Merge pull request #60 from slingamn/ircevent_truncation

fix #59, add a message sanitizer

ircevent/irc.go

@@ -351,7 +351,7 @@ func (irc *Connection) sendInternal(b []byte) (err error) {
 // Send a built ircmsg.Message.
 func (irc *Connection) SendIRCMessage(msg ircmsg.Message) error {
 	b, err := msg.LineBytesStrict(true, irc.MaxLineLen)
-	if err != nil {
+	if err != nil && !(irc.AllowTruncation && err == ircmsg.ErrorBodyTooLong) {
 		if irc.Debug {
 			irc.Log.Printf("couldn't assemble message: %v\n", err)
 		}

ircevent/irc_struct.go

@@ -41,28 +41,29 @@ type capResult struct {
 
 type Connection struct {
 	// config data, user-settable
-	Server        string
-	TLSConfig     *tls.Config
-	Nick          string
-	User          string
-	RealName      string   // IRC realname/gecos
-	WebIRC        []string // parameters for the WEBIRC command
-	Password      string   // server password (PASS command)
-	RequestCaps   []string // IRCv3 capabilities to request (failure is non-fatal)
-	SASLLogin     string   // SASL credentials to log in with (failure is fatal)
-	SASLPassword  string
-	SASLMech      string
-	QuitMessage   string
-	Version       string
-	Timeout       time.Duration
-	KeepAlive     time.Duration
-	ReconnectFreq time.Duration
-	MaxLineLen    int // maximum line length, not including tags
-	UseTLS        bool
-	UseSASL       bool
-	EnableCTCP    bool
-	Debug         bool
-	AllowPanic    bool // if set, don't recover() from panics in callbacks
+	Server          string
+	TLSConfig       *tls.Config
+	Nick            string
+	User            string
+	RealName        string   // IRC realname/gecos
+	WebIRC          []string // parameters for the WEBIRC command
+	Password        string   // server password (PASS command)
+	RequestCaps     []string // IRCv3 capabilities to request (failure is non-fatal)
+	SASLLogin       string   // SASL credentials to log in with (failure is fatal)
+	SASLPassword    string
+	SASLMech        string
+	QuitMessage     string
+	Version         string
+	Timeout         time.Duration
+	KeepAlive       time.Duration
+	ReconnectFreq   time.Duration
+	MaxLineLen      int // maximum line length, not including tags
+	UseTLS          bool
+	UseSASL         bool
+	EnableCTCP      bool
+	Debug           bool
+	AllowPanic      bool // if set, don't recover() from panics in callbacks
+	AllowTruncation bool // if set, truncate lines exceeding MaxLineLen and send them
 
 	// networking and synchronization
 	stateMutex sync.Mutex     // innermost mutex: don't block while holding this

ircutils/unicode.go

@@ -4,6 +4,8 @@
 package ircutils
 
 import (
+	"strings"
+	"unicode"
 	"unicode/utf8"
 )
 
@@ -23,3 +25,38 @@ func TruncateUTF8Safe(message string, byteLimit int) (result string) {
 	}
 	return message
 }
+
+// Sanitizes human-readable text to make it safe for IRC;
+// assumes UTF-8 and uses the replacement character where
+// applicable.
+func SanitizeText(message string, byteLimit int) (result string) {
+	var buf strings.Builder
+
+	for _, r := range message {
+		if r == '\x00' || r == '\r' {
+			continue
+		} else if r == '\n' {
+			if buf.Len()+2 <= byteLimit {
+				buf.WriteString("  ")
+				continue
+			} else {
+				break
+			}
+		} else if unicode.IsSpace(r) {
+			if buf.Len()+1 <= byteLimit {
+				buf.WriteString(" ")
+			} else {
+				break
+			}
+		} else {
+			rLen := utf8.RuneLen(r)
+			if buf.Len()+rLen <= byteLimit {
+				buf.WriteRune(r)
+			} else {
+				break
+			}
+		}
+	}
+
+	return buf.String()
+}

ircutils/unicode_test.go

@@ -31,3 +31,16 @@ func TestTruncateUTF8(t *testing.T) {
 	// shouldn't truncate the whole string
 	assertEqual(TruncateUTF8Safe("\xff\xff\xff\xff\xff\xff", 5), "\xff\xff")
 }
+
+func TestSanitize(t *testing.T) {
+	assertEqual(SanitizeText("abc", 10), "abc")
+	assertEqual(SanitizeText("abcdef", 5), "abcde")
+
+	assertEqual(SanitizeText("shivaram\x00shivaram\x00shivarampassphrase", 400), "shivaramshivaramshivarampassphrase")
+
+	assertEqual(SanitizeText("the quick brown fox\xffjumps over the lazy dog", 400), "the quick brown fox\xef\xbf\xbdjumps over the lazy dog")
+
+	// \r ignored, \n is two spaces
+	assertEqual(SanitizeText("the quick brown fox\r\njumps over the lazy dog", 400), "the quick brown fox  jumps over the lazy dog")
+	assertEqual(SanitizeText("the quick brown fox\njumps over the lazy dog", 400), "the quick brown fox  jumps over the lazy dog")
+}