README.md

[![JSON for Modern C++](https://raw.githubusercontent.com/nlohmann/json/master/doc/json.gif)](https://github.com/nlohmann/json/releases)

[![Build Status](https://travis-ci.org/nlohmann/json.svg?branch=master)](https://travis-ci.org/nlohmann/json)
[![Build Status](https://ci.appveyor.com/api/projects/status/1acb366xfyg3qybk/branch/develop?svg=true)](https://ci.appveyor.com/project/nlohmann/json)
[![Coverage Status](https://img.shields.io/coveralls/nlohmann/json.svg)](https://coveralls.io/r/nlohmann/json)
[![Coverity Scan Build Status](https://scan.coverity.com/projects/5550/badge.svg)](https://scan.coverity.com/projects/nlohmann-json)
[![Codacy Badge](https://api.codacy.com/project/badge/Grade/f3732b3327e34358a0e9d1fe9f661f08)](https://www.codacy.com/app/nlohmann/json?utm_source=github.com&utm_medium=referral&utm_content=nlohmann/json&utm_campaign=Badge_Grade)
[![Language grade: C/C++](https://img.shields.io/lgtm/grade/cpp/g/nlohmann/json.svg?logo=lgtm&logoWidth=18)](https://lgtm.com/projects/g/nlohmann/json/context:cpp)
[![Try online](https://img.shields.io/badge/try-online-blue.svg)](https://wandbox.org/permlink/TarF5pPn9NtHQjhf)
[![Documentation](https://img.shields.io/badge/docs-doxygen-blue.svg)](http://nlohmann.github.io/json)
[![GitHub license](https://img.shields.io/badge/license-MIT-blue.svg)](https://raw.githubusercontent.com/nlohmann/json/master/LICENSE.MIT)
[![GitHub Releases](https://img.shields.io/github/release/nlohmann/json.svg)](https://github.com/nlohmann/json/releases)
[![GitHub Issues](https://img.shields.io/github/issues/nlohmann/json.svg)](http://github.com/nlohmann/json/issues)
[![Average time to resolve an issue](http://isitmaintained.com/badge/resolution/nlohmann/json.svg)](http://isitmaintained.com/project/nlohmann/json "Average time to resolve an issue")
[![CII Best Practices](https://bestpractices.coreinfrastructure.org/projects/289/badge)](https://bestpractices.coreinfrastructure.org/projects/289)

- [Design goals](#design-goals)
- [Integration](#integration)
  - [CMake](#cmake)
  - [Package Managers](#package-managers)
- [Examples](#examples)
  - [JSON as first-class data type](#json-as-first-class-data-type)
  - [Serialization / Deserialization](#serialization--deserialization)
  - [STL-like access](#stl-like-access)
  - [Conversion from STL containers](#conversion-from-stl-containers)
  - [JSON Pointer and JSON Patch](#json-pointer-and-json-patch)
  - [JSON Merge Patch](#json-merge-patch)
  - [Implicit conversions](#implicit-conversions)
  - [Conversions to/from arbitrary types](#arbitrary-types-conversions)
  - [Specializing enum conversion](#specializing-enum-conversion)
  - [Binary formats (BSON, CBOR, MessagePack, and UBJSON)](#binary-formats-bson-cbor-messagepack-and-ubjson)
- [Supported compilers](#supported-compilers)
- [License](#license)
- [Contact](#contact)
- [Thanks](#thanks)
- [Used third-party tools](#used-third-party-tools)
- [Projects using JSON for Modern C++](#projects-using-json-for-modern-c)
- [Notes](#notes)
- [Execute unit tests](#execute-unit-tests)

## Design goals

There are myriads of [JSON](http://json.org) libraries out there, and each may even have its reason to exist. Our class had these design goals:

- **Intuitive syntax**. In languages such as Python, JSON feels like a first class data type. We used all the operator magic of modern C++ to achieve the same feeling in your code. Check out the [examples below](#examples) and you'll know what I mean.

- **Trivial integration**. Our whole code consists of a single header file [`json.hpp`](https://github.com/nlohmann/json/blob/develop/single_include/nlohmann/json.hpp). That's it. No library, no subproject, no dependencies, no complex build system. The class is written in vanilla C++11. All in all, everything should require no adjustment of your compiler flags or project settings.

- **Serious testing**. Our class is heavily [unit-tested](https://github.com/nlohmann/json/tree/develop/test/src) and covers [100%](https://coveralls.io/r/nlohmann/json) of the code, including all exceptional behavior. Furthermore, we checked with [Valgrind](http://valgrind.org) and the [Clang Sanitizers](https://clang.llvm.org/docs/index.html) that there are no memory leaks. [Google OSS-Fuzz](https://github.com/google/oss-fuzz/tree/master/projects/json) additionally runs fuzz tests against all parsers 24/7, effectively executing billions of tests so far. To maintain high quality, the project is following the [Core Infrastructure Initiative (CII) best practices](https://bestpractices.coreinfrastructure.org/projects/289).

Other aspects were not so important to us:

- **Memory efficiency**. Each JSON object has an overhead of one pointer (the maximal size of a union) and one enumeration element (1 byte). The default generalization uses the following C++ data types: `std::string` for strings, `int64_t`, `uint64_t` or `double` for numbers, `std::map` for objects, `std::vector` for arrays, and `bool` for Booleans. However, you can template the generalized class `basic_json` to your needs.

