Repository files navigation

• Design goals
• Sponsors
• Support (documentation,FAQ,discussions,API,bug issues)
• Examples
  • Read JSON from a file
  • Creatingjson objects from JSON literals
  • JSON as first-class data type
  • Serialization / Deserialization
  • STL-like access
  • Conversion from STL containers
  • JSON Pointer and JSON Patch
  • JSON Merge Patch
  • Implicit conversions
  • Conversions to/from arbitrary types
  • Specializing enum conversion
  • Binary formats (BSON, CBOR, MessagePack, UBJSON, and BJData)
• Supported compilers
• Integration
  • CMake
  • Package Managers
  • Pkg-config
• License
• Contact
• Thanks
• Used third-party tools
• Projects using JSON for Modern C++
• Notes
• Execute unit tests

There are myriads ofJSON 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 theexamples below and you'll know what I mean.

• Trivial integration. Our whole code consists of a single header filejson.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 code is heavilyunit-tested and covers100% of the code, including all exceptional behavior. Furthermore, we checked withValgrind and theClang Sanitizers that there are no memory leaks.Google OSS-Fuzz additionally runs fuzz tests against all parsers 24/7, effectively executing billions of tests so far. To maintain high quality, the project is following theCore Infrastructure Initiative (CII) best practices.

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 ordouble for numbers,std::map for objects,std::vector for arrays, andbool for Booleans. However, you can template the generalized classbasic_json to your needs.

• Speed. There are certainlyfaster JSON libraries 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 astd::vector orstd::map, you are already set.

See thecontribution guidelines for more information.

You can sponsor this library atGitHub Sponsors.

• Michael Hartmann
• Stefan Hagen
• Steve Sperandeo
• Robert Jefe Lindstädt
• Steve Wagner

Thanks everyone!

❓ If you have aquestion, please check if it is already answered in theFAQ or theQ&A section. If not, pleaseask a new question there.

📚 If you want tolearn more about how to use the library, check out the rest of theREADME, have a look atcode examples, or browse through thehelp pages.

🚧 If you want to understand theAPI better, check out theAPI Reference.

🐛 If you found abug, please check theFAQ if it is a known issue or the result of a design decision. Please also have a look at theissue list before youcreate a new issue. Please provide as much information as possible to help us understand and reproduce your issue.

There is also adocset for the documentation browsersDash,Velocity, andZeal that contains the fulldocumentation as offline resource.

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

Beside the examples below, you may want to:

→ Check thedocumentation
→ Browse thestandalone example files

Every API function (documented in theAPI Documentation) has a corresponding standalone example file. For example, theemplace() function has a matchingemplace.cpp example file.

Thejson class provides an API for manipulating a JSON value. To create ajson object by reading a JSON file:

#include<fstream>#include<nlohmann/json.hpp>using json = nlohmann::json;// ...std::ifstreamf("example.json");json data = json::parse(f);

Assume you want to create hard-code this literal JSON value in a file, as ajson object:

{"pi":3.141,"happy":true}

There are various options:

// Using (raw) string literals and json::parsejson ex1 = json::parse(R"(  {    "pi": 3.141,    "happy": true  })");// Using user-defined (raw) string literalsusingnamespacenlohmann::literals;json ex2 =R"(  {    "pi": 3.141,    "happy": true  })"_json;// Using initializer listsjson ex3 = {  {"happy",true},  {"pi",3.141},};

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

Assume you want to create the JSON object

{"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:

// 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 boolj["happy"] =true;// add a string that is stored as std::stringj["name"] ="Niels";// add another null object by passing nullptrj["nothing"] =nullptr;// add an object inside the objectj["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 functionsjson::array() andjson::object() will help:

// 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} });

You can create a JSON value (deserialization) by appending_json to a string literal:

// create object from string literaljson j ="{\"happy\": true,\"pi\": 3.141 }"_json;// or even nicer with a raw string literalauto 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 stringvalue. 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 string literal should be brought into scope with withusing namespace nlohmann::literals;(seejson::parse()).

