Index: llvm/trunk/lib/Support/YAMLTraits.cpp =================================================================== --- llvm/trunk/lib/Support/YAMLTraits.cpp +++ llvm/trunk/lib/Support/YAMLTraits.cpp @@ -657,7 +657,12 @@ } 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")); Index: llvm/trunk/unittests/Support/YAMLIOTest.cpp =================================================================== --- llvm/trunk/unittests/Support/YAMLIOTest.cpp +++ llvm/trunk/unittests/Support/YAMLIOTest.cpp @@ -2541,3 +2541,31 @@ 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); +}