dataverse++

This library provides a C++ API for uploading files to a Dataverse instance. Its main purpose is enabling direct archival of research data from your simulation code or benchmarking experiments.

Note that this library is currently in development. We eventually plan to support Windows and Linux, but it is currently only running on Windows.

Building

CMake is required for building. It should obtain all dependencies automatically when configuring.

Usage

dataverse_connection dataverse;
dataverse.base_path(L"https://darus.uni-stuttgart.de/api")
    .api_key(L"YOUR API KEY")
    .upload(L"doi:10.18419/darus-3044",
        L"T:\\data\\rtx4090.csv",
        L"Measurement results for GeForce RTX4090",
        L"/raw/",
        nullptr, 0, false,
        [](const blob& result, void *context) {
            std::string output(result.as<char>(), r.size());
            std::cout << convert<char>(convert<wchar_t>(output, CP_UTF8), CP_OEMCP) << std::endl;
            std::cout << std::endl;
        },
        [](const int error, const char_type *msg, const char_type *cat, void *context) {
            std::cerr << msg << std::endl << std::endl;
        });

When using narrow strings, you must tell the API about their encoding:

dataverse.base_path(const_narrow_string("https://darus.uni-stuttgart.de/api/", CP_ACP))
    .api_key(make_narrow_string("YOUR API KEY", CP_ACP))
    .upload(make_narrow_string("doi:10.18419/darus-3044", CP_ACP),
        make_narrow_string("T:\\data\\rtx4090.csv", CP_ACP),
        make_narrow_string("Measurement results for GeForce RTX4090", CP_ACP),
        make_narrow_string("/raw/", CP_ACP),
        std::vector<const_narrow_string> { narrow_string("#bota", CP_ACP), narrow_string("#boschofthemall", CP_ACP) },
        false,
        [](const blob& result, void *context) {
            std::string output(result.as<char>(), r.size());
            std::cout << convert<char>(convert<wchar_t>(output, CP_UTF8), CP_OEMCP) << std::endl;
            std::cout << std::endl;
        },
        [](const int error, const char_type *msg, const char_type *cat, void *context) {
            std::cerr << msg << std::endl << std::endl;
        });

⚠️ The above code is only for illustration purposes. You should never hard-code the code page like this in production code! Unless you are a MegaMol developer, you should track the encoding from its origin. For instance, console input on Windows is CP_OEMCP. On Linux, you need to determine that using system calls or via libicu. The encoding of literal strings is compiler-dependent, but usually CP_ACP on Windows. When using file input, you must ask the user or apply heuristics like detecing UTF-8 BOMs. While you might get away with hard coding for URLs and your API key, because these are most likely 7-bit ASCII anyway, going the hard coding way has the potential to turn your metadata into rubbish. The library will convert all metadata to UTF-8 based on your input in order to create the JSON descriptor objects. Specifying the wrong code page may corrupt the output. Specifying UTF-8 as the code page while the data are not actually UTF-8 may crash your application.

Provided you have nlohmann/json installed, you can also make requests like for the creation of a new data set:

auto data_set = nlohmann::json({ });

data_set["datasetVersion"]["license"]["name"] = "CC BY 4.0";
data_set["datasetVersion"]["license"]["uri"] = "http://creativecommons.org/licenses/by/4.0/";
data_set["datasetVersion"]["metadataBlocks"]["citation"] = visus::dataverse::json::make_citation_metadata(
    visus::dataverse::json::make_meta_field(
        L"title",
        L"primitive",
        false,
        visus::dataverse::to_utf8(L"Energy consumption of scientific visualisation and data visualisation algorithms")),

    visus::dataverse::json::make_meta_field(L"author", L"compound", true,
        visus::dataverse::json::make_author(L"Müller, Christoph"),
        visus::dataverse::json::make_author(L"Heinemann, Moritz"),
        visus::dataverse::json::make_author(L"Weiskopf, Daniel"),
        visus::dataverse::json::make_author(L"Ertl, Thomas")),

    visus::dataverse::json::make_meta_field(L"datasetContact", L"compound", true,
        visus::dataverse::json::make_contact(L"Querulant", L"querulant@visus.uni-stuttgart.de")),

    visus::dataverse::json::make_meta_field(L"dsDescription", L"compound", true,
        visus::dataverse::json::make_data_desc(L"This data set comprises a series of measurements of GPU power consumption.")),

    visus::dataverse::json::make_meta_field(L"subject",
        L"controlledVocabulary",
        true,
        visus::dataverse::to_utf8(L"Computer and Information Science"))
);

Make sure that all string content is UTF-8. Just assuming that string literals are UTF-8 is not sufficient. The construction of the meta data will fail if any non-UTF-8 code unit is discovered. It is best using the to_utf8 functions provided by the library.