The above example can also be expressed explicitly usingjson::parse():

// parse explicitlyauto j3 = json::parse(R"({"happy": true, "pi": 3.141})");

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

// explicit conversion to stringstd::string s = j.dump();// {"happy":true,"pi":3.141}// serialization with pretty printing// pass in the amount of spaces to indentstd::cout << j.dump(4) << std::endl;// {//     "happy": true,//     "pi": 3.141// }

Note the difference between serialization and assignment:

// store a string in a JSON valuejson j_string ="this is a string";// retrieve the string valueauto cpp_string = j_string.get<std::string>();// retrieve the string value (alternative when a 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 stringstd::cout << cpp_string <<" ==" << cpp_string2 <<" ==" << j_string.get<std::string>() <<'\n';// output of serialized valuestd::cout << j_string <<" ==" << serialized_string << std::endl;

.dump() returns the originally stored string value.

Note the library only supports UTF-8. When you store strings with different encodings in the library, callingdump() may throw an exception unlessjson::error_handler_t::replace orjson::error_handler_t::ignore are used as error handlers.

You can also use streams to serialize and deserialize:

// deserialize from standard inputjson j;std::cin >> j;// serialize to standard outputstd::cout << j;// the setw manipulator was overloaded to set the indentation for pretty printingstd::cout << std::setw(4) << j << std::endl;

These operators work for any subclasses ofstd::istream orstd::ostream. Here is the same example with files:

// read a JSON filestd::ifstreami("file.json");json j;i >> j;// write prettified JSON to another filestd::ofstreamo("pretty.json");o << std::setw(4) << j << std::endl;

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

You can also parse JSON from an iterator range; that is, from any container accessible by iterators whosevalue_type is an integral type of 1, 2 or 4 bytes, which will be interpreted as UTF-8, UTF-16 and UTF-32 respectively. For instance, astd::vector<std::uint8_t>, or astd::list<std::uint16_t>:

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):

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

Since the parse function accepts arbitrary iterator ranges, you can provide your own data sources by implementing theLegacyInputIterator concept.

structMyContainer {voidadvance();constchar&get_current();};structMyIterator {using difference_type = std::ptrdiff_t;using value_type =char;using pointer =constchar*;using reference =constchar&;using iterator_category = std::input_iterator_tag;    MyIterator&operator++() {        MyContainer.advance();return *this;    }booloperator!=(const MyIterator& rhs)const {return rhs.target != target;    }    referenceoperator*()const {return target.get_current();    }    MyContainer* target =nullptr;};MyIteratorbegin(MyContainer& tgt) {return MyIterator{&tgt};}MyIteratorend(const MyContainer&) {return {};}voidfoo() {    MyContainer c;    json j =json::parse(c);}

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

// called when null is parsedboolnull();// called when a boolean is parsed; value is passedboolboolean(bool val);// called when a signed or unsigned integer number is parsed; value is passedboolnumber_integer(number_integer_t val);boolnumber_unsigned(number_unsigned_t val);// called when a floating-point number is parsed; value and original string is passedboolnumber_float(number_float_t val,conststring_t& s);// called when a string is parsed; value is passed and can be safely moved awayboolstring(string_t& val);// called when a binary value is parsed; value is passed and can be safely moved awayboolbinary(binary_t& val);// called when an object or array begins or ends, resp. The number of elements is passed (or -1 if not known)boolstart_object(std::size_t elements);boolend_object();boolstart_array(std::size_t elements);boolend_array();// called when an object key is parsed; value is passed and can be safely moved awayboolkey(string_t& val);// called when a parse error occurs; byte position, the last token, and an exception is passedboolparse_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 classnlohmann::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. Callbool 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 thesax_parse function only returns abool indicating the result of the last executed SAX event. It does not return ajson 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 yourparse_error implementation. Internally, the SAX interface is used for the DOM parser (classjson_sax_dom_parser) as well as the acceptor (json_sax_acceptor), see filejson_sax.hpp.

