[Parse] Add support for multiline strings inside interpolations

Adds support for multiline string literals to appear inside of string interpolations. Tests added.

Author: johnno1962

include/swift/AST/DiagnosticsParse.def

@@ -134,8 +134,6 @@ ERROR(lex_illegal_multiline_string_end,none,
       "invalid end of multi-line string literal", ())
 ERROR(lex_ambiguous_string_indent,none,
       "invalid mix of multi-line string literal indentation", ())
-ERROR(lex_multiline_inside_interpolation,none,
-      "multi-line string literals are not allowed inside an interpolation", ())
 
 
 ERROR(lex_invalid_character,none,

include/swift/Parse/Lexer.h

@@ -519,9 +519,6 @@ class Lexer {
   /// Try to lex conflict markers by checking for the presence of the start and
   /// end of the marker in diff3 or Perforce style respectively.
   bool tryLexConflictMarker();
-
-  /// Check multiline string literal is indented correctly.
-  void validateMultilineIndents(const Token &Str);
 };
 
 } // end namespace swift

lib/Parse/Lexer.cpp

@@ -1240,7 +1240,8 @@ unsigned Lexer::lexCharacter(const char *&CurPtr, char StopQuote,
 /// outstanding delimiters as it scans the string.
 static const char *skipToEndOfInterpolatedExpression(const char *CurPtr,
                                                      const char *EndPtr,
-                                                     DiagnosticEngine *Diags) {
+                                                     DiagnosticEngine *Diags,
+                                                     bool MultilineString) {
   llvm::SmallVector<char, 4> OpenDelimiters;
   auto inStringLiteral = [&]() {
     return !OpenDelimiters.empty() &&
@@ -1257,9 +1258,11 @@ static const char *skipToEndOfInterpolatedExpression(const char *CurPtr,
     // issues with malformed tokens or other problems.
     switch (*CurPtr++) {
     // String literals in general cannot be split across multiple lines;
-    // interpolated ones are no exception.
+    // interpolated ones are no exception - unless multiline literals.
     case '\n':
     case '\r':
+      if (MultilineString)
+        continue;
       // Will be diagnosed as an unterminated string literal.
       return CurPtr-1;
 
@@ -1272,12 +1275,12 @@ static const char *skipToEndOfInterpolatedExpression(const char *CurPtr,
         }
         // Otherwise it's an ordinary character; treat it normally.
       } else {
-        if (*CurPtr == '"' && *(CurPtr + 1) == '"') {
-          Diags->diagnose(Lexer::getSourceLoc(CurPtr-1),
-                          diag::lex_multiline_inside_interpolation);
-        }
         OpenDelimiters.push_back(CurPtr[-1]);
       }
+      if (*CurPtr == '"' && *(CurPtr + 1) == '"' && *(CurPtr - 1) == '"') {
+        MultilineString = true;
+        CurPtr += 2;
+      }
       continue;
     case '\\':
       if (inStringLiteral()) {
@@ -1373,7 +1376,8 @@ static StringRef getMultilineTrailingIndent(const Token &Str,
 
 /// validateMultilineIndents:
 /// Diagnose contents of string literal that have inconsistent indentation.
-void Lexer::validateMultilineIndents(const Token &Str) {
+static void validateMultilineIndents(const Token &Str,
+                                     DiagnosticEngine *Diags) {
   StringRef Indent = getMultilineTrailingIndent(Str, Diags);
   if (Indent.empty())
     return;
@@ -1385,7 +1389,8 @@ void Lexer::validateMultilineIndents(const Token &Str) {
     size_t nextpos = pos + 1;
     if (BytesPtr[nextpos] != '\n' && BytesPtr[nextpos] != '\r') {
       if (Bytes.substr(nextpos, Indent.size()) != Indent)
-        diagnose(BytesPtr + nextpos, diag::lex_ambiguous_string_indent);
+        Diags->diagnose(Lexer::getSourceLoc(BytesPtr + nextpos),
+                        diag::lex_ambiguous_string_indent);
     }
     pos = nextpos;
   }
@@ -1415,7 +1420,8 @@ void Lexer::lexStringLiteral() {
       // Consume tokens until we hit the corresponding ')'.
       CurPtr += 2;
       const char *EndPtr =
-          skipToEndOfInterpolatedExpression(CurPtr, BufferEnd, Diags);
+          skipToEndOfInterpolatedExpression(CurPtr, BufferEnd,
+                                            Diags, MultilineString);
 
       if (*EndPtr == ')') {
         // Successfully scanned the body of the expression literal.
@@ -1483,7 +1489,8 @@ void Lexer::lexStringLiteral() {
         if (*CurPtr == '"' && *(CurPtr + 1) == '"' && *(CurPtr + 2) != '"') {
           CurPtr += 2;
           formToken(tok::string_literal, TokStart, MultilineString);
-          validateMultilineIndents(NextToken);
+          if (Diags)
+            validateMultilineIndents(NextToken, Diags);
           return;
         }
         else
@@ -1778,7 +1785,7 @@ void Lexer::getStringLiteralSegments(
     // Find the closing ')'.
     const char *End = skipToEndOfInterpolatedExpression(BytesPtr,
                                                         Str.getText().end(),
-                                                        Diags);
+                                                        Diags, MultilineString);
     assert(*End == ')' && "invalid string literal interpolations should"
            " not be returned as string literals");
     ++End;

test/Parse/multiline_errors.swift

@@ -49,10 +49,3 @@ _ = """""" // expected-error@-0{{invalid start of multi-line string literal}}
 
 _ = """ """ // expected-error@-0{{invalid start of multi-line string literal}}
 // newline currently required after opening """
-
-_ = "\("""
-  not valid
-  """)" // expected-error@-2{{multi-line string literals are not allowed inside an interpolation}}
-// expected-error@-3{{unterminated string literal}}
-// expected-error@-2{{invalid start of multi-line string literal}}
-// expected-error@-3{{unterminated string literal}}

test/Parse/multiline_string.swift

@@ -135,3 +135,28 @@ _ = """
 _ = """
     """
 // CHECK: ""
+
+_ = "\("""
+  \("a" + """
+   valid
+  """)
+  """) literal"
+// CHECK: "a"
+// CHECK: " valid"
+// CHECK: " literal"
+
+_ = "hello\("""
+  world
+  """)"
+// CHECK: "hello"
+// CHECK: "world"
+
+_ = """
+  hello\("""
+     world
+     """)
+  abc
+  """
+// CHECK: "hello"
+// CHECK: "world"
+// CHECK: "\nabc"