- **Speed**. There are certainly [faster JSON libraries](https://github.com/miloyip/nativejson-benchmark#parsing-time) out there. However, if your goal is to speed up your development by adding JSON support with a single header, then this library is the way to go. If you know how to use a `std::vector` or `std::map`, you are already set.

See the [contribution guidelines](https://github.com/nlohmann/json/blob/master/.github/CONTRIBUTING.md#please-dont) for more information.


## Integration

[`json.hpp`](https://github.com/nlohmann/json/blob/develop/single_include/nlohmann/json.hpp) is the single required file in `single_include/nlohmann` or [released here](https://github.com/nlohmann/json/releases). You need to add

```cpp
#include <nlohmann/json.hpp>

// for convenience
using json = nlohmann::json;
```

to the files you want to process JSON and set the necessary switches to enable C++11 (e.g., `-std=c++11` for GCC and Clang).

You can further use file [`include/nlohmann/json_fwd.hpp`](https://github.com/nlohmann/json/blob/develop/include/nlohmann/json_fwd.hpp) for forward-declarations. The installation of json_fwd.hpp (as part of cmake's install step), can be achieved by setting `-DJSON_MultipleHeaders=ON`.

### CMake

You can also use the `nlohmann_json::nlohmann_json` interface target in CMake.  This target populates the appropriate usage requirements for `INTERFACE_INCLUDE_DIRECTORIES` to point to the appropriate include directories and `INTERFACE_COMPILE_FEATURES` for the necessary C++11 flags.

#### External

To use this library from a CMake project, you can locate it directly with `find_package()` and use the namespaced imported target from the generated package configuration:

```cmake
# CMakeLists.txt
find_package(nlohmann_json 3.2.0 REQUIRED)
...
add_library(foo ...)
...
target_link_libraries(foo PRIVATE nlohmann_json::nlohmann_json)
```

The package configuration file, `nlohmann_jsonConfig.cmake`, can be used either from an install tree or directly out of the build tree.

#### Embedded

To embed the library directly into an existing CMake project, place the entire source tree in a subdirectory and call `add_subdirectory()` in your `CMakeLists.txt` file:

```cmake
# Typically you don't care so much for a third party library's tests to be
# run from your own project's code.
set(JSON_BuildTests OFF CACHE INTERNAL "")

# Don't use include(nlohmann_json/CMakeLists.txt) since that carries with it
# unintended consequences that will break the build.  It's generally
# discouraged (although not necessarily well documented as such) to use
# include(...) for pulling in other CMake projects anyways.
add_subdirectory(nlohmann_json)
...
add_library(foo ...)
...
target_link_libraries(foo PRIVATE nlohmann_json::nlohmann_json)
```

#### Supporting Both

To allow your project to support either an externally supplied or an embedded JSON library, you can use a pattern akin to the following:

``` cmake
# Top level CMakeLists.txt
project(FOO)
...
option(FOO_USE_EXTERNAL_JSON "Use an external JSON library" OFF)
...
add_subdirectory(thirdparty)
...
add_library(foo ...)
...
# Note that the namespaced target will always be available regardless of the
# import method
target_link_libraries(foo PRIVATE nlohmann_json::nlohmann_json)
```
```cmake
# thirdparty/CMakeLists.txt
...
if(FOO_USE_EXTERNAL_JSON)
  find_package(nlohmann_json 3.2.0 REQUIRED)
else()
  set(JSON_BuildTests OFF CACHE INTERNAL "")
  add_subdirectory(nlohmann_json)
endif()
...
```

`thirdparty/nlohmann_json` is then a complete copy of this source tree.

### Package Managers

:beer: If you are using OS X and [Homebrew](http://brew.sh), just type `brew tap nlohmann/json` and `brew install nlohmann_json` and you're set. If you want the bleeding edge rather than the latest release, use `brew install nlohmann_json --HEAD`.

If you are using the [Meson Build System](http://mesonbuild.com), then you can get a wrap file by downloading it from [Meson WrapDB](https://wrapdb.mesonbuild.com/nlohmann_json), or simply use `meson wrap install nlohmann_json`.

If you are using [Conan](https://www.conan.io/) to manage your dependencies, merely add `jsonformoderncpp/x.y.z@vthiery/stable` to your `conanfile.py`'s requires, where `x.y.z` is the release version you want to use. Please file issues [here](https://github.com/vthiery/conan-jsonformoderncpp/issues) if you experience problems with the packages.

If you are using [Spack](https://www.spack.io/) to manage your dependencies, you can use the `nlohmann_json` package. Please see the [spack project](https://github.com/spack/spack) for any issues regarding the packaging.

If you are using [hunter](https://github.com/ruslo/hunter/) on your project for external dependencies, then you can use the [nlohmann_json package](https://docs.hunter.sh/en/latest/packages/pkg/nlohmann_json.html). Please see the hunter project for any issues regarding the packaging.

If you are using [Buckaroo](https://buckaroo.pm), you can install this library's module with `buckaroo install nlohmann/json`. Please file issues [here](https://github.com/LoopPerfect/buckaroo-recipes/issues/new?title=nlohmann/nlohmann/json).

