Skip to content

Add support for Jujutsu conflict markers #64750

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Draft
Ryan Fitzgerald (rf-) wants to merge 1 commit into
base: main
Choose a base branch
from
+857 −69
Draft

Add support for Jujutsu conflict markers #64750

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
224 changes: 155 additions & 69 deletions tsc/internal/scanner/scanner.go
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -198,6 +198,11 @@ var textToToken = func() map[string]ast.Kind {
return m
}()

type ConflictState struct {
markerLength int
scanNextSnapshot bool
}

type ScannerState struct {
pos int // Current position in text (and ending position of current token)
fullStartPos int // Starting position of current token including preceding whitespace
Expand All @@ -207,6 +212,7 @@ type ScannerState struct {
tokenFlags ast.TokenFlags // Flags for current token
commentDirectives []ast.CommentDirective
skipJSDocLeadingAsterisks int // Leading asterisks to skip when scanning types inside JSDoc. Should be 0 outside JSDoc
conflictState *ConflictState
}

type Scanner struct {
Expand Down Expand Up @@ -530,6 +536,15 @@ func (s *Scanner) Scan() ast.Kind {
case '`':
s.token = s.scanTemplateAndSetTokenValue(false /*shouldEmitInvalidEscapeError*/)
case '%':
if s.charAt(1) == '%' && isConflictMarkerTrivia(s.text, s.pos, s.conflictState) {
s.pos, s.conflictState = scanConflictMarkerTrivia(s.text, s.pos, s.conflictState, s.errorAt)
if s.skipTrivia {
continue
} else {
s.token = ast.KindConflictMarkerTrivia
return s.token
}
}
if s.charAt(1) == '=' {
s.pos += 2
s.token = ast.KindPercentEqualsToken
Expand Down Expand Up @@ -584,6 +599,15 @@ func (s *Scanner) Scan() ast.Kind {
s.token = ast.KindAsteriskToken
}
case '+':
if s.charAt(1) == '+' && isConflictMarkerTrivia(s.text, s.pos, s.conflictState) {
s.pos, s.conflictState = scanConflictMarkerTrivia(s.text, s.pos, s.conflictState, s.errorAt)
if s.skipTrivia {
continue
} else {
s.token = ast.KindConflictMarkerTrivia
return s.token
}
}
next := s.charAt(1)
if next == '=' {
s.pos += 2
Expand All @@ -599,6 +623,15 @@ func (s *Scanner) Scan() ast.Kind {
s.pos++
s.token = ast.KindCommaToken
case '-':
if s.charAt(1) == '-' && isConflictMarkerTrivia(s.text, s.pos, s.conflictState) {
s.pos, s.conflictState = scanConflictMarkerTrivia(s.text, s.pos, s.conflictState, s.errorAt)
if s.skipTrivia {
continue
} else {
s.token = ast.KindConflictMarkerTrivia
return s.token
}
}
next := s.charAt(1)
if next == '=' {
s.pos += 2
Expand Down Expand Up @@ -764,8 +797,8 @@ func (s *Scanner) Scan() ast.Kind {
s.pos++
s.token = ast.KindSemicolonToken
case '<':
if s.charAt(1) == '<' && isConflictMarkerTrivia(s.text, s.pos) {
s.pos = scanConflictMarkerTrivia(s.text, s.pos, s.errorAt)
if s.charAt(1) == '<' && isConflictMarkerTrivia(s.text, s.pos, s.conflictState) {
s.pos, s.conflictState = scanConflictMarkerTrivia(s.text, s.pos, s.conflictState, s.errorAt)
if s.skipTrivia {
continue
} else {
Expand All @@ -792,8 +825,8 @@ func (s *Scanner) Scan() ast.Kind {
s.token = ast.KindLessThanToken
}
case '=':
if s.charAt(1) == '=' && isConflictMarkerTrivia(s.text, s.pos) {
s.pos = scanConflictMarkerTrivia(s.text, s.pos, s.errorAt)
if s.charAt(1) == '=' && isConflictMarkerTrivia(s.text, s.pos, s.conflictState) {
s.pos, s.conflictState = scanConflictMarkerTrivia(s.text, s.pos, s.conflictState, s.errorAt)
if s.skipTrivia {
continue
} else {
Expand All @@ -817,8 +850,8 @@ func (s *Scanner) Scan() ast.Kind {
s.token = ast.KindEqualsToken
}
case '>':
if s.charAt(1) == '>' && isConflictMarkerTrivia(s.text, s.pos) {
s.pos = scanConflictMarkerTrivia(s.text, s.pos, s.errorAt)
if s.charAt(1) == '>' && isConflictMarkerTrivia(s.text, s.pos, s.conflictState) {
s.pos, s.conflictState = scanConflictMarkerTrivia(s.text, s.pos, s.conflictState, s.errorAt)
if s.skipTrivia {
continue
} else {
Expand Down Expand Up @@ -862,8 +895,8 @@ func (s *Scanner) Scan() ast.Kind {
s.pos++
s.token = ast.KindOpenBraceToken
case '|':
if s.charAt(1) == '|' && isConflictMarkerTrivia(s.text, s.pos) {
s.pos = scanConflictMarkerTrivia(s.text, s.pos, s.errorAt)
if s.charAt(1) == '|' && isConflictMarkerTrivia(s.text, s.pos, s.conflictState) {
s.pos, s.conflictState = scanConflictMarkerTrivia(s.text, s.pos, s.conflictState, s.errorAt)
if s.skipTrivia {
continue
} else {
Expand Down Expand Up @@ -896,6 +929,15 @@ func (s *Scanner) Scan() ast.Kind {
s.pos++
s.token = ast.KindAtToken
case '\\':
if s.charAt(1) == '\\' && isConflictMarkerTrivia(s.text, s.pos, s.conflictState) {
s.pos, s.conflictState = scanConflictMarkerTrivia(s.text, s.pos, s.conflictState, s.errorAt)
if s.skipTrivia {
continue
} else {
s.token = ast.KindConflictMarkerTrivia
return s.token
}
}
if s.scanIdentifier(0, identifierVariantStandard) {
s.token = GetIdentifierToken(s.tokenValue)
} else {
Expand Down Expand Up @@ -1276,8 +1318,8 @@ func (s *Scanner) ScanJsxTokenEx(allowMultilineJsxText bool) ast.Kind {
break
}
if ch == '<' {
if isConflictMarkerTrivia(s.text, s.pos) {
s.pos = scanConflictMarkerTrivia(s.text, s.pos, s.errorAt)
if isConflictMarkerTrivia(s.text, s.pos, s.conflictState) {
s.pos, s.conflictState = scanConflictMarkerTrivia(s.text, s.pos, s.conflictState, s.errorAt)
Comment on lines +1321 to +1322

