Diagnose if the last newline of a multi-line string literal is escaped

Author: ahoppen

Sources/SwiftParser/StringLiterals.swift

@@ -86,7 +86,6 @@ fileprivate class StringLiteralExpressionIndentationChecker {
   }
 }
 
-
 extension Parser {
   /// Consumes a raw string delimiter that has the same number of `#` as `openDelimiter`.
   private mutating func parseStringDelimiter(openDelimiter: RawTokenSyntax?) -> (unexpectedBeforeCheckedDelimiter: RawUnexpectedNodesSyntax?, checkedDelimiter: RawTokenSyntax?) {
@@ -185,9 +184,25 @@ extension Parser {
       if lastMiddleSegment.content.tokenText.hasSuffix("\n") {
         // The newline at the end of the last line in the string literal is not part of the represented string.
         // Mark it as trivia.
+        var content = lastMiddleSegment.content.reclassifyAsTrailingTrivia([.newlines(1)], arena: self.arena)
+        var unexpectedBeforeContent: RawTokenSyntax?
+        if content.tokenText.hasSuffix("\\") {
+          // The newline on the last line must not be escaped
+          unexpectedBeforeContent = content
+          content = RawTokenSyntax(
+            missing: .stringSegment,
+            text: SyntaxText(rebasing: content.tokenText[0..<content.tokenText.count - 1]),
+            leadingTriviaPieces: content.leadingTriviaPieces,
+            trailingTriviaPieces: content.trailingTriviaPieces,
+            arena: self.arena
+          )
+        }
+
         middleSegments[middleSegments.count - 1] = .stringSegment(
           RawStringSegmentSyntax(
-            content: lastMiddleSegment.content.reclassifyAsTrailingTrivia([.newlines(1)], arena: self.arena),
+            RawUnexpectedNodesSyntax(combining: lastMiddleSegment.unexpectedBeforeContent, unexpectedBeforeContent, arena: self.arena),
+            content: content,
+            lastMiddleSegment.unexpectedAfterContent,
             arena: self.arena
           )
         )
@@ -261,14 +276,10 @@ extension Parser {
     for (index, segment) in middleSegments.enumerated() {
       switch segment {
       case .stringSegment(var segment):
-        if segment.content.isMissing {
-          // Don't diagnose incorrect indentation for segments that we synthesized
-          break
-        }
-        // We are not considering unexpected and leading trivia for indentation
-        // computation. If these assertions are violated, we can probably lift
-        // them but we would need to check the produce the expected results.
-        assert(segment.unexpectedBeforeContent == nil && segment.content.leadingTriviaByteLength == 0)
+        // We are not considering leading trivia for indentation computation.
+        // If these assertions are violated, we can probably lift them but we
+        // would need to check the produce the expected results.
+        assert(segment.content.leadingTriviaByteLength == 0)
 
         // Re-classify indentation as leading trivia
         if isSegmentOnNewLine {

Sources/SwiftParserDiagnostics/DiagnosticExtensions.swift

@@ -98,7 +98,8 @@ extension FixIt.Changes {
       !firstToken.tokenKind.isPunctuation,
       !previousToken.tokenKind.isPunctuation,
       firstToken.leadingTrivia.isEmpty,
-      (previousToken.presence == .missing ? BasicFormat().visit(previousToken).trailingTrivia : previousToken.trailingTrivia).isEmpty
+      (previousToken.presence == .missing ? BasicFormat().visit(previousToken).trailingTrivia : previousToken.trailingTrivia).isEmpty,
+      leadingTrivia == nil
     {
       /// If neither this nor the previous token are punctionation make sure they
       /// are separated by a space.

Sources/SwiftParserDiagnostics/ParseDiagnosticsGenerator.swift

@@ -712,6 +712,26 @@ public class ParseDiagnosticsGenerator: SyntaxAnyVisitor {
     for (diagnostic, handledNodes) in MultiLineStringLiteralIndentatinDiagnosticsGenerator.diagnose(node) {
       addDiagnostic(diagnostic, handledNodes: handledNodes)
     }
+    if case .stringSegment(let segment) = node.segments.last {
+      if let invalidContent = segment.unexpectedBeforeContent?.onlyToken(where: { $0.text.hasSuffix("\\") }) {
+        let leadingTrivia = segment.content.leadingTrivia
+        let trailingTrivia = segment.content.trailingTrivia
+        let fixIt = FixIt(
+          message: .removeBackslash,
+          changes: [
+            .makePresent(segment.content, leadingTrivia: leadingTrivia, trailingTrivia: trailingTrivia),
+            .makeMissing(invalidContent, transferTrivia: false),
+          ]
+        )
+        addDiagnostic(
+          invalidContent,
+          position: invalidContent.endPositionBeforeTrailingTrivia.advanced(by: -1),
+          .escapedNewlineAtLatlineOfMultiLineStringLiteralNotAllowed,
+          fixIts: [fixIt],
+          handledNodes: [segment.id]
+        )
+      }
+    }
     return .visitChildren
   }
 