If you are using [vcpkg](https://github.com/Microsoft/vcpkg/) on your project for external dependencies, then you can use the [nlohmann-json package](https://github.com/Microsoft/vcpkg/tree/master/ports/nlohmann-json). Please see the vcpkg project for any issues regarding the packaging.

If you are using [cget](http://cget.readthedocs.io/en/latest/), you can install the latest development version with `cget install nlohmann/json`. A specific version can be installed with `cget install nlohmann/json@v3.1.0`. Also, the multiple header version can be installed by adding the `-DJSON_MultipleHeaders=ON` flag (i.e., `cget install nlohmann/json -DJSON_MultipleHeaders=ON`).

If you are using [CocoaPods](https://cocoapods.org), you can use the library by adding pod `"nlohmann_json", '~>3.1.2'` to your podfile (see [an example](https://bitbucket.org/benman/nlohmann_json-cocoapod/src/master/)). Please file issues [here](https://bitbucket.org/benman/nlohmann_json-cocoapod/issues?status=new&status=open).

If you are using [NuGet](https://www.nuget.org), you can use the package [nlohmann.json](https://www.nuget.org/packages/nlohmann.json/). Please check [this extensive description](https://github.com/nlohmann/json/issues/1132#issuecomment-452250255) on how to use the package. Please files issues [here](https://github.com/hnkb/nlohmann-json-nuget/issues).

If you are using [conda](https://conda.io/), you can use the package [nlohmann_json](https://github.com/conda-forge/nlohmann_json-feedstock) from [conda-forge](https://conda-forge.org) executing `conda install -c conda-forge nlohmann_json`. Please file issues [here](https://github.com/conda-forge/nlohmann_json-feedstock/issues).

## Examples

Beside the examples below, you may want to check the [documentation](https://nlohmann.github.io/json/) where each function contains a separate code example (e.g., check out [`emplace()`](https://nlohmann.github.io/json/classnlohmann_1_1basic__json_a5338e282d1d02bed389d852dd670d98d.html#a5338e282d1d02bed389d852dd670d98d)). All [example files](https://github.com/nlohmann/json/tree/develop/doc/examples) can be compiled and executed on their own (e.g., file [emplace.cpp](https://github.com/nlohmann/json/blob/develop/doc/examples/emplace.cpp)).

### JSON as first-class data type

Here are some examples to give you an idea how to use the class.

Assume you want to create the JSON object

```json
{
  "pi": 3.141,
  "happy": true,
  "name": "Niels",
  "nothing": null,
  "answer": {
    "everything": 42
  },
  "list": [1, 0, 2],
  "object": {
    "currency": "USD",
    "value": 42.99
  }
}
```

With this library, you could write:

```cpp
// create an empty structure (null)
json j;

// add a number that is stored as double (note the implicit conversion of j to an object)
j["pi"] = 3.141;

// add a Boolean that is stored as bool
j["happy"] = true;

// add a string that is stored as std::string
j["name"] = "Niels";

// add another null object by passing nullptr
j["nothing"] = nullptr;

// add an object inside the object
j["answer"]["everything"] = 42;

// add an array that is stored as std::vector (using an initializer list)
j["list"] = { 1, 0, 2 };

// add another object (using an initializer list of pairs)
j["object"] = { {"currency", "USD"}, {"value", 42.99} };

// instead, you could also write (which looks very similar to the JSON above)
json j2 = {
  {"pi", 3.141},
  {"happy", true},
  {"name", "Niels"},
  {"nothing", nullptr},
  {"answer", {
    {"everything", 42}
  }},
  {"list", {1, 0, 2}},
  {"object", {
    {"currency", "USD"},
    {"value", 42.99}
  }}
};
```

Note that in all these cases, you never need to "tell" the compiler which JSON value type you want to use. If you want to be explicit or express some edge cases, the functions [`json::array`](https://nlohmann.github.io/json/classnlohmann_1_1basic__json_aa80485befaffcadaa39965494e0b4d2e.html#aa80485befaffcadaa39965494e0b4d2e) and [`json::object`](https://nlohmann.github.io/json/classnlohmann_1_1basic__json_aa13f7c0615867542ce80337cbcf13ada.html#aa13f7c0615867542ce80337cbcf13ada) will help:

```cpp
// a way to express the empty array []
json empty_array_explicit = json::array();

// ways to express the empty object {}
json empty_object_implicit = json({});
json empty_object_explicit = json::object();

// a way to express an _array_ of key/value pairs [["currency", "USD"], ["value", 42.99]]
json array_not_object = json::array({ {"currency", "USD"}, {"value", 42.99} });
```

### Serialization / Deserialization

#### To/from strings

You can create a JSON value (deserialization) by appending `_json` to a string literal:

```cpp
// create object from string literal
json j = "{ \"happy\": true, \"pi\": 3.141 }"_json;

// or even nicer with a raw string literal
auto j2 = R"(
  {
    "happy": true,
    "pi": 3.141
  }
)"_json;
```

Note that without appending the `_json` suffix, the passed string literal is not parsed, but just used as JSON string value. That is, `json j = "{ \"happy\": true, \"pi\": 3.141 }"` would just store the string `"{ "happy": true, "pi": 3.141 }"` rather than parsing the actual object.