@rf- Ryan Fitzgerald (rf-) Oct 26, 2026 •
edited
Loading

Copy link
Copy Markdown
Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Is this related to this PR? If the existing JSX scanner doesn't handle intermediate markers properly in general, then fixing that doesn't seem in scope here 🤔

s.token = ast.KindConflictMarkerTrivia
return s.token
}
Expand Down Expand Up @@ -2282,22 +2324,6 @@ func GetViableKeywordSuggestions() []string {
return result
}

func couldStartTrivia(text string, pos int) bool {
// Keep in sync with skipTrivia
switch ch := text[pos]; ch {
// Characters that could start normal trivia
case '\r', '\n', '\t', '\v', '\f', ' ', '/',
// Characters that could start conflict marker trivia
'<', '|', '=', '>':
return true
case '#':
// Only if its the beginning can we have #! trivia
return pos == 0
default:
return ch > maxAsciiCharacter
}
}

type SkipTriviaOptions struct {
StopAfterLineBreak bool
StopAtComments bool
Expand All @@ -2318,7 +2344,7 @@ func SkipTriviaEx(text string, pos int, options *SkipTriviaOptions) int {

textLen := len(text)
canConsumeStar := false
// Keep in sync with couldStartTrivia
var conflictState *ConflictState
for {
if pos >= textLen {
return pos
Expand Down Expand Up @@ -2371,9 +2397,9 @@ func SkipTriviaEx(text string, pos int, options *SkipTriviaOptions) int {
continue
}
}
case '<', '|', '=', '>':
if isConflictMarkerTrivia(text, pos) {
pos = scanConflictMarkerTrivia(text, pos, nil)
case '<', '|', '=', '%', '\\', '+', '-', '>':
if isConflictMarkerTrivia(text, pos, conflictState) {
pos, conflictState = scanConflictMarkerTrivia(text, pos, conflictState, nil)
canConsumeStar = false
continue
}
Expand All @@ -2399,21 +2425,21 @@ func SkipTriviaEx(text string, pos int, options *SkipTriviaOptions) int {
}
}