We designed the JSON class to behave just like an STL container. In fact, it satisfies theReversibleContainer requirement.

// create an array using push_backjson j;j.push_back("foo");j.push_back(1);j.push_back(true);// also use emplace_backj.emplace_back(1.78);// iterate the arrayfor (json::iterator it = j.begin(); it != j.end(); ++it) {  std::cout << *it <<'\n';}// range-based forfor (auto& element : j) {  std::cout << element <<'\n';}// getter/setterconstauto tmp = j[0].get<std::string>();j[1] =42;bool foo = j.at(2);// comparisonj ==R"(["foo", 1, true, 1.78])"_json;// true// other stuffj.size();// 4 entriesj.empty();// falsej.type();// json::value_t::arrayj.clear();// the array is empty again// convenience type checkersj.is_null();j.is_boolean();j.is_number();j.is_object();j.is_array();j.is_string();// create an objectjson o;o["foo"] =23;o["bar"] =false;o["baz"] =3.141;// also use emplaceo.emplace("weather","sunny");// special iterator member functions for objectsfor (json::iterator it = o.begin(); it != o.end(); ++it) {  std::cout << it.key() <<" :" << it.value() <<"\n";}// the same code as range forfor (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 entryif (o.contains("foo")) {// there is an entry with key "foo"}// or via find and an iteratorif (o.find("foo") != o.end()) {// there is an entry with key "foo"}// or simpler using count()int foo_present = o.count("foo");// 1int fob_present = o.count("fob");// 0// delete an entryo.erase("foo");

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.

std::vector<int> c_vector {1,2,3,4};jsonj_vec(c_vector);// [1, 2, 3, 4]std::deque<double> c_deque {1.2,2.3,3.4,5.6};jsonj_deque(c_deque);// [1.2, 2.3, 3.4, 5.6]std::list<bool> c_list {true,true,false,true};jsonj_list(c_list);// [true, true, false, true]std::forward_list<int64_t> c_flist {12345678909876,23456789098765,34567890987654,45678909876543};jsonj_flist(c_flist);// [12345678909876, 23456789098765, 34567890987654, 45678909876543]std::array<unsignedlong,4> c_array {{1,2,3,4}};jsonj_array(c_array);// [1, 2, 3, 4]std::set<std::string> c_set {"one","two","three","four","one"};jsonj_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"};jsonj_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"};jsonj_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"};jsonj_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 anstd::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.

std::map<std::string,int> c_map { {"one",1}, {"two",2}, {"three",3} };jsonj_map(c_map);// {"one": 1, "three": 3, "two": 2 }std::unordered_map<constchar*,double> c_umap { {"one",1.2}, {"two",2.3}, {"three",3.4} };jsonj_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} };jsonj_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} };jsonj_ummap(c_ummap);// only one entry for key "three" is used// maybe {"one": true, "two": true, "three": true}

The library supportsJSON Pointer (RFC 6901) as alternative means to address structured values. On top of this,JSON Patch (RFC 6902) allows describing differences between two JSON values - effectively allowing patch and diff operations known from Unix.

// a JSON valuejson 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 patchjson j_result = j_original.patch(j_patch);// {//    "baz": "boo",//    "hello": ["world"]// }// calculate a JSON patch from two JSON valuesjson::diff(j_result, j_original);// [//   { "op":" replace", "path": "/baz", "value": ["one", "two", "three"] },//   { "op": "remove","path": "/hello" },//   { "op": "add", "path": "/foo", "value": "bar" }// ]

The library supportsJSON Merge Patch (RFC 7386) 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.

// a JSON valuejson j_document =R"({  "a": "b",  "c": {    "d": "e",    "f": "g"  }})"_json;// a patchjson j_patch =R"({  "a":"z",  "c": {    "f": null  }})"_json;// apply the patchj_document.merge_patch(j_patch);// {//  "a": "z",//  "c": {//    "d": "e"//  }// }

Supported types can be implicitly converted to JSON values.