The above example can also be expressed explicitly using [`json::parse()`](https://nlohmann.github.io/json/classnlohmann_1_1basic__json_a5a0339361f3282cb8fd2f9ede6e17d72.html#a5a0339361f3282cb8fd2f9ede6e17d72):

```cpp
// parse explicitly
auto j3 = json::parse("{ \"happy\": true, \"pi\": 3.141 }");
```

You can also get a string representation of a JSON value (serialize):

```cpp
// explicit conversion to string
std::string s = j.dump();    // {\"happy\":true,\"pi\":3.141}

// serialization with pretty printing
// pass in the amount of spaces to indent
std::cout << j.dump(4) << std::endl;
// {
//     "happy": true,
//     "pi": 3.141
// }
```

Note the difference between serialization and assignment:

```cpp
// store a string in a JSON value
json j_string = "this is a string";

// retrieve the string value
auto cpp_string = j_string.get<std::string>();
// retrieve the string value (alternative when an variable already exists)
std::string cpp_string2;
j_string.get_to(cpp_string2);

// retrieve the serialized value (explicit JSON serialization)
std::string serialized_string = j_string.dump();

// output of original string
std::cout << cpp_string << " == " << cpp_string2 << " == " << j_string.get<std::string>() << '\n';
// output of serialized value
std::cout << j_string << " == " << serialized_string << std::endl;
```

[`.dump()`](https://nlohmann.github.io/json/classnlohmann_1_1basic__json_a50ec80b02d0f3f51130d4abb5d1cfdc5.html#a50ec80b02d0f3f51130d4abb5d1cfdc5) always returns the serialized value, and [`.get<std::string>()`](https://nlohmann.github.io/json/classnlohmann_1_1basic__json_a16f9445f7629f634221a42b967cdcd43.html#a16f9445f7629f634221a42b967cdcd43) returns the originally stored string value.

Note the library only supports UTF-8. When you store strings with different encodings in the library, calling [`dump()`](https://nlohmann.github.io/json/classnlohmann_1_1basic__json_a50ec80b02d0f3f51130d4abb5d1cfdc5.html#a50ec80b02d0f3f51130d4abb5d1cfdc5) may throw an exception unless `json::error_handler_t::replace` or `json::error_handler_t::ignore` are used as error handlers.

#### To/from streams (e.g. files, string streams)

You can also use streams to serialize and deserialize:

```cpp
// deserialize from standard input
json j;
std::cin >> j;

// serialize to standard output
std::cout << j;

// the setw manipulator was overloaded to set the indentation for pretty printing
std::cout << std::setw(4) << j << std::endl;
```

These operators work for any subclasses of `std::istream` or `std::ostream`. Here is the same example with files:

```cpp
// read a JSON file
std::ifstream i("file.json");
json j;
i >> j;

// write prettified JSON to another file
std::ofstream o("pretty.json");
o << std::setw(4) << j << std::endl;
```

Please note that setting the exception bit for `failbit` is inappropriate for this use case. It will result in program termination due to the `noexcept` specifier in use.

#### Read from iterator range

You can also parse JSON from an iterator range; that is, from any container accessible by iterators whose content is stored as contiguous byte sequence, for instance a `std::vector<std::uint8_t>`:

```cpp
std::vector<std::uint8_t> v = {'t', 'r', 'u', 'e'};
json j = json::parse(v.begin(), v.end());
```

You may leave the iterators for the range [begin, end):

```cpp
std::vector<std::uint8_t> v = {'t', 'r', 'u', 'e'};
json j = json::parse(v);
```

#### SAX interface

The library uses a SAX-like interface with the following functions:

```cpp
// called when null is parsed
bool null();

// called when a boolean is parsed; value is passed
bool boolean(bool val);

// called when a signed or unsigned integer number is parsed; value is passed
bool number_integer(number_integer_t val);
bool number_unsigned(number_unsigned_t val);

// called when a floating-point number is parsed; value and original string is passed
bool number_float(number_float_t val, const string_t& s);

// called when a string is parsed; value is passed and can be safely moved away
bool string(string_t& val);

// called when an object or array begins or ends, resp. The number of elements is passed (or -1 if not known)
bool start_object(std::size_t elements);
bool end_object();
bool start_array(std::size_t elements);
bool end_array();
// called when an object key is parsed; value is passed and can be safely moved away
bool key(string_t& val);

// called when a parse error occurs; byte position, the last token, and an exception is passed
bool parse_error(std::size_t position, const std::string& last_token, const detail::exception& ex);
```

The return value of each function determines whether parsing should proceed.

To implement your own SAX handler, proceed as follows:

1. Implement the SAX interface in a class. You can use class `nlohmann::json_sax<json>` as base class, but you can also use any class where the functions described above are implemented and public.
2. Create an object of your SAX interface class, e.g. `my_sax`.
3. Call `bool json::sax_parse(input, &my_sax)`; where the first parameter can be any input like a string or an input stream and the second parameter is a pointer to your SAX interface.

