text: handle hyperlink embedded text correctly; fixes #329 (#334)

text/escape_seq_parser.go

@@ -0,0 +1,201 @@
+package text
+
+import (
+	"fmt"
+	"sort"
+	"strconv"
+	"strings"
+)
+
+// Constants
+const (
+	EscapeReset        = EscapeResetCSI
+	EscapeResetCSI     = EscapeStartCSI + "0" + EscapeStopCSI
+	EscapeResetOSI     = EscapeStartOSI + "0" + EscapeStopOSI
+	EscapeStart        = EscapeStartCSI
+	EscapeStartCSI     = "\x1b["
+	EscapeStartOSI     = "\x1b]"
+	EscapeStartRune    = rune(27) // \x1b
+	EscapeStartRuneCSI = '['      // [
+	EscapeStartRuneOSI = ']'      // ]
+	EscapeStop         = EscapeStopCSI
+	EscapeStopCSI      = "m"
+	EscapeStopOSI      = "\\"
+	EscapeStopRune     = EscapeStopRuneCSI
+	EscapeStopRuneCSI  = 'm'
+	EscapeStopRuneOSI  = '\\'
+)
+
+// Deprecated Constants
+const (
+	CSIStartRune = EscapeStartRuneCSI
+	CSIStopRune  = EscapeStopRuneCSI
+	OSIStartRune = EscapeStartRuneOSI
+	OSIStopRune  = EscapeStopRuneOSI
+)
+
+type escSeqKind int
+
+const (
+	escSeqKindUnknown escSeqKind = iota
+	escSeqKindCSI
+	escSeqKindOSI
+)
+
+type escSeqParser struct {
+	codes map[int]bool
+
+	// consume specific
+	inEscSeq   bool
+	escSeqKind escSeqKind
+	escapeSeq  string
+}
+
+func (s *escSeqParser) Codes() []int {
+	codes := make([]int, 0)
+	for code, val := range s.codes {
+		if val {
+			codes = append(codes, code)
+		}
+	}
+	sort.Ints(codes)
+	return codes
+}
+
+func (s *escSeqParser) Consume(char rune) {
+	if !s.inEscSeq && char == EscapeStartRune {
+		s.inEscSeq = true
+		s.escSeqKind = escSeqKindUnknown
+		s.escapeSeq = ""
+	} else if s.inEscSeq && s.escSeqKind == escSeqKindUnknown {
+		if char == EscapeStartRuneCSI {
+			s.escSeqKind = escSeqKindCSI
+		} else if char == EscapeStartRuneOSI {
+			s.escSeqKind = escSeqKindOSI
+		}
+	}
+
+	if s.inEscSeq {
+		s.escapeSeq += string(char)
+
+		if s.isEscapeStopRune(char) {
+			s.ParseSeq(s.escapeSeq, s.escSeqKind)
+			s.Reset()
+		}
+	}
+}
+
+func (s *escSeqParser) InSequence() bool {
+	return s.inEscSeq
+}
+
+func (s *escSeqParser) IsOpen() bool {
+	return len(s.codes) > 0
+}
+
+func (s *escSeqParser) Reset() {
+	s.inEscSeq = false
+	s.escSeqKind = escSeqKindUnknown
+	s.escapeSeq = ""
+}
+
+const (
+	escCodeResetAll        = 0
+	escCodeResetIntensity  = 22
+	escCodeResetItalic     = 23
+	escCodeResetUnderline  = 24
+	escCodeResetBlink      = 25
+	escCodeResetReverse    = 27
+	escCodeResetCrossedOut = 29
+	escCodeBold            = 1
+	escCodeDim             = 2
+	escCodeItalic          = 3
+	escCodeUnderline       = 4
+	escCodeBlinkSlow       = 5
+	escCodeBlinkRapid      = 6
+	escCodeReverse         = 7
+	escCodeConceal         = 8
+	escCodeCrossedOut      = 9
+)
+
+func (s *escSeqParser) ParseSeq(seq string, seqKind escSeqKind) {
+	if s.codes == nil {
+		s.codes = make(map[int]bool)
+	}
+
+	if seqKind == escSeqKindOSI {
+		seq = strings.Replace(seq, EscapeStartOSI, "", 1)
+		seq = strings.Replace(seq, EscapeStopOSI, "", 1)
+	} else { // escSeqKindCSI
+		seq = strings.Replace(seq, EscapeStartCSI, "", 1)
+		seq = strings.Replace(seq, EscapeStopCSI, "", 1)
+	}
+
+	codes := strings.Split(seq, ";")
+	for _, code := range codes {
+		code = strings.TrimSpace(code)
+		if codeNum, err := strconv.Atoi(code); err == nil {
+			switch codeNum {
+			case escCodeResetAll:
+				s.codes = make(map[int]bool) // clear everything
+			case escCodeResetIntensity:
+				delete(s.codes, escCodeBold)
+				delete(s.codes, escCodeDim)
+			case escCodeResetItalic:
+				delete(s.codes, escCodeItalic)
+			case escCodeResetUnderline:
+				delete(s.codes, escCodeUnderline)
+			case escCodeResetBlink:
+				delete(s.codes, escCodeBlinkSlow)
+				delete(s.codes, escCodeBlinkRapid)
+			case escCodeResetReverse:
+				delete(s.codes, escCodeReverse)
+			case escCodeResetCrossedOut:
+				delete(s.codes, escCodeCrossedOut)
+			default:
+				s.codes[codeNum] = true
+			}
+		}
+	}
+}
+
+func (s *escSeqParser) ParseString(str string) string {
+	s.escapeSeq, s.inEscSeq, s.escSeqKind = "", false, escSeqKindUnknown
+	for _, char := range str {
+		s.Consume(char)
+	}
+	return s.Sequence()
+}
+
+func (s *escSeqParser) Sequence() string {
+	out := strings.Builder{}
+	if s.IsOpen() {
+		out.WriteString(EscapeStart)
+		for idx, code := range s.Codes() {
+			if idx > 0 {
+				out.WriteRune(';')
+			}
+			out.WriteString(fmt.Sprint(code))
+		}
+		out.WriteString(EscapeStop)
+	}
+
+	return out.String()
+}
+
+const (
+	escapeStartConcealOSI = "\x1b]8;"
+	escapeStopConcealOSI  = "\x1b\\"
+)
+
+func (s *escSeqParser) isEscapeStopRune(char rune) bool {
+	if strings.HasPrefix(s.escapeSeq, escapeStartConcealOSI) {
+		if strings.HasSuffix(s.escapeSeq, escapeStopConcealOSI) {
+			return true
+		}
+	} else if (s.escSeqKind == escSeqKindCSI && char == EscapeStopRuneCSI) ||
+		(s.escSeqKind == escSeqKindOSI && char == EscapeStopRuneOSI) {
+		return true
+	}
+	return false
+}