Sources/SwiftParserDiagnostics/ParserDiagnosticMessages.swift

@@ -122,6 +122,9 @@ extension DiagnosticMessage where Self == StaticParserError {
   public static var editorPlaceholderInSourceFile: Self {
     .init("editor placeholder in source file")
   }
+  public static var escapedNewlineAtLatlineOfMultiLineStringLiteralNotAllowed: Self {
+    .init("escaped newline at the last line of a multi-line string literal is not allowed")
+  }
   public static var expectedExpressionAfterTry: Self {
     .init("expected expression after 'try'")
   }
@@ -415,6 +418,9 @@ extension FixItMessage where Self == StaticParserFixIt {
   public static var joinIdentifiersWithCamelCase: Self {
     .init("join the identifiers together with camel-case")
   }
+  public static var removeBackslash: Self {
+    .init("remove '\'")
+  }
   public static var removeExtraneousDelimiters: Self {
     .init("remove extraneous delimiters")
   }

Sources/SwiftParserDiagnostics/PresenceUtils.swift

@@ -49,7 +49,7 @@ class PresentMaker: SyntaxRewriter {
     if token.presence == .missing {
       let presentToken: TokenSyntax
       let (rawKind, text) = token.tokenKind.decomposeToRaw()
-      if let text = text, !text.isEmpty {
+      if let text = text, (!text.isEmpty || rawKind == .stringSegment) {  // string segments can have empty text
         presentToken = TokenSyntax(token.tokenKind, presence: .present)
       } else {
         let newKind = TokenKind.fromRaw(kind: rawKind, text: rawKind.defaultText.map(String.init) ?? "<#\(rawKind.nameForDiagnostics)#>")

Sources/SwiftSyntax/Raw/RawSyntax.swift

@@ -785,7 +785,7 @@ extension RawSyntax: CustomDebugStringConvertible {
     case .parsedToken(let dat):
       target.write(".parsedToken(")
       target.write(String(describing: dat.tokenKind))
-      target.write(" wholeText=\(dat.tokenText.debugDescription)")
+      target.write(" wholeText=\(dat.wholeText.debugDescription)")
       target.write(" textRange=\(dat.textRange.description)")
     case .materializedToken(let dat):
       target.write(".materializedToken(")

Tests/SwiftParserTest/ExpressionTests.swift

@@ -918,8 +918,8 @@ final class ExpressionTests: XCTestCase {
         StringLiteralExprSyntax(
           openQuote: .multilineStringQuoteToken(leadingTrivia: .spaces(2), trailingTrivia: .newline),
           segments: StringLiteralSegmentsSyntax([
-            .stringSegment(StringSegmentSyntax(leadingTrivia: .spaces(2), content: .stringSegment("line 1\n"))),
-            .stringSegment(StringSegmentSyntax(leadingTrivia: .spaces(2), content: .stringSegment("line 2"), trailingTrivia: .newline)),
+            .stringSegment(StringSegmentSyntax(content: .stringSegment("line 1\n", leadingTrivia: .spaces(2)))),
+            .stringSegment(StringSegmentSyntax(content: .stringSegment("line 2", leadingTrivia: .spaces(2), trailingTrivia: .newline))),
           ]),
           closeQuote: .multilineStringQuoteToken(leadingTrivia: .spaces(2))
         )
@@ -938,8 +938,8 @@ final class ExpressionTests: XCTestCase {
         StringLiteralExprSyntax(
           openQuote: .multilineStringQuoteToken(leadingTrivia: .spaces(2), trailingTrivia: .newline),
           segments: StringLiteralSegmentsSyntax([
-            .stringSegment(StringSegmentSyntax(leadingTrivia: .spaces(2), content: .stringSegment("line 1 "), trailingTrivia: [.unexpectedText("\\"), .newlines(1)])),
-            .stringSegment(StringSegmentSyntax(leadingTrivia: .spaces(2), content: .stringSegment("line 2"), trailingTrivia: .newline)),
+            .stringSegment(StringSegmentSyntax(content: .stringSegment("line 1 ", leadingTrivia: .spaces(2), trailingTrivia: [.unexpectedText("\\"), .newlines(1)]))),
+            .stringSegment(StringSegmentSyntax(content: .stringSegment("line 2", leadingTrivia: .spaces(2), trailingTrivia: .newline))),
           ]),
           closeQuote: .multilineStringQuoteToken(leadingTrivia: .spaces(2))
         )
@@ -951,20 +951,28 @@ final class ExpressionTests: XCTestCase {
       #"""
         """
         line 1
-        line 2 \
+        line 2 1️⃣\
         """
       """#,
       substructure: Syntax(
         StringLiteralExprSyntax(
           openQuote: .multilineStringQuoteToken(leadingTrivia: .spaces(2), trailingTrivia: .newline),
           segments: StringLiteralSegmentsSyntax([
-            .stringSegment(StringSegmentSyntax(leadingTrivia: .spaces(2), content: .stringSegment("line 1\n"))),
-            .stringSegment(StringSegmentSyntax(leadingTrivia: .spaces(2), content: .stringSegment("line 2 \\"), trailingTrivia: .newline)),
+            .stringSegment(StringSegmentSyntax(content: .stringSegment("line 1\n", leadingTrivia: .spaces(2)))),
+            .stringSegment(
+              StringSegmentSyntax(
+                UnexpectedNodesSyntax([Syntax(TokenSyntax.stringSegment("  line 2 \\", trailingTrivia: .newline))]),
+                content: .stringSegment("line 2 ", leadingTrivia: .spaces(2), trailingTrivia: .newline, presence: .missing)
+              )
+            ),
           ]),
           closeQuote: .multilineStringQuoteToken(leadingTrivia: .spaces(2))
         )
       ),
-      substructureCheckTrivia: true
+      substructureCheckTrivia: true,
+      diagnostics: [
+        DiagnosticSpec(message: "escaped newline at the last line of a multi-line string literal is not allowed")
+      ]
     )
   }
 }

Tests/SwiftParserTest/translated/MultilineErrorsTests.swift

@@ -468,12 +468,18 @@ final class MultilineErrorsTests: XCTestCase {
       #"""
       _ = """
         line one
-        line two\
+        line two1️⃣\
         """
       """#,
       diagnostics: [
-        // TODO: Old parser expected error on line 3: escaped newline at the last line is not allowed, Fix-It replacements: 11 - 12 = ''
-      ]
+        DiagnosticSpec(message: "escaped newline at the last line of a multi-line string literal is not allowed", fixIts: ["remove '\'"])
+      ],
+      fixedSource: #"""
+        _ = """
+          line one
+          line two
+          """
+        """#
     )
   }
 
@@ -507,25 +513,35 @@ final class MultilineErrorsTests: XCTestCase {
     AssertParse(
       #"""
       _ = """
-        foo\
+        foo1️⃣\
         """
       """#,
       diagnostics: [
-        // TODO: Old parser expected error on line 2: escaped newline at the last line is not allowed, Fix-It replacements: 6 - 7 = ''
-      ]
+        DiagnosticSpec(message: "escaped newline at the last line of a multi-line string literal is not allowed")
+      ],
+      fixedSource: #"""
+        _ = """
+          foo
+          """
+        """#
     )
   }
 
   func testMultilineErrors25() {
     AssertParse(
       #"""
       _ = """
-        foo\
+        foo1️⃣\
         """
       """#,
       diagnostics: [
-        // TODO: Old parser expected error on line 3: escaped newline at the last line is not allowed, Fix-It replacements: 6 - 7 = ''
-      ]
+        DiagnosticSpec(message: "escaped newline at the last line of a multi-line string literal is not allowed")
+      ],
+      fixedSource: #"""
+        _ = """
+          foo
+          """
+        """#
     )
   }
 
@@ -549,12 +565,12 @@ final class MultilineErrorsTests: XCTestCase {
         """
       """#,
       diagnostics: [
-        DiagnosticSpec(message: "insufficient indentation of line in multi-line string literal")
+        DiagnosticSpec(message: "escaped newline at the last line of a multi-line string literal is not allowed")
         // TODO: Old parser expected error on line 2: escaped newline at the last line is not allowed, Fix-It replacements: 1 - 2 = ''
       ],
       fixedSource: #"""
         _ = """
-          \
+          
           """
         """#
     )

Tests/SwiftParserTest/translated/RawStringTests.swift

@@ -74,9 +74,12 @@ final class RawStringTests: XCTestCase {
       ##"""
       _ = #"""
           Three\r
-          Gamma\
+          Gamma1️⃣\
         """#
-      """##
+      """##,
+      diagnostics: [
+        DiagnosticSpec(message: "escaped newline at the last line of a multi-line string literal is not allowed")
+      ]
     )
   }