Note the `sax_parse` function only returns a `bool` indicating the result of the last executed SAX event. It does not return a  `json` value - it is up to you to decide what to do with the SAX events. Furthermore, no exceptions are thrown in case of a parse error - it is up to you what to do with the exception object passed to your `parse_error` implementation. Internally, the SAX interface is used for the DOM parser (class `json_sax_dom_parser`) as well as the acceptor (`json_sax_acceptor`), see file [`json_sax.hpp`](https://github.com/nlohmann/json/blob/develop/include/nlohmann/detail/input/json_sax.hpp).

### STL-like access

We designed the JSON class to behave just like an STL container. In fact, it satisfies the [**ReversibleContainer**](https://en.cppreference.com/w/cpp/named_req/ReversibleContainer) requirement.

```cpp
// create an array using push_back
json j;
j.push_back("foo");
j.push_back(1);
j.push_back(true);

// also use emplace_back
j.emplace_back(1.78);

// iterate the array
for (json::iterator it = j.begin(); it != j.end(); ++it) {
  std::cout << *it << '\n';
}

// range-based for
for (auto& element : j) {
  std::cout << element << '\n';
}

// getter/setter
const auto tmp = j[0].get<std::string>();
j[1] = 42;
bool foo = j.at(2);

// comparison
j == "[\"foo\", 1, true]"_json;  // true

// other stuff
j.size();     // 3 entries
j.empty();    // false
j.type();     // json::value_t::array
j.clear();    // the array is empty again

// convenience type checkers
j.is_null();
j.is_boolean();
j.is_number();
j.is_object();
j.is_array();
j.is_string();

// create an object
json o;
o["foo"] = 23;
o["bar"] = false;
o["baz"] = 3.141;

// also use emplace
o.emplace("weather", "sunny");

// special iterator member functions for objects
for (json::iterator it = o.begin(); it != o.end(); ++it) {
  std::cout << it.key() << " : " << it.value() << "\n";
}

// the same code as range for
for (auto& el : o.items()) {
  std::cout << el.key() << " : " << el.value() << "\n";
}

// even easier with structured bindings (C++17)
for (auto& [key, value] : o.items()) {
  std::cout << key << " : " << value << "\n";
}

// find an entry
if (o.find("foo") != o.end()) {
  // there is an entry with key "foo"
}

// or simpler using count()
int foo_present = o.count("foo"); // 1
int fob_present = o.count("fob"); // 0

// delete an entry
o.erase("foo");
```


### Conversion from STL containers

Any sequence container (`std::array`, `std::vector`, `std::deque`, `std::forward_list`, `std::list`) whose values can be used to construct JSON values (e.g., integers, floating point numbers, Booleans, string types, or again STL containers described in this section) can be used to create a JSON array. The same holds for similar associative containers (`std::set`, `std::multiset`, `std::unordered_set`, `std::unordered_multiset`), but in these cases the order of the elements of the array depends on how the elements are ordered in the respective STL container.

```cpp
std::vector<int> c_vector {1, 2, 3, 4};
json j_vec(c_vector);
// [1, 2, 3, 4]

std::deque<double> c_deque {1.2, 2.3, 3.4, 5.6};
json j_deque(c_deque);
// [1.2, 2.3, 3.4, 5.6]

std::list<bool> c_list {true, true, false, true};
json j_list(c_list);
// [true, true, false, true]

std::forward_list<int64_t> c_flist {12345678909876, 23456789098765, 34567890987654, 45678909876543};
json j_flist(c_flist);
// [12345678909876, 23456789098765, 34567890987654, 45678909876543]

std::array<unsigned long, 4> c_array {{1, 2, 3, 4}};
json j_array(c_array);
// [1, 2, 3, 4]

std::set<std::string> c_set {"one", "two", "three", "four", "one"};
json j_set(c_set); // only one entry for "one" is used
// ["four", "one", "three", "two"]

std::unordered_set<std::string> c_uset {"one", "two", "three", "four", "one"};
json j_uset(c_uset); // only one entry for "one" is used
// maybe ["two", "three", "four", "one"]

std::multiset<std::string> c_mset {"one", "two", "one", "four"};
json j_mset(c_mset); // both entries for "one" are used
// maybe ["one", "two", "one", "four"]

std::unordered_multiset<std::string> c_umset {"one", "two", "one", "four"};
json j_umset(c_umset); // both entries for "one" are used
// maybe ["one", "two", "one", "four"]
```

Likewise, any associative key-value containers (`std::map`, `std::multimap`, `std::unordered_map`, `std::unordered_multimap`) whose keys can construct an `std::string` and whose values can be used to construct JSON values (see examples above) can be used to create a JSON object. Note that in case of multimaps only one key is used in the JSON object and the value depends on the internal order of the STL container.

```cpp
std::map<std::string, int> c_map { {"one", 1}, {"two", 2}, {"three", 3} };
json j_map(c_map);
// {"one": 1, "three": 3, "two": 2 }

std::unordered_map<const char*, double> c_umap { {"one", 1.2}, {"two", 2.3}, {"three", 3.4} };
json j_umap(c_umap);
// {"one": 1.2, "two": 2.3, "three": 3.4}

std::multimap<std::string, bool> c_mmap { {"one", true}, {"two", true}, {"three", false}, {"three", true} };
json j_mmap(c_mmap); // only one entry for key "three" is used
// maybe {"one": true, "two": true, "three": true}

std::unordered_multimap<std::string, bool> c_ummap { {"one", true}, {"two", true}, {"three", false}, {"three", true} };
json j_ummap(c_ummap); // only one entry for key "three" is used
// maybe {"one": true, "two": true, "three": true}
```