It is recommended toNOT USE implicit conversionsFROM a JSON value.You can find more details about this recommendationhere.You can switch off implicit conversions by definingJSON_USE_IMPLICIT_CONVERSIONS to0 before including thejson.hpp header. When using CMake, you can also achieve this by setting the optionJSON_ImplicitConversions toOFF.

// stringsstd::string s1 ="Hello, world!";json js = s1;auto s2 = js.get<std::string>();// NOT RECOMMENDEDstd::string s3 = js;std::string s4;s4 = js;// Booleansbool b1 =true;json jb = b1;auto b2 = jb.get<bool>();// NOT RECOMMENDEDbool b3 = jb;bool b4;b4 = jb;// numbersint i =42;json jn = i;auto f = jn.get<double>();// NOT RECOMMENDEDdouble f2 = jb;double f3;f3 = jb;// etc.

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

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

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

namespacens {// a simple struct to model a personstructperson {        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 objectjson j;j["name"] = p.name;j["address"] = p.address;j["age"] = p.age;// ...// convert from JSON: copy each value from the JSON objectns::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:

// create a personns::person p {"Ned Flanders","744 Evergreen Terrace",60};// conversion: person -> jsonjson j = p;std::cout << j << std::endl;// {"address":"744 Evergreen Terrace","age":60,"name":"Ned Flanders"}// conversion: json -> personauto p2 = j.get<ns::person>();// that's itassert(p == p2);

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

using json = nlohmann::json;namespacens {voidto_json(json& j,const person& p) {        j = json{{"name", p.name}, {"address", p.address}, {"age", p.age}};    }voidfrom_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 thejson constructor with your type, your customto_json method will be automatically called.Likewise, when callingget<your_type>() orget_to(your_type&), thefrom_json method will be called.

Some important things:

• Those methodsMUST 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 namespacens, whereperson is defined).
• Those methodsMUST be available (e.g., proper headers must be included) everywhere you use these conversions. Look atissue 1108 for errors that may occur otherwise.
• When usingget<your_type>(),your_typeMUST beDefaultConstructible. (There is a way to bypass this requirement described later.)
• In functionfrom_json, use functionat() to access the object values rather thanoperator[]. In case a key does not exist,at throws an exception that you can handle, whereasoperator[] exhibits undefined behavior.
• You do not need to add serializers or deserializers for STL types likestd::vector: the library already implements these.

If you just want to serialize/deserialize some structs, theto_json/from_json functions can be a lot of boilerplate.

There are two macros to make your life easier as long as you (1) want to use a JSON object as serialization and (2) want to use the member variable names as object keys in that object:

• NLOHMANN_DEFINE_TYPE_NON_INTRUSIVE(name, member1, member2, ...) is to be defined inside the namespace of the class/struct to create code for.
• NLOHMANN_DEFINE_TYPE_INTRUSIVE(name, member1, member2, ...) is to be defined inside the class/struct to create code for. This macro can also access private members.

In both macros, the first parameter is the name of the class/struct, and all remaining parameters name the members.

Theto_json/from_json functions for theperson struct above can be created with:

namespacens {NLOHMANN_DEFINE_TYPE_NON_INTRUSIVE(person, name, address, age)}

Here is an example with private members, whereNLOHMANN_DEFINE_TYPE_INTRUSIVE is needed:

namespacens {classaddress {private:        std::string street;int housenumber;int postcode;public:NLOHMANN_DEFINE_TYPE_INTRUSIVE(address, street, housenumber, postcode)    };}

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

The library usesJSON Serializers to convert types to json.The default serializer fornlohmann::json isnlohmann::adl_serializer (ADL meansArgument-Dependent Lookup).

It is implemented like this (simplified):

template<typename T>structadl_serializer {staticvoidto_json(json& j,const T& value) {// calls the "to_json" method in T's namespace    }staticvoidfrom_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 aboutboost::optional orstd::filesystem::path (C++17)? Hijacking theboost namespace is pretty bad, and it's illegal to add something other than template specializations tostd...

