Diagnose source conflict markers

Author: ahoppen

Sources/SwiftParser/Lexer/Cursor.swift

@@ -1027,7 +1027,7 @@ extension Lexer.Cursor {
       let start = self
 
       switch self.advance() {
-      // 'continue' - the character is a part of the trivia9.
+      // 'continue' - the character is a part of the trivia.
       // 'break' - the character should a part of token text.
       case nil:
         break
@@ -1070,10 +1070,10 @@ extension Lexer.Cursor {
         self.advanceToEndOfLine()
         continue
       case UInt8(ascii: "<"), UInt8(ascii: ">"):
-        guard self.tryLexConflictMarker(start: start) else {
-          break
+        if self.tryLexConflictMarker(start: start) {
+          error = (.sourceConflictMarker, start)
+          continue
         }
-        continue
       // Start character of tokens.
       //        case (char)-1: case (char)-2:
       case  // Punctuation.
@@ -2195,7 +2195,7 @@ extension Lexer.Cursor {
   }
   mutating func tryLexConflictMarker(start: Lexer.Cursor) -> Bool {
     // Only a conflict marker if it starts at the beginning of a line.
-    guard start.previous == UInt8(ascii: "\n") || start.previous == UInt8(ascii: "\r") else {
+    guard start.previous == UInt8(ascii: "\n") || start.previous == UInt8(ascii: "\r") || start.previous == 0 else {
       return false
     }
 
@@ -2211,7 +2211,6 @@ extension Lexer.Cursor {
     }
 
     // Diagnose at the conflict marker, then jump ahead to the end.
-    //    diagnose(CurPtr, diag::lex_conflict_marker_in_file);
     self = end
 
     // Skip ahead to the end of the marker.

Sources/SwiftParserDiagnostics/LexerDiagnosticMessages.swift

@@ -51,6 +51,7 @@ public enum StaticLexerError: String, DiagnosticMessage {
   case lexerErrorOffsetOverflow = "the lexer dicovered an error in this token but was not able to represent its offset due to overflow; please split the token"
   case nonBreakingSpace = "non-breaking space (U+00A0) used instead of regular space"
   case nulCharacter = "nul character embedded in middle of file"
+  case sourceConflictMarker = "source control conflict marker in source file"
   case unexpectedBlockCommentEnd = "unexpected end of block comment"
   case unicodeCurlyQuote = #"unicode curly quote found; use '"' instead"#
 
@@ -142,6 +143,7 @@ public extension SwiftSyntax.LexerError {
     case .lexerErrorOffsetOverflow: return StaticLexerError.lexerErrorOffsetOverflow
     case .nonBreakingSpace: return StaticLexerError.nonBreakingSpace
     case .nulCharacter: return StaticLexerError.nulCharacter
+    case .sourceConflictMarker: return StaticLexerError.sourceConflictMarker
     case .unexpectedBlockCommentEnd: return StaticLexerError.unexpectedBlockCommentEnd
     case .unicodeCurlyQuote: return StaticLexerError.unicodeCurlyQuote
     }

Sources/SwiftSyntax/LexerError.swift

@@ -38,6 +38,7 @@ public struct LexerError: Hashable {
     case lexerErrorOffsetOverflow
     case nonBreakingSpace
     case nulCharacter
+    case sourceConflictMarker
     case unexpectedBlockCommentEnd
     case unicodeCurlyQuote
   }

Tests/SwiftParserTest/LexerTests.swift

@@ -453,7 +453,7 @@ public class LexerTests: XCTestCase {
       """
       // diff3-style conflict markers
 
-      <<<<<<< HEAD:conflict_markers.swift // expected-error {{source control conflict marker in source file}}
+      1️⃣<<<<<<< HEAD:conflict_markers.swift // expected-error {{source control conflict marker in source file}}
       var a : String = "A"
       var b : String = "b"
       =======
@@ -477,6 +477,7 @@ public class LexerTests: XCTestCase {
             >>>>>>> 18844bc65229786b96b89a9fc7739c0fc897905e:conflict_markers.swift
             """,
           text: "",
+          error: "source control conflict marker in source file",
           flags: [.isAtStartOfLine]
         )
       ]
@@ -486,7 +487,7 @@ public class LexerTests: XCTestCase {
       """
       // Perforce-style conflict markers
 
-      >>>> ORIGINAL
+      1️⃣>>>> ORIGINAL
       var a : String = "A"
       var b : String = "B"
       ==== THEIRS
@@ -518,6 +519,7 @@ public class LexerTests: XCTestCase {
 
             """,
           text: "",
+          error: "source control conflict marker in source file",
           flags: [.isAtStartOfLine]
         )
       ]