### JSON Pointer and JSON Patch

The library supports **JSON Pointer** ([RFC 6901](https://tools.ietf.org/html/rfc6901)) as alternative means to address structured values. On top of this, **JSON Patch** ([RFC 6902](https://tools.ietf.org/html/rfc6902)) allows to describe differences between two JSON values - effectively allowing patch and diff operations known from Unix.

```cpp
// a JSON value
json j_original = R"({
  "baz": ["one", "two", "three"],
  "foo": "bar"
})"_json;

// access members with a JSON pointer (RFC 6901)
j_original["/baz/1"_json_pointer];
// "two"

// a JSON patch (RFC 6902)
json j_patch = R"([
  { "op": "replace", "path": "/baz", "value": "boo" },
  { "op": "add", "path": "/hello", "value": ["world"] },
  { "op": "remove", "path": "/foo"}
])"_json;

// apply the patch
json j_result = j_original.patch(j_patch);
// {
//    "baz": "boo",
//    "hello": ["world"]
// }

// calculate a JSON patch from two JSON values
json::diff(j_result, j_original);
// [
//   { "op":" replace", "path": "/baz", "value": ["one", "two", "three"] },
//   { "op": "remove","path": "/hello" },
//   { "op": "add", "path": "/foo", "value": "bar" }
// ]
```

### JSON Merge Patch

The library supports **JSON Merge Patch** ([RFC 7386](https://tools.ietf.org/html/rfc7386)) as a patch format. Instead of using JSON Pointer (see above) to specify values to be manipulated, it describes the changes using a syntax that closely mimics the document being modified.

```cpp
// a JSON value
json j_document = R"({
  "a": "b",
  "c": {
    "d": "e",
    "f": "g"
  }
})"_json;

// a patch
json j_patch = R"({
  "a":"z",
  "c": {
    "f": null
  }
})"_json;

// apply the patch
j_original.merge_patch(j_patch);
// {
//  "a": "z",
//  "c": {
//    "d": "e"
//  }
// }
```

### Implicit conversions

Supported types can be implicitly converted to JSON values.

It is recommended to **NOT USE** implicit conversions **FROM** a JSON value.
You can find more details about this recommendation [here](https://www.github.com/nlohmann/json/issues/958). 

```cpp
// strings
std::string s1 = "Hello, world!";
json js = s1;
auto s2 = js.get<std::string>();
// NOT RECOMMENDED
std::string s3 = js;
std::string s4;
s4 = js;

// Booleans
bool b1 = true;
json jb = b1;
auto b2 = jb.get<bool>();
// NOT RECOMMENDED
bool b3 = jb;
bool b4;
b4 = jb;

// numbers
int i = 42;
json jn = i;
auto f = jn.get<double>();
// NOT RECOMMENDED
double f2 = jb;
double f3;
f3 = jb;

// etc.
```

Note that `char` types are not automatically converted to JSON strings, but to integer numbers. A conversion to a string must be specified explicitly:

```cpp
char ch = 'A';                       // ASCII value 65
json j_default = ch;                 // stores integer number 65
json j_string = std::string(1, ch);  // stores string "A"
```

### Arbitrary types conversions

Every type can be serialized in JSON, not just STL containers and scalar types. Usually, you would do something along those lines:

```cpp
namespace ns {
    // a simple struct to model a person
    struct person {
        std::string name;
        std::string address;
        int age;
    };
}

ns::person p = {"Ned Flanders", "744 Evergreen Terrace", 60};

// convert to JSON: copy each value into the JSON object
json j;
j["name"] = p.name;
j["address"] = p.address;
j["age"] = p.age;

// ...

// convert from JSON: copy each value from the JSON object
ns::person p {
    j["name"].get<std::string>(),
    j["address"].get<std::string>(),
    j["age"].get<int>()
};
```

It works, but that's quite a lot of boilerplate... Fortunately, there's a better way:

```cpp
// create a person
ns::person p {"Ned Flanders", "744 Evergreen Terrace", 60};

// conversion: person -> json
json j = p;

std::cout << j << std::endl;
// {"address":"744 Evergreen Terrace","age":60,"name":"Ned Flanders"}

// conversion: json -> person
auto p2 = j.get<ns::person>();

// that's it
assert(p == p2);
```

#### Basic usage

To make this work with one of your types, you only need to provide two functions:

```cpp
using nlohmann::json;

namespace ns {
    void to_json(json& j, const person& p) {
        j = json{{"name", p.name}, {"address", p.address}, {"age", p.age}};
    }

    void from_json(const json& j, person& p) {
        j.at("name").get_to(p.name);
        j.at("address").get_to(p.address);
        j.at("age").get_to(p.age);
    }
} // namespace ns
```

That's all! When calling the `json` constructor with your type, your custom `to_json` method will be automatically called.
Likewise, when calling `get<your_type>()` or `get_to(your_type&)`, the `from_json` method will be called.