// All conflict markers consist of the same character repeated seven times. If it is
// a <<<<<<< or >>>>>>> marker then it is also followed by a space.
// All conflict markers consist of the same character repeated seven or more times,
// followed by either a space or a newline.
var (
mergeConflictMarkerLength = len("<<<<<<<")
maxAsciiCharacter byte = 127
mergeConflictMarkerMinLength = len("<<<<<<<")
maxAsciiCharacter byte = 127
)

func isConflictMarkerTrivia(text string, pos int) bool {
func isConflictMarkerTrivia(text string, pos int, conflictState *ConflictState) bool {
if pos < 0 {
panic("pos < 0")
}

// Fast reject: a conflict marker is the same byte repeated seven times. If the
// second byte differs (the overwhelmingly common case for `<`, `>`, `=`, `|`
// tokens), it cannot be a marker, so skip the line-start check entirely.
// Fast reject: a conflict marker is the same byte repeated at least seven times. If
// the second byte differs (the overwhelmingly common case), it cannot be a marker, so
// skip the line-start check entirely.
if pos+1 >= len(text) || text[pos+1] != text[pos] {
return false
}
Expand All @@ -2424,52 +2450,112 @@ func isConflictMarkerTrivia(text string, pos int) bool {
prev, _ := utf8.DecodeLastRuneInString(text[:pos-2])
atLineStart = stringutil.IsLineBreak(prev)
}
if atLineStart {
ch := text[pos]
if !atLineStart {
return false
}

if (pos + mergeConflictMarkerLength) < len(text) {
for i := range mergeConflictMarkerLength {
if text[pos+i] != ch {
return false
}
}
ch := text[pos]
length := 0
for pos+length < len(text) && text[pos+length] == ch {
length += 1
}

return ch == '=' || text[pos+mergeConflictMarkerLength] == ' '
}
if length < mergeConflictMarkerMinLength {
return false
}

if conflictState != nil && length != conflictState.markerLength {
return false
}

if pos+length >= len(text) {
return true
}

if text[pos+length] == ' ' {
return true
}

nextChar, _ := utf8.DecodeRuneInString(text[pos+length:])
if stringutil.IsLineBreak(nextChar) {
return true
}

return false
}

func scanConflictMarkerTrivia(text string, pos int, reportError func(diag *diagnostics.Message, pos int, length int, args ...any)) int {
if reportError != nil {
reportError(diagnostics.Merge_conflict_marker_encountered, pos, mergeConflictMarkerLength)
}
func scanConflictMarkerTrivia(text string, pos int, conflictState *ConflictState, reportError func(diag *diagnostics.Message, pos int, length int, args ...any)) (int, *ConflictState) {
ch, size := utf8.DecodeRuneInString(text[pos:])
length := len(text)

if ch == '<' || ch == '>' {
for pos < length && !stringutil.IsLineBreak(ch) {
pos += size
ch, size = utf8.DecodeRuneInString(text[pos:])
if ch != '<' && ch != '>' && ch != '|' && ch != '=' && ch != '%' && ch != '\\' && ch != '+' && ch != '-' {
panic("Assertion failed: unexpected marker char")
}

markerChar := ch
startPos := pos

// First, consume and measure the marker itself and report the error.
for pos < length && ch == markerChar {
pos += size
ch, size = utf8.DecodeRuneInString(text[pos:])
}
markerLength := pos - startPos

if reportError != nil {
reportError(diagnostics.Merge_conflict_marker_encountered, startPos, markerLength)
}

// Then consume up to and including the next line break.
for pos < length && !stringutil.IsLineBreak(ch) {
pos += size
ch, size = utf8.DecodeRuneInString(text[pos:])
}
if pos < length && ch == '\r' {
pos += size
ch, size = utf8.DecodeRuneInString(text[pos:])
}
if pos < length && ch == '\n' {
pos += size
ch, _ = utf8.DecodeRuneInString(text[pos:])
}

if markerChar == '<' {
conflictState = new(ConflictState)
conflictState.markerLength = markerLength