To solve this, you need to add a specialization ofadl_serializer to thenlohmann namespace, here's an example:

// partial specialization (full specialization works too)namespacenlohmann {template<typename T>structadl_serializer<boost::optional<T>> {staticvoidto_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!            }        }staticvoidfrom_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            }        }    };}

There is a way, if your type isMoveConstructible. You will need to specialize theadl_serializer as well, but with a specialfrom_json overload:

structmove_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;};namespacenlohmann {template<>structadl_serializer<move_only_type> {// note: the return type is no longer 'void', and the method only takes// one argumentstatic move_only_typefrom_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 typestaticvoidto_json(json& j, move_only_type t) {            j = t.i;        }    };}

Yes. You might want to take a look atunit-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 differentbasic_json alias thannlohmann::json (the last template parameter ofbasic_json is theJSONSerializer)
• use yourbasic_json alias (or a template parameter) in all yourto_json/from_json methods
• usenlohmann::to_json andnlohmann::from_json when you need ADL

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

// You should use void as a second template argument// if you don't need compile-time checks on Ttemplate<typename T,typename SFINAE =typename std::enable_if<sizeof(T) <=32>::type>structless_than_32_serializer {template<typename BasicJsonType>staticvoidto_json(BasicJsonType& j, T value) {// we want to use ADL, and call the correct to_json overloadusing nlohmann::to_json;// this method is called by adl_serializer,// this is where the magic happensto_json(j, value);    }template<typename BasicJsonType>staticvoidfrom_json(const BasicJsonType& j, T& value) {// same thing hereusing nlohmann::from_json;from_json(j, value);    }};

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

template<typename T,void>structbad_serializer{template<typename BasicJsonType>staticvoidto_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>staticvoidto_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.templateget<T>();// oops!    }};

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:

// example enum type declarationenum TaskState {    TS_STOPPED,    TS_RUNNING,    TS_COMPLETED,    TS_INVALID=-1,};// map TaskState values to JSON as stringsNLOHMANN_JSON_SERIALIZE_ENUM( TaskState, {    {TS_INVALID,nullptr},    {TS_STOPPED,"stopped"},    {TS_RUNNING,"running"},    {TS_COMPLETED,"completed"},})

TheNLOHMANN_JSON_SERIALIZE_ENUM() macro declares a set ofto_json() /from_json() functions for typeTaskState while avoiding repetition and boilerplate serialization code.

Usage:

// enum to JSON as stringjson j = TS_STOPPED;assert(j =="stopped");// json string to enumjson 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 inArbitrary Type 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 usingget<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.

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 supportsBSON (Binary JSON),CBOR (Concise Binary Object Representation),MessagePack,UBJSON (Universal Binary JSON Specification) andBJData (Binary JData) to efficiently encode JSON values to byte vectors and to decode such vectors.

// create a JSON valuejson j =R"({"compact": true, "schema": 0})"_json;// serialize to BSONstd::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// roundtripjson j_from_bson = json::from_bson(v_bson);// serialize to CBORstd::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// roundtripjson j_from_cbor = json::from_cbor(v_cbor);// serialize to MessagePackstd::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// roundtripjson j_from_msgpack = json::from_msgpack(v_msgpack);// serialize to UBJSONstd::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// roundtripjson j_from_ubjson = json::from_ubjson(v_ubjson);

The library also supports binary types from BSON, CBOR (byte strings), and MessagePack (bin, ext, fixext). They are stored by default asstd::vector<std::uint8_t> to be processed outside the library.