Tests/SwiftParserTest/translated/ConflictMarkersTests.swift

@@ -73,18 +73,14 @@ final class ConflictMarkersTests: XCTestCase {
       1️⃣<<<<<<< HEAD:conflict_markers.swift 
       var a : String = "A"
       var b : String = "b"
-      =======2️⃣
+      =======
       var a : String = "a"
       var b : String = "B"
-      >>>>>>> 188443️⃣bc65229786b96b89a9fc7739c0fc897905e4️⃣:conflict_markers.swift
+      >>>>>>> 18844bc65229786b96b89a9fc7739c0fc897905e:conflict_markers.swift
       print(a + b)
       """#,
       diagnostics: [
-        // TODO: Old parser expected error on line 1: source control conflict marker in source file
-        DiagnosticSpec(locationMarker: "1️⃣", message: "unexpected code '<<<<<<< HEAD:conflict_markers.swift' before variable"),
-        DiagnosticSpec(locationMarker: "2️⃣", message: "expected expression after operator"),
-        DiagnosticSpec(locationMarker: "3️⃣", message: "'b' is not a valid digit in integer literal"),
-        DiagnosticSpec(locationMarker: "4️⃣", message: "extraneous code at top level"),
+        DiagnosticSpec(message: "source control conflict marker in source file")
       ]
     )
   }
@@ -95,14 +91,11 @@ final class ConflictMarkersTests: XCTestCase {
       1️⃣<<<<<<< HEAD:conflict_markers.swift 
       ======= 
       var d : String = "D"
-      >>>>>>> 188442️⃣bc65229786b96b89a9fc7739c0fc897905e3️⃣:conflict_markers.swift
+      >>>>>>> 18844bc65229786b96b89a9fc7739c0fc897905e:conflict_markers.swift
       print(d)
       """#,
       diagnostics: [
-        // TODO: Old parser expected error on line 1: source control conflict marker in source file
-        DiagnosticSpec(locationMarker: "1️⃣", message: "unexpected code before variable"),
-        DiagnosticSpec(locationMarker: "2️⃣", message: "'b' is not a valid digit in integer literal"),
-        DiagnosticSpec(locationMarker: "3️⃣", message: "extraneous code at top level"),
+        DiagnosticSpec(message: "source control conflict marker in source file")
       ]
     )
   }
@@ -119,22 +112,19 @@ final class ConflictMarkersTests: XCTestCase {
   func testConflictMarkers10() {
     AssertParse(
       #"""
-      1️⃣<<<<<<< HEAD:conflict_markers.swift 
+      1️⃣<<<<<<< HEAD:conflict_markers.swift
       <<<<<<<"HEAD:fake_conflict_markers.swift"
       var fake_b : String = "a"
       >>>>>>>"18844bc65229786b96b89a9fc7739c0fc897905e:fake_conflict_markers.swift"
       =======
       <<<<<<<"HEAD:fake_conflict_markers.swift"
       var fake_c : String = "a"
       >>>>>>>"18844bc65229786b96b89a9fc7739c0fc897905e:fake_conflict_markers.swift"
-      >>>>>>> 188442️⃣bc65229786b96b89a9fc7739c0fc897905e3️⃣:conflict_markers.swift
+      >>>>>>> 18844bc65229786b96b89a9fc7739c0fc897905e:conflict_markers.swift
       print(fake_b + fake_c)
       """#,
       diagnostics: [
-        // TODO: Old parser expected error on line 1: source control conflict marker in source file
-        DiagnosticSpec(locationMarker: "1️⃣", message: "unexpected code before variable"),
-        DiagnosticSpec(locationMarker: "2️⃣", message: "'b' is not a valid digit in integer literal"),
-        DiagnosticSpec(locationMarker: "3️⃣", message: "extraneous code at top level"),
+        DiagnosticSpec(message: "source control conflict marker in source file")
       ]
     )
   }
@@ -154,7 +144,7 @@ final class ConflictMarkersTests: XCTestCase {
       // Conflict marker.
       let a = "a", b = "b"
       a 
-      <<<<<<< b 
+      1️⃣<<<<<<< b
       a
       >>>>>>> b
       // Not a conflict marker.
@@ -165,7 +155,7 @@ final class ConflictMarkersTests: XCTestCase {
       }()
       """#,
       diagnostics: [
-        // TODO: Old parser expected error on line 5: source control conflict marker in source file
+        DiagnosticSpec(message: "source control conflict marker in source file")
       ]
     )
   }
@@ -194,8 +184,7 @@ final class ConflictMarkersTests: XCTestCase {
       print(a + b)
       """#,
       diagnostics: [
-        // TODO: Old parser expected error on line 1: source control conflict marker in source file
-        DiagnosticSpec(message: "unexpected code '>>>> ORIGINAL' before variable")
+        DiagnosticSpec(message: "source control conflict marker in source file")
       ]
     )
   }
@@ -211,8 +200,7 @@ final class ConflictMarkersTests: XCTestCase {
       print(d)
       """#,
       diagnostics: [
-        // TODO: Old parser expected error on line 1: source control conflict marker in source file
-        DiagnosticSpec(message: "unexpected code before variable")
+        DiagnosticSpec(message: "source control conflict marker in source file")
       ]
     )
   }