[clang-format] Handle escaped " in C# string-literals

[krasimir]

clang/lib/Format/FormatTokenLexer.cpp

197

This merging is only valid if CSharpStringLiteral starts with @" or with $@"; if the literal stars with $", it's invalid.
Consider this example:
$"""Hello"
In this case these really are 2 different tokens (I don't know C# so I might be wrong, but vim syntax highlighting suggests this).

Consider also adding tests for these combinations.

213

Nice catch!

[krasimir]

clang/unittests/Format/FormatTestCSharp.cpp

415

maybe add an example where these are applied a couple of times, like in @"""Hello, world""" (to make sure the state of the token stream after the merging is good).

[jbcoe]

Extend tests and explicit check for verbatim strings not just C# string literals.

[krasimir]

Thank you!
Should I commit this for you?

[jbcoe]

@krasimir Would be great if you could commit this.

clang/lib/Format/FormatTokenLexer.cpp

 }
 
 // Search for verbatim or interpolated string literals @"ABC" or
 // $"aaaaa{abc}aaaaa" i and mark the token as TT_CSharpStringLiteral, and to
 // prevent splitting of @, $ and ".
 bool FormatTokenLexer::tryMergeCSharpVerbatimStringLiteral() {
   if (Tokens.size() < 2)
     return false;
-  auto &At = *(Tokens.end() - 2);
+
   auto &String = *(Tokens.end() - 1);
+  if (!String->is(tok::string_literal))
+    return false;
+
+  // verbatim strings could contain "" which C# sees as an escaped ".
+  // @"""Hello""" will have been tokenized as @"" "Hello" "" and needs
+  // merging into a single string literal.
+  auto &CSharpStringLiteral = *(Tokens.end() - 2);
+  if (CSharpStringLiteral->Type == TT_CSharpStringLiteral &&
+      (CSharpStringLiteral->TokenText.startswith(R"(@")") ||
+       CSharpStringLiteral->TokenText.startswith(R"($@")"))) {
+    CSharpStringLiteral->TokenText = StringRef(
+        CSharpStringLiteral->TokenText.begin(),
+        String->TokenText.end() - CSharpStringLiteral->TokenText.begin());
+    CSharpStringLiteral->ColumnWidth += String->ColumnWidth;
+    Tokens.erase(Tokens.end() - 1);
+    return true;
+  }
+
+  auto &At = *(Tokens.end() - 2);
 
-  // Look for $"aaaaaa" @"aaaaaa".
-  if (!(At->is(tok::at) || At->TokenText == "$") ||
-      !String->is(tok::string_literal))
+  // Look for @"aaaaaa" or $"aaaaaa".
+  if (!(At->is(tok::at) || At->TokenText == "$"))
     return false;
 
-  if (Tokens.size() >= 2 && At->is(tok::at)) {
+  if (Tokens.size() > 2 && At->is(tok::at)) {
     auto &Dollar = *(Tokens.end() - 3);
     if (Dollar->TokenText == "$") {
       // This looks like $@"aaaaa" so we need to combine all 3 tokens.
       Dollar->Tok.setKind(tok::string_literal);
       Dollar->TokenText =
           StringRef(Dollar->TokenText.begin(),
                     String->TokenText.end() - Dollar->TokenText.begin());
       Dollar->ColumnWidth += (At->ColumnWidth + String->ColumnWidth);

clang/unittests/Format/FormatTestCSharp.cpp

TEST_F(FormatTestCSharp, CSharpSpaceAfterCStyleCast) {
   FormatStyle Style = getGoogleStyle(FormatStyle::LK_CSharp);
 
   verifyFormat("(int)x / y;", Style);
 
   Style.SpaceAfterCStyleCast = true;
   verifyFormat("(int) x / y;", Style);
 }
 
+TEST_F(FormatTestCSharp, CSharpEscapedQuotesInVerbatimStrings) {
+  FormatStyle Style = getGoogleStyle(FormatStyle::LK_CSharp);
+
+  verifyFormat(R"(string str = @"""")", Style);
+  verifyFormat(R"(string str = @"""Hello world""")", Style);
+  verifyFormat(R"(string str = $@"""Hello {friend}""")", Style);
+}
+
 } // namespace format
 } // end namespace clang