json::dump() silently crashes if items contain accented letters

[pietro68]

• If a string field contains accented letters (e.g. à è ), a json::dump to std::string will silently crash

• run this simple snippet:
  void foobar() { using namespace nlohmann; json foo = { "resumè" }; std::string bar = foo.dump(4); }

• I expect the .dump operation to complete without problems...

• ...but instead, it silently crashes (i.e. control flow goes out of scope of foobar() )

• compiler: Visual Studio 2015 version 14.0.25431.01 update 3; Windows 10

• Did you use a released version of the library or the version from the develop branch? yes

[nlohmann]

Can you provide more information on the crash? An exception? A stack trace?

[pietro68]

Sure. Code snippet upgraded to:
{ try { json foo = { "resumè" }; std::string bar = foo.dump(4); } catch (std::exception& e) { std::cout << e.what() << std::endl; } }

And it prints out:

[json.exception.type_error.316] incomplete UTF-8 string; last byte: 0xE8

[nlohmann]

This is the description of the error: your compiler seems not to encode strings as UTF-8. Hence we have an exception, because we encountered an invalid byte. You can fix this by making your compiler use UTF-8, or add an error handler to ignore or replace these invalid bytes (see https://nlohmann.github.io/json/classnlohmann_1_1basic__json_a50ec80b02d0f3f51130d4abb5d1cfdc5.html#a50ec80b02d0f3f51130d4abb5d1cfdc5) which will remove the exception.

[pietro68]

Thanks, that solved it ( File | Advanced Saved Options... | UTF-8 without signature ).
Apologies for not looking deeper. VisualStudio comes out of the box with codepage 1252 as the default encoding, I did not realize it.