,6 +2,7 @@ package sds
 
 import (
 	"fmt"
+	"regexp"
 	"time"
 
 	"golang.org/x/text/encoding"
@@ -190,3 +191,36 @@ func AppendEncodedPayloadText(bytes []byte, bits int, text string, textEncoding
 	bits += encodedBits
 	return bytes, bits
 }
+
+var leadingOPTA = regexp.MustCompile(`^[A-Za-z ]+#[0-9]{16}`)
+
+func SplitLeadingOPTA(s string) (string, string) {
+	opta := leadingOPTA.FindString(s)
+	return opta, s[len(opta):]
+}
+
+func RemoveLeadingOPTA(s string) string {
+	_, result := SplitLeadingOPTA(s)
+	return result
+}
+
+var trailingITSI = regexp.MustCompile(`((\x1a\x00)|(\x0d\x0d))([0-9]{16})$`)
+
+func SplitTrailingITSI(s string) (string, string) {
+	groups := trailingITSI.FindStringSubmatch(s)
+	var itsi string
+	var matchLen int
+	if len(groups) == 0 {
+		itsi = ""
+		matchLen = 0
+	} else {
+		itsi = groups[len(groups)-1]
+		matchLen = len(groups[0])
+	}
+	return s[0 : len(s)-matchLen], itsi
+}
+
+func RemoveTrailingITSI(s string) string {
+	result, _ := SplitTrailingITSI(s)
+	return result
+}
diff --git a/sds/text_test.go b/sds/text_test.go
new file mode 100644
index 0000000..ba12ae8
--- /dev/null
+++ b/sds/text_test.go
@@ -0,0 +1,77 @@
+package sds
+
+import (
+	"testing"
+
+	"github.com/stretchr/testify/assert"
+)
+
+func TestSplitLeadingOPTA(t *testing.T) {
+	tt := []struct {
+		desc         string
+		value        string
+		expectedOPTA string
+		expectedTail string
+	}{
+		{
+			desc:         "no OPTA",
+			value:        "testmessage",
+			expectedOPTA: "",
+			expectedTail: "testmessage",
+		},
+		{
+			desc:         "only OPTA",
+			value:        "ABCD FG#1234567890123456",
+			expectedOPTA: "ABCD FG#1234567890123456",
+			expectedTail: "",
+		},
+		{
+			desc:         "OPTA and tail",
+			value:        "ABCD FG#1234567890123456testmessage",
+			expectedOPTA: "ABCD FG#1234567890123456",
+			expectedTail: "testmessage",
+		},
+	}
+	for _, tc := range tt {
+		t.Run(tc.desc, func(t *testing.T) {
+			actualOPTA, actualTail := SplitLeadingOPTA(tc.value)
+			assert.Equal(t, tc.expectedOPTA, actualOPTA)
+			assert.Equal(t, tc.expectedTail, actualTail)
+		})
+	}
+}
+
+func TestSplitTrailingITSI(t *testing.T) {
+	tt := []struct {
+		desc         string
+		value        string
+		expectedHead string
+		expectedITSI string
+	}{
+		{
+			desc:         "no ITSI",
+			value:        "testmessage",
+			expectedHead: "testmessage",
+			expectedITSI: "",
+		},
+		{
+			desc:         "cr cr",
+			value:        "testmessage\r\r1234567890123456",
+			expectedHead: "testmessage",
+			expectedITSI: "1234567890123456",
+		},
+		{
+			desc:         "ctrl-z nul",
+			value:        "testmessage\x1a\x001234567890123456",
+			expectedHead: "testmessage",
+			expectedITSI: "1234567890123456",
+		},
+	}
+	for _, tc := range tt {
+		t.Run(tc.desc, func(t *testing.T) {
+			actualHead, actualITSI := SplitTrailingITSI(tc.value)
+			assert.Equal(t, tc.expectedHead, actualHead)
+			assert.Equal(t, tc.expectedITSI, actualITSI)
+		})
+	}
+}