// CBOR byte string with payload 0xCAFEstd::vector<std::uint8_t> v = {0x42,0xCA,0xFE};// read valuejson j = json::from_cbor(v);// the JSON value has type binaryj.is_binary();// true// get reference to stored binary valueauto& binary = j.get_binary();// the binary value has no subtype (CBOR has no binary subtypes)binary.has_subtype();// false// access std::vector<std::uint8_t> member functionsbinary.size();// 2binary[0];// 0xCAbinary[1];// 0xFE// set subtype to 0x10binary.set_subtype(0x10);// serialize to MessagePackauto cbor = json::to_msgpack(j);// 0xD5 (fixext2), 0x10, 0xCA, 0xFE

Though it's 2022 already, the support for C++11 is still a bit sparse. Currently, the following compilers are known to work:

• GCC 4.8 - 12.0 (and possibly later)
• Clang 3.4 - 15.0 (and possibly later)
• Apple Clang 9.1 - 13.1 (and possibly later)
• Intel C++ Compiler 17.0.2 (and possibly later)
• Nvidia CUDA Compiler 11.0.221 (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)
• Microsoft Visual C++ 2019 / Build Tools 16.3.1+1def00d3d (and possibly later)
• Microsoft Visual C++ 2022 / Build Tools 19.30.30709.0 (and possibly later)

I would be happy to learn about other compilers/versions.

Please note:

• GCC 4.8 has a bug57824): multiline raw strings cannot be the arguments to macros. Don't use multiline raw strings directly in macros with this compiler.

• Android defaults to using very old compilers and C++ libraries. To fix this, add the following to yourApplication.mk. This will switch to the LLVM C++ library, the Clang compiler, and enable C++11 and other features disabled by default.

  APP_STL := c++_sharedNDK_TOOLCHAIN_VERSION := clang3.6APP_CPPFLAGS += -frtti -fexceptions

  The code compiles successfully withAndroid NDK, Revision 9 - 11 (and possibly later) andCrystaX's Android NDK version 10.

• For GCC running on MinGW or Android SDK, the error'to_string' is not a member of 'std' (or similarly, forstrtod orstrtof) may occur. Note this is not an issue with the code, but rather with the compiler itself. On Android, see above to build with a newer environment. For MinGW, please refer tothis site andthis discussion for information on how to fix this bug. For Android NDK usingAPP_STL := gnustl_static, please refer tothis discussion.

• Unsupported versions of GCC and Clang are rejected by#error directives. This can be switched off by definingJSON_SKIP_UNSUPPORTED_COMPILER_CHECK. Note that you can expect no support in this case.

The following compilers are currently used in continuous integration atAppVeyor,Drone CI, andGitHub Actions:

json.hpp is the single required file insingle_include/nlohmann orreleased here. You need to add

#include<nlohmann/json.hpp>// for convenienceusing 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 fileinclude/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.

You can also use thenlohmann_json::nlohmann_json interface target in CMake. This target populates the appropriate usage requirements forINTERFACE_INCLUDE_DIRECTORIES to point to the appropriate include directories andINTERFACE_COMPILE_FEATURES for the necessary C++11 flags.

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

# CMakeLists.txtfind_package(nlohmann_json 3.2.0 REQUIRED)...add_library(foo ...)...target_link_libraries(fooPRIVATE 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.

To embed the library directly into an existing CMake project, place the entire source tree in a subdirectory and calladd_subdirectory() in yourCMakeLists.txt file:

# 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_BuildTestsOFFCACHEINTERNAL"")# If you only include this third party in PRIVATE source files, you do not# need to install it when your main project gets installed.# set(JSON_Install 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(fooPRIVATE nlohmann_json::nlohmann_json)

Since CMake v3.11,FetchContent canbe used to automatically download a release as a dependency at configure time.

Example:

include(FetchContent)FetchContent_Declare(json URL https://github.com/nlohmann/json/releases/download/v3.11.2/json.tar.xz)FetchContent_MakeAvailable(json)target_link_libraries(fooPRIVATE nlohmann_json::nlohmann_json)

Note: It is recommended to use the URL approach described above which is supported as of version 3.10.0. Seehttps://json.nlohmann.me/integration/cmake/#fetchcontent for more information.

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

# Top level CMakeLists.txtproject(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 methodtarget_link_libraries(fooPRIVATE nlