Diagnose unprintable characters in string literal

Author: ahoppen

Sources/SwiftParser/Lexer/Cursor.swift

@@ -1563,20 +1563,17 @@ extension Lexer.Cursor {
       }
     default:
       _ = self.advance()
-      // Normal characters are part of the string.
-      // If this is a "high" UTF-8 character, validate it.
-      //      if ((signed char)(CurPtr[-1]) >= 0) {
-      //        if (isPrintable(CurPtr[-1]) == 0)
-      //          if (!(IsMultilineString && (CurPtr[-1] == '\t')))
-      //            if (EmitDiagnostics)
-      //              diagnose(CharStart, diag::lex_unprintable_ascii_character)
-      //        return CurPtr[-1]
-      //      }
       self = charStart
       guard let charValue = self.advanceValidatingUTF8Character() else {
         return .error(.invalidUtf8)
       }
-      return .success(charValue)
+      // We disallow non-printable ASCII characters in a string literal, with
+      // the exception of \t, which is valid only in multi-line string literals.
+      if !charValue.isASCII || charValue.isPrintableASCII || stringLiteralKind == .multiLine && charValue == "\t" {
+        return .success(charValue)
+      } else {
+        return .error(.unprintableAsciiCharacter)
+      }
     }
   }
 

Sources/SwiftParserDiagnostics/LexerDiagnosticMessages.swift

@@ -54,6 +54,7 @@ public enum StaticLexerError: String, DiagnosticMessage {
   case sourceConflictMarker = "source control conflict marker in source file"
   case unexpectedBlockCommentEnd = "unexpected end of block comment"
   case unicodeCurlyQuote = #"unicode curly quote found; use '"' instead"#
+  case unprintableAsciiCharacter = "unprintable ASCII character found in source file"
 
   public var message: String { self.rawValue }
 
@@ -146,6 +147,7 @@ public extension SwiftSyntax.LexerError {
     case .sourceConflictMarker: return StaticLexerError.sourceConflictMarker
     case .unexpectedBlockCommentEnd: return StaticLexerError.unexpectedBlockCommentEnd
     case .unicodeCurlyQuote: return StaticLexerError.unicodeCurlyQuote
+    case .unprintableAsciiCharacter: return StaticLexerError.unprintableAsciiCharacter
     }
   }
 

Sources/SwiftSyntax/LexerError.swift

@@ -41,6 +41,7 @@ public struct LexerError: Hashable {
     case sourceConflictMarker
     case unexpectedBlockCommentEnd
     case unicodeCurlyQuote
+    case unprintableAsciiCharacter
   }
 
   public let kind: Kind

Tests/SwiftParserTest/ExpressionTests.swift

@@ -1163,6 +1163,26 @@ final class ExpressionTests: XCTestCase {
       fixedSource: "a  + 2"
     )
   }
+
+  func testTabsIndentationInMultilineStringLiteral() {
+    AssertParse(
+      #"""
+      _ = """
+      \#taq
+      \#t"""
+      """#
+    )
+  }
+
+  func testMixedIndentationInMultilineStringLiteral() {
+    AssertParse(
+      #"""
+      _ = """
+      \#t aq
+      \#t """
+      """#
+    )
+  }
 }
 
 final class MemberExprTests: XCTestCase {

Tests/SwiftParserTest/LexerTests.swift

@@ -1226,4 +1226,16 @@ public class LexerTests: XCTestCase {
     )
   }
 
+  func testUnprintableAsciiCharactersInStringLiteral() {
+    AssertLexemes(
+      """
+      "1️⃣\u{7}"
+      """,
+      lexemes: [
+        LexemeSpec(.stringQuote, text: #"""#),
+        LexemeSpec(.stringSegment, text: "\u{7}", error: "unprintable ASCII character found in source file"),
+        LexemeSpec(.stringQuote, text: #"""#),
+      ]
+    )
+  }
 }