Merge pull request #63048 from lorentey/fix-String-debugDescription

[stdlib] String.debugDescription: Fix quoting behavior

Author: lorentey

stdlib/public/core/String.swift

@@ -676,11 +676,49 @@ extension String: ExpressibleByStringLiteral {
 extension String: CustomDebugStringConvertible {
   /// A representation of the string that is suitable for debugging.
   public var debugDescription: String {
+    func hasBreak(between left: String, and right: Unicode.Scalar) -> Bool {
+      // Note: we know `left` ends with an ASCII character, so we only need to
+      // look at its last scalar.
+      var state = _GraphemeBreakingState()
+      return state.shouldBreak(between: left.unicodeScalars.last!, and: right)
+    }
+
+    // Prevent unquoted scalars in the string from combining with the opening
+    // `"` or the tail of the preceding quoted scalar.
     var result = "\""
+    var wantBreak = true // true if next scalar must not combine with the last
     for us in self.unicodeScalars {
-      result += us.escaped(asASCII: false)
+      if let escaped = us._escaped(asASCII: false) {
+        result += escaped
+        wantBreak = true
+      } else if wantBreak && !hasBreak(between: result, and: us) {
+        result += us.escaped(asASCII: true)
+        wantBreak = true
+      } else {
+        result.unicodeScalars.append(us)
+        wantBreak = false
+      }
+    }
+    // Also prevent the last scalar from combining with the closing `"`.
+    var suffix = "\"".unicodeScalars
+    while !result.isEmpty {
+      // Append first scalar of suffix, then check if it combines.
+      result.unicodeScalars.append(suffix.first!)
+      let i = result.index(before: result.endIndex)
+      let j = result.unicodeScalars.index(before: result.endIndex)
+      if i >= j {
+        // All good; append the rest and we're done.
+        result.unicodeScalars.append(contentsOf: suffix.dropFirst())
+        break
+      }
+      // Cancel appending the scalar, then quote the last scalar in `result` and
+      // prepend it to `suffix`.
+      result.unicodeScalars.removeLast()
+      let last = result.unicodeScalars.removeLast()
+      suffix.insert(
+        contentsOf: last.escaped(asASCII: true).unicodeScalars,
+        at: suffix.startIndex)
     }
-    result += "\""
     return result
   }
 }

stdlib/public/core/UnicodeScalar.swift

@@ -192,6 +192,10 @@ extension Unicode.Scalar :
   ///   ASCII characters; otherwise, pass `false`.
   /// - Returns: A string representation of the scalar.
   public func escaped(asASCII forceASCII: Bool) -> String {
+    _escaped(asASCII: forceASCII) ?? String(self)
+  }
+
+  internal func _escaped(asASCII forceASCII: Bool) -> String? {
     func lowNibbleAsHex(_ v: UInt32) -> String {
       let nibble = v & 15
       if nibble < 10 {
@@ -208,7 +212,7 @@ extension Unicode.Scalar :
     } else if self == "\"" {
       return "\\\""
     } else if _isPrintableASCII {
-      return String(self)
+      return nil
     } else if self == "\0" {
       return "\\0"
     } else if self == "\n" {
@@ -222,7 +226,7 @@ extension Unicode.Scalar :
         + lowNibbleAsHex(UInt32(self) >> 4)
         + lowNibbleAsHex(UInt32(self)) + "}"
     } else if !forceASCII {
-      return String(self)
+      return nil
     } else if UInt32(self) <= 0xFFFF {
       var result = "\\u{"
       result += lowNibbleAsHex(UInt32(self) >> 12)

test/stdlib/PrintString.swift

@@ -29,16 +29,60 @@ PrintTests.test("Printable") {
 
   let us1: UnicodeScalar = "\\"
   expectPrinted("\\", us1)
-  expectEqual("\"\\\\\"", us1.description)
+  expectEqual("\\", us1.description)
   expectDebugPrinted("\"\\\\\"", us1)
 
   let us2: UnicodeScalar = "あ"
   expectPrinted("あ", us2)
-  expectEqual("\"あ\"", us2.description)
+  expectEqual("あ", us2.description)
   expectDebugPrinted("\"\\u{3042}\"", us2)
 }
 
-PrintTests.test("Printable") {
+PrintTests.test("TrickyQuoting") {
+  guard #available(SwiftStdlib 5.9, *) else { return }
+  // U+301: COMBINING ACUTE ACCENT (Grapheme_Cluster_Break = Extend)
+  let s1 = "\u{301}Foo"
+  expectPrinted(s1, s1)
+  expectDebugPrinted("\"\\u{0301}Foo\"", s1)
+
+  // U+302: COMBINING CIRCUMFLEX ACCENT (Grapheme_Cluster_Break = Extend)
+  let s2 = "\u{301}\u{302}Foo"
+  expectPrinted(s2, s2)
+  expectDebugPrinted("\"\\u{0301}\\u{0302}Foo\"", s2)
+
+  let s3 = "Foo\n\u{301}\u{302}Foo"
+  expectPrinted(s3, s3)
+  expectDebugPrinted("\"Foo\\n\\u{0301}\\u{0302}Foo\"", s3)
+
+  // U+200D: ZERO WIDTH JOINER (Grapheme_Cluster_Break = ZWJ)
+  let s4 = "\u{200d}Foo"
+  expectPrinted(s4, s4)
+  expectDebugPrinted("\"\\u{200D}Foo\"", s4)
+
+  // U+110BD: KAITHI NUMBER SIGN (Grapheme_Cluster_Break = Prepend)
+  let s5 = "Foo\u{110BD}"
+  expectPrinted(s5, s5)
+  expectDebugPrinted("\"Foo\\u{000110BD}\"", s5)
+
+  // U+070F: SYRIAC ABBREVIATION MARK (Grapheme_Cluster_Break = Prepend)
+  let s6 = "Foo\u{070F}\u{110BD}"
+  expectPrinted(s6, s6)
+  expectDebugPrinted("\"Foo\\u{070F}\\u{000110BD}\"", s6)
+
+  let s7 = "Foo\u{301}\u{070F}\u{110BD}"
+  expectPrinted(s7, s7)
+  expectDebugPrinted("\"Foo\u{301}\\u{070F}\\u{000110BD}\"", s7)
+
+  let s8 = "Foo\u{301}\u{302}\u{070F}\u{110BD}Foo"
+  expectPrinted(s8, s8)
+  expectDebugPrinted("\"Foo\u{0301}\u{0302}\u{070F}\u{110BD}Foo\"", s8)
+
+  let s9 = "Foo\u{301}"
+  expectPrinted(s9, s9)
+  expectDebugPrinted("\"Foo\u{0301}\"", s9)
+}
+
+PrintTests.test("Optional") {
   expectPrinted("Optional(\"meow\")", String?("meow"))
 }