add support for OSC sequences

escapes.go

@@ -22,28 +22,28 @@ type EscapeItem struct {
 // Required: The line shall not contain "\n"
 func ExtractTermEscapes(line string) (string, []EscapeItem) {
 	var termEscapes []EscapeItem
+	var ed EscapeDetector
 	var line1 strings.Builder
 
 	pos := 0
 	item := ""
 	occupiedRuneCount := 0
-	inEscape := false
 	for i, r := range []rune(line) {
-		if r == '\x1b' {
+		ed.Witness(r)
+		if ed.Started() {
 			pos = i
 			item = string(r)
-			inEscape = true
 			continue
 		}
-		if inEscape {
+		if ed.InEscape() {
 			item += string(r)
-			if r == 'm' {
+			if ed.Ended() {
 				termEscapes = append(termEscapes, EscapeItem{item, pos - occupiedRuneCount})
 				occupiedRuneCount += utf8.RuneCountInString(item)
-				inEscape = false
 			}
 			continue
 		}
+
 		line1.WriteRune(r)
 	}
 
@@ -93,3 +93,64 @@ func OffsetEscapes(escapes []EscapeItem, offset int) []EscapeItem {
 	}
 	return result
 }
+
+// EscapeDetector detect escape sequences in a stream of runes.
+// Supported sequences are:
+// - Select Graphic Rendition (SGR)
+// - Operating System Command (OSC)
+//
+// See https://chromium.googlesource.com/apps/libapps/+/refs/heads/master/hterm/doc/ControlSequences.md
+type EscapeDetector struct {
+	inEscape  bool
+	started   bool
+	ended     bool
+	firstRune rune
+}
+
+func (ed *EscapeDetector) Witness(r rune) {
+	if ed.inEscape && ed.ended {
+		ed.inEscape = false
+	}
+
+	if !ed.inEscape {
+		switch r {
+		case '\x1b': // SGR + OSC
+			ed.inEscape = true
+			ed.started = true
+			ed.ended = false
+			ed.firstRune = r
+		default:
+			ed.ended = false
+		}
+	} else {
+		switch {
+		case ed.firstRune == '\x1b' && r == 'm': // SGR
+			ed.inEscape = true
+			ed.started = false
+			ed.ended = true
+			ed.firstRune = rune(0)
+		case ed.firstRune == '\x1b' && r == '\x07': // OSC
+			ed.inEscape = true
+			ed.started = false
+			ed.ended = true
+			ed.firstRune = rune(0)
+		default:
+			ed.started = false
+		}
+	}
+}
+
+// InEscape indicate that the last rune was part of a sequence
+func (ed *EscapeDetector) InEscape() bool {
+	return ed.inEscape
+}
+
+// Started indicate that the last rune started a sequence
+func (ed *EscapeDetector) Started() bool {
+	return ed.started
+}
+
+// Ended indicate that the last rune ended a sequence
+func (ed *EscapeDetector) Ended() bool {
+	return ed.ended
+}

escapes_test.go

@@ -4,6 +4,7 @@ import (
 	"testing"
 
 	"github.com/stretchr/testify/assert"
+	"github.com/stretchr/testify/require"
 )
 
 func TestExtractApplyTermEscapes(t *testing.T) {
@@ -137,3 +138,49 @@ func TestOffsetEscapes(t *testing.T) {
 		assert.Equal(t, tc.output, result)
 	}
 }
+
+func TestEscapeDetector(t *testing.T) {
+	input := "This \u001B[31mis an\u001B[0m example."
+	states := []struct {
+		inEscape bool
+		started  bool
+		ended    bool
+	}{
+		{false, false, false}, // T
+		{false, false, false}, // h
+		{false, false, false}, // i
+		{false, false, false}, // s
+		{false, false, false}, //
+		{true, true, false},   // \u001b
+		{true, false, false},  // [
+		{true, false, false},  // 3
+		{true, false, false},  // 1
+		{true, false, true},   // m
+		{false, false, false}, // i
+		{false, false, false}, // s
+		{false, false, false}, //
+		{false, false, false}, // a
+		{false, false, false}, // n
+		{true, true, false},   // \u001b
+		{true, false, false},  // [
+		{true, false, false},  // 0
+		{true, false, true},   // m
+		{false, false, false}, //
+		{false, false, false}, // e
+		{false, false, false}, // x
+		{false, false, false}, // a
+		{false, false, false}, // m
+		{false, false, false}, // p
+		{false, false, false}, // l
+		{false, false, false}, // e
+		{false, false, false}, // .
+	}
+
+	var ed EscapeDetector
+	for i, r := range input {
+		ed.Witness(r)
+		require.Equal(t, states[i].inEscape, ed.InEscape())
+		require.Equal(t, states[i].started, ed.Started())
+		require.Equal(t, states[i].ended, ed.Ended())
+	}
+}

len.go

@@ -10,17 +10,12 @@ import (
 // escape sequences.
 func Len(text string) int {
 	length := 0
-	escape := false
+	var ed EscapeDetector
 
-	for _, char := range text {
-		if char == '\x1b' {
-			escape = true
-		}
-		if !escape {
-			length += runewidth.RuneWidth(char)
-		}
-		if char == 'm' {
-			escape = false
+	for _, r := range []rune(text) {
+		ed.Witness(r)
+		if !ed.InEscape() {
+			length += runewidth.RuneWidth(r)
 		}
 	}
 

wrap.go

@@ -389,16 +389,14 @@ func splitWord(word string, length int) (string, string) {
 	runes := []rune(word)
 	var result []rune
 	added := 0
-	escape := false
 
 	if length == 0 {
 		return "", word
 	}
 
+	var ed EscapeDetector
 	for _, r := range runes {
-		if r == '\x1b' {
-			escape = true
-		}
+		ed.Witness(r)
 
 		width := runewidth.RuneWidth(r)
 		if width+added > length {
@@ -408,16 +406,12 @@ func splitWord(word string, length int) (string, string) {
 
 		result = append(result, r)
 
-		if !escape {
+		if !ed.InEscape() {
 			added += width
 			if added >= length {
 				break
 			}
 		}
-
-		if r == 'm' {
-			escape = false
-		}
 	}
 
 	leftover := runes[len(result):]