text/escape_seq_parser_test.go

@@ -0,0 +1,74 @@
+package text
+
+import (
+	"github.com/stretchr/testify/assert"
+	"testing"
+)
+
+func Test_escSeqParser(t *testing.T) {
+	t.Run("extract csi", func(t *testing.T) {
+		es := escSeqParser{}
+
+		assert.Equal(t, "\x1b[1;3;4;5;7;9;91m", es.ParseString("\x1b[91m\x1b[1m\x1b[3m\x1b[4m\x1b[5m\x1b[7m\x1b[9m Spicy"))
+		assert.Equal(t, "\x1b[3;4;5;7;9;91m", es.ParseString("\x1b[22m No Bold"))
+		assert.Equal(t, "\x1b[4;5;7;9;91m", es.ParseString("\x1b[23m No Italic"))
+		assert.Equal(t, "\x1b[5;7;9;91m", es.ParseString("\x1b[24m No Underline"))
+		assert.Equal(t, "\x1b[7;9;91m", es.ParseString("\x1b[25m No Blink"))
+		assert.Equal(t, "\x1b[9;91m", es.ParseString("\x1b[27m No Reverse"))
+		assert.Equal(t, "\x1b[91m", es.ParseString("\x1b[29m No Crossed-Out"))
+		assert.Equal(t, "", es.ParseString("\x1b[0m Resetted"))
+	})
+
+	t.Run("extract osi", func(t *testing.T) {
+		es := escSeqParser{}
+
+		assert.Equal(t, "\x1b[1;3;4;5;7;9;91m", es.ParseString("\x1b]91\\\x1b]1\\\x1b]3\\\x1b]4\\\x1b]5\\\x1b]7\\\x1b]9\\ Spicy"))
+		assert.Equal(t, "\x1b[3;4;5;7;9;91m", es.ParseString("\x1b]22\\ No Bold"))
+		assert.Equal(t, "\x1b[4;5;7;9;91m", es.ParseString("\x1b]23\\ No Italic"))
+		assert.Equal(t, "\x1b[5;7;9;91m", es.ParseString("\x1b]24\\ No Underline"))
+		assert.Equal(t, "\x1b[7;9;91m", es.ParseString("\x1b]25\\ No Blink"))
+		assert.Equal(t, "\x1b[9;91m", es.ParseString("\x1b]27\\ No Reverse"))
+		assert.Equal(t, "\x1b[91m", es.ParseString("\x1b]29\\ No Crossed-Out"))
+		assert.Equal(t, "", es.ParseString("\x1b[0m Resetted"))
+	})
+
+	t.Run("parse csi", func(t *testing.T) {
+		es := escSeqParser{}
+
+		es.ParseSeq("\x1b[91m", escSeqKindCSI) // color
+		es.ParseSeq("\x1b[1m", escSeqKindCSI)  // bold
+		assert.Len(t, es.Codes(), 2)
+		assert.True(t, es.IsOpen())
+		assert.Equal(t, "\x1b[1;91m", es.Sequence())
+
+		es.ParseSeq("\x1b[22m", escSeqKindCSI) // un-bold
+		assert.Len(t, es.Codes(), 1)
+		assert.True(t, es.IsOpen())
+		assert.Equal(t, "\x1b[91m", es.Sequence())
+
+		es.ParseSeq("\x1b[0m", escSeqKindCSI) // reset
+		assert.Empty(t, es.Codes())
+		assert.False(t, es.IsOpen())
+		assert.Empty(t, es.Sequence())
+	})
+
+	t.Run("parse osi", func(t *testing.T) {
+		es := escSeqParser{}
+
+		es.ParseSeq("\x1b]91\\", escSeqKindOSI) // color
+		es.ParseSeq("\x1b]1\\", escSeqKindOSI)  // bold
+		assert.Len(t, es.Codes(), 2)
+		assert.True(t, es.IsOpen())
+		assert.Equal(t, "\x1b[1;91m", es.Sequence())
+
+		es.ParseSeq("\x1b]22\\", escSeqKindOSI) // un-bold
+		assert.Len(t, es.Codes(), 1)
+		assert.True(t, es.IsOpen())
+		assert.Equal(t, "\x1b[91m", es.Sequence())
+
+		es.ParseSeq("\x1b]0\\", escSeqKindOSI) // reset
+		assert.Empty(t, es.Codes())
+		assert.False(t, es.IsOpen())
+		assert.Empty(t, es.Sequence())
+	})
+}