Some important things:

* Those methods **MUST** be in your type's namespace (which can be the global namespace), or the library will not be able to locate them (in this example, they are in namespace `ns`, where `person` is defined).
* Those methods **MUST** be available (e.g., properly headers must be included) everywhere you use these conversions. Look at [issue 1108](https://github.com/nlohmann/json/issues/1108) for errors that may occur otherwise.
* When using `get<your_type>()`, `your_type` **MUST** be [DefaultConstructible](https://en.cppreference.com/w/cpp/named_req/DefaultConstructible). (There is a way to bypass this requirement described later.)
* In function `from_json`, use function [`at()`](https://nlohmann.github.io/json/classnlohmann_1_1basic__json_a93403e803947b86f4da2d1fb3345cf2c.html#a93403e803947b86f4da2d1fb3345cf2c) to access the object values rather than `operator[]`. In case a key does not exist, `at` throws an exception that you can handle, whereas `operator[]` exhibits undefined behavior.
* You do not need to add serializers or deserializers for STL types like `std::vector`: the library already implements these.


#### How do I convert third-party types?

This requires a bit more advanced technique. But first, let's see how this conversion mechanism works:

The library uses **JSON Serializers** to convert types to json.
The default serializer for `nlohmann::json` is `nlohmann::adl_serializer` (ADL means [Argument-Dependent Lookup](https://en.cppreference.com/w/cpp/language/adl)).

It is implemented like this (simplified):

```cpp
template <typename T>
struct adl_serializer {
    static void to_json(json& j, const T& value) {
        // calls the "to_json" method in T's namespace
    }

    static void from_json(const json& j, T& value) {
        // same thing, but with the "from_json" method
    }
};
```

This serializer works fine when you have control over the type's namespace. However, what about `boost::optional` or `std::filesystem::path` (C++17)? Hijacking the `boost` namespace is pretty bad, and it's illegal to add something other than template specializations to `std`...

To solve this, you need to add a specialization of `adl_serializer` to the `nlohmann` namespace, here's an example:

```cpp
// partial specialization (full specialization works too)
namespace nlohmann {
    template <typename T>
    struct adl_serializer<boost::optional<T>> {
        static void to_json(json& j, const boost::optional<T>& opt) {
            if (opt == boost::none) {
                j = nullptr;
            } else {
              j = *opt; // this will call adl_serializer<T>::to_json which will
                        // find the free function to_json in T's namespace!
            }
        }

        static void from_json(const json& j, boost::optional<T>& opt) {
            if (j.is_null()) {
                opt = boost::none;
            } else {
                opt = j.get<T>(); // same as above, but with
                                  // adl_serializer<T>::from_json
            }
        }
    };
}
```

#### How can I use `get()` for non-default constructible/non-copyable types?

There is a way, if your type is [MoveConstructible](https://en.cppreference.com/w/cpp/named_req/MoveConstructible). You will need to specialize the `adl_serializer` as well, but with a special `from_json` overload:

```cpp
struct move_only_type {
    move_only_type() = delete;
    move_only_type(int ii): i(ii) {}
    move_only_type(const move_only_type&) = delete;
    move_only_type(move_only_type&&) = default;

    int i;
};

namespace nlohmann {
    template <>
    struct adl_serializer<move_only_type> {
        // note: the return type is no longer 'void', and the method only takes
        // one argument
        static move_only_type from_json(const json& j) {
            return {j.get<int>()};
        }

        // Here's the catch! You must provide a to_json method! Otherwise you
        // will not be able to convert move_only_type to json, since you fully
        // specialized adl_serializer on that type
        static void to_json(json& j, move_only_type t) {
            j = t.i;
        }
    };
}
```

#### Can I write my own serializer? (Advanced use)

Yes. You might want to take a look at [`unit-udt.cpp`](https://github.com/nlohmann/json/blob/develop/test/src/unit-udt.cpp) in the test suite, to see a few examples.

If you write your own serializer, you'll need to do a few things:

- use a different `basic_json` alias than `nlohmann::json` (the last template parameter of `basic_json` is the `JSONSerializer`)
- use your `basic_json` alias (or a template parameter) in all your `to_json`/`from_json` methods
- use `nlohmann::to_json` and `nlohmann::from_json` when you need ADL

Here is an example, without simplifications, that only accepts types with a size <= 32, and uses ADL.

```cpp
// You should use void as a second template argument
// if you don't need compile-time checks on T
template<typename T, typename SFINAE = typename std::enable_if<sizeof(T) <= 32>::type>
struct less_than_32_serializer {
    template <typename BasicJsonType>
    static void to_json(BasicJsonType& j, T value) {
        // we want to use ADL, and call the correct to_json overload
        using nlohmann::to_json; // this method is called by adl_serializer,
                                 // this is where the magic happens
        to_json(j, value);
    }

    template <typename BasicJsonType>
    static void from_json(const BasicJsonType& j, T& value) {
        // same thing here
        using nlohmann::from_json;
        from_json(j, value);
    }
};
```

Be **very** careful when reimplementing your serializer, you can stack overflow if you don't pay attention:

```cpp
template <typename T, void>
struct bad_serializer
{
    template <typename BasicJsonType>
    static void to_json(BasicJsonType& j, const T& value) {
      // this calls BasicJsonType::json_serializer<T>::to_json(j, value);
      // if BasicJsonType::json_serializer == bad_serializer ... oops!
      j = value;
    }

    template <typename BasicJsonType>
    static void to_json(const BasicJsonType& j, T& value) {
      // this calls BasicJsonType::json_serializer<T>::from_json(j, value);
      // if BasicJsonType::json_serializer == bad_serializer ... oops!
      value = j.template get<T>(); // oops!
    }
};
```

### Specializing enum conversion

By default, enum values are serialized to JSON as integers. In some cases this could result in undesired behavior. If an enum is modified or re-ordered after data has been serialized to JSON, the later de-serialized JSON data may be undefined or a different enum value than was originally intended.