// If the next line starts with a % or + marker, this is a Jujutsu
// conflict and we should remember to scan the next snapshot section.
// Otherwise, it's a Git conflict and we just need to scan the section
// immediately following this marker line.
if (ch == '%' || ch == '+') && isConflictMarkerTrivia(text, pos, conflictState) {
conflictState.scanNextSnapshot = true
}
} else {
if ch != '|' && ch != '=' {
panic("Assertion failed: ch must be either '|' or '='")
}
// Consume everything from the start of a ||||||| or ======= marker to the start
// of the next ======= or >>>>>>> marker.
for pos < length {
currentChar := text[pos]
if (currentChar == '=' || currentChar == '>') && rune(currentChar) != ch && isConflictMarkerTrivia(text, pos) {
break
}

pos++
return pos, conflictState
}

if markerChar == '+' && conflictState != nil && conflictState.scanNextSnapshot {
nextState := *conflictState
nextState.scanNextSnapshot = false
return pos, &nextState
}

if markerChar == '>' {
return pos, nil
}

// Now that we've established we don't want to scan this section, consume
// everything up to the start of the next marker.
for pos < length {

@rf- Ryan Fitzgerald (rf-) Oct 26, 2026

Copy link
Copy Markdown
Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Hmm, OK. This seems very edge casey IIUC, and it's still not solving the broader issue that SkipTriviaEx doesn't know the expected marker length if it starts halfway through a conflict area, but it seems reasonable to initialize the state if we start at an intermediate marker.

@rf- Ryan Fitzgerald (rf-) Oct 26, 2026

Copy link
Copy Markdown
Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Actually the hard part here is the test case—SkipTriviaEx itself doesn't have tests, and I don't know what call site would make sense to try to test this from. It would probably make sense to use one where this situation is likely to come up and would have a meaningful impact, but I'm not sure if that exists.

currentChar := text[pos]
if (currentChar == '=' || currentChar == '%' || currentChar == '\\' || currentChar == '+' || currentChar == '-' || currentChar == '>') && isConflictMarkerTrivia(text, pos, conflictState) {
break
}
pos++
}

return pos
return pos, conflictState
}

func isShebangTrivia(text string, pos int) bool {
Expand Down
36 changes: 36 additions & 0 deletions tsc/testdata/baselines/reference/compiler/conflictMarkerJujutsuTrivia1.errors.txt
View file
Open in desktop
Original file line number Diff line number Diff line change
@@ -0,0 +1,36 @@
conflictMarkerJujutsuTrivia1.ts(3,1): error TS1185: Merge conflict marker encountered.
conflictMarkerJujutsuTrivia1.ts(4,1): error TS1185: Merge conflict marker encountered.
conflictMarkerJujutsuTrivia1.ts(5,1): error TS1185: Merge conflict marker encountered.
conflictMarkerJujutsuTrivia1.ts(8,1): error TS1185: Merge conflict marker encountered.
conflictMarkerJujutsuTrivia1.ts(9,6): error TS2304: Cannot find name 'b'.
conflictMarkerJujutsuTrivia1.ts(10,1): error TS1185: Merge conflict marker encountered.


==== conflictMarkerJujutsuTrivia1.ts (6 errors) ====
class C {
foo() {
<<<<<<< conflict 1 of 1
~~~~~~~
!!! error TS1185: Merge conflict marker encountered.
%%%%%%% diff from: puqzxvrm 9be6077e "call c()"
~~~~~~~
!!! error TS1185: Merge conflict marker encountered.
\\\\\\\ to: lnqukurm 3504bbcc "call a()"
~~~~~~~
!!! error TS1185: Merge conflict marker encountered.
- c();
+ a();
+++++++ uxvuvmmx 392ce8b5 "call b()"
~~~~~~~
!!! error TS1185: Merge conflict marker encountered.
b();
~
!!! error TS2304: Cannot find name 'b'.
>>>>>>> conflict 1 of 1 ends
~~~~~~~
!!! error TS1185: Merge conflict marker encountered.
}

public bar() { }
}

Loading