[YAML] Fix UTF-8 handling

[thegameg]

Summary

Previous YAML quoting patches broke UTF-8 printing in YAML: see https://reviews.llvm.org/D41290#961801.

[JDevlieghere]

Makes sense to me. LGTM!

[JDevlieghere]

unittests/Support/YAMLIOTest.cpp

2546 ↗

(On Diff #127880)

I'd remove the double single quotes here though, it doesn't really add anything to the test case except confusion.

2559 ↗

(On Diff #127880)

Maybe merge this with the previous test case?

llvm/trunk/lib/Support/YAMLTraits.cpp

if (S[j] == QuoteChar) {                  // Escape quotes.
       if (MustQuote == QuotingType::Double) { // Print it as \"
         output(StringLiteral("\\"));
         output(StringRef(Quote, 1));
       } else {                       // Single
         output(StringLiteral("''")); // Print it as ''
       }
       i = j + 1;
     } else if (MustQuote == QuotingType::Double &&
-               !sys::unicode::isPrintable(S[j])) {
+               !sys::unicode::isPrintable(S[j]) && (S[j] & 0x80) == 0) {
+      // If we're double quoting non-printable characters, we prefer printing
+      // them as "\x" + their hex representation. Note that special casing is
+      // needed for UTF-8, where a byte may be part of a UTF-8 sequence and
+      // appear as non-printable, in which case we want to print the correct
+      // unicode character and not its hex representation.
       output(StringRef(&Base[i], j - i)); // "flush"
       output(StringLiteral("\\x"));
 
       // Output the byte 0x0F as \x0f.
       auto FormattedHex = format_hex_no_prefix(S[j], 2);
       Out << FormattedHex;
       Column += 4; // one for the '\', one for the 'x', and two for the hex
 

llvm/trunk/unittests/Support/YAMLIOTest.cpp

TEST(YAMLIO, TestEscapedSingleQuoteInsideSingleQuote) {
   Output xout(ostr, nullptr, 0);
 
   llvm::yaml::EmptyContext Ctx;
   yamlize(xout, Id, true, Ctx);
 
   ostr.flush();
   EXPECT_EQ("'abc''fdf'", out);
 }
+
+TEST(YAMLIO, TestEscapedUTF8SingleQuoteInsideDoubleQuote) {
+  std::string Id = "parameter 'параметр' is unused";
+
+  std::string out;
+  llvm::raw_string_ostream ostr(out);
+  Output xout(ostr, nullptr, 0);
+
+  llvm::yaml::EmptyContext Ctx;
+  yamlize(xout, Id, true, Ctx);
+
+  ostr.flush();
+  EXPECT_EQ("\"parameter 'параметр' is unused\"", out);
+}
+
+TEST(YAMLIO, TestEscapedUTF8) {
+  std::string Id = "/*параметр*/";
+
+  std::string out;
+  llvm::raw_string_ostream ostr(out);
+  Output xout(ostr, nullptr, 0);
+
+  llvm::yaml::EmptyContext Ctx;
+  yamlize(xout, Id, true, Ctx);
+
+  ostr.flush();
+  EXPECT_EQ("\"/*параметр*/\"", out);
+}