It is possible to more precisely specify how a given enum is mapped to and from JSON as shown below:

```cpp
// example enum type declaration
enum TaskState {
    TS_STOPPED,
    TS_RUNNING,
    TS_COMPLETED,
    TS_INVALID=-1,
};

// map TaskState values to JSON as strings
NLOHMANN_JSON_SERIALIZE_ENUM( TaskState, {
    {TS_INVALID, nullptr},
    {TS_STOPPED, "stopped"},
    {TS_RUNNING, "running"},
    {TS_COMPLETED, "completed"},
});
```

The `NLOHMANN_JSON_SERIALIZE_ENUM()` macro declares a set of `to_json()` / `from_json()` functions for type `TaskState` while avoiding repetition and boilerplate serilization code.

**Usage:**

```cpp
// enum to JSON as string
json j = TS_STOPPED;
assert(j == "stopped");

// json string to enum
json j3 = "running";
assert(j3.get<TaskState>() == TS_RUNNING);

// undefined json value to enum (where the first map entry above is the default)
json jPi = 3.14;
assert(jPi.get<TaskState>() == TS_INVALID );
```

Just as in [Arbitrary Type Conversions](#arbitrary-types-conversions) above,
- `NLOHMANN_JSON_SERIALIZE_ENUM()` MUST be declared in your enum type's namespace (which can be the global namespace), or the library will not be able to locate it and it will default to integer serialization.
- It MUST be available (e.g., proper headers must be included) everywhere you use the conversions.

Other Important points:
- When using `get<ENUM_TYPE>()`, undefined JSON values will default to the first pair specified in your map. Select this default pair carefully.
- If an enum or JSON value is specified more than once in your map, the first matching occurrence from the top of the map will be returned when converting to or from JSON.

### Binary formats (BSON, CBOR, MessagePack, and UBJSON)

Though JSON is a ubiquitous data format, it is not a very compact format suitable for data exchange, for instance over a network. Hence, the library supports [BSON](http://bsonspec.org) (Binary JSON), [CBOR](http://cbor.io) (Concise Binary Object Representation), [MessagePack](http://msgpack.org), and [UBJSON](http://ubjson.org) (Universal Binary JSON Specification) to efficiently encode JSON values to byte vectors and to decode such vectors.

```cpp
// create a JSON value
json j = R"({"compact": true, "schema": 0})"_json;

// serialize to BSON
std::vector<std::uint8_t> v_bson = json::to_bson(j);

// 0x1B, 0x00, 0x00, 0x00, 0x08, 0x63, 0x6F, 0x6D, 0x70, 0x61, 0x63, 0x74, 0x00, 0x01, 0x10, 0x73, 0x63, 0x68, 0x65, 0x6D, 0x61, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00

// roundtrip
json j_from_bson = json::from_bson(v_bson);

// serialize to CBOR
std::vector<std::uint8_t> v_cbor = json::to_cbor(j);

// 0xA2, 0x67, 0x63, 0x6F, 0x6D, 0x70, 0x61, 0x63, 0x74, 0xF5, 0x66, 0x73, 0x63, 0x68, 0x65, 0x6D, 0x61, 0x00

// roundtrip
json j_from_cbor = json::from_cbor(v_cbor);

// serialize to MessagePack
std::vector<std::uint8_t> v_msgpack = json::to_msgpack(j);

// 0x82, 0xA7, 0x63, 0x6F, 0x6D, 0x70, 0x61, 0x63, 0x74, 0xC3, 0xA6, 0x73, 0x63, 0x68, 0x65, 0x6D, 0x61, 0x00

// roundtrip
json j_from_msgpack = json::from_msgpack(v_msgpack);

// serialize to UBJSON
std::vector<std::uint8_t> v_ubjson = json::to_ubjson(j);

// 0x7B, 0x69, 0x07, 0x63, 0x6F, 0x6D, 0x70, 0x61, 0x63, 0x74, 0x54, 0x69, 0x06, 0x73, 0x63, 0x68, 0x65, 0x6D, 0x61, 0x69, 0x00, 0x7D

// roundtrip
json j_from_ubjson = json::from_ubjson(v_ubjson);
```


## Supported compilers

Though it's 2019 already, the support for C++11 is still a bit sparse. Currently, the following compilers are known to work:

- GCC 4.8 - 9.0 (and possibly later)
- Clang 3.4 - 8.0 (and possibly later)
- Intel C++ Compiler 17.0.2 (and possibly later)
- Microsoft Visual C++ 2015 / Build Tools 14.0.25123.0 (and possibly later)
- Microsoft Visual C++ 2017 / Build Tools 15.5.180.51428 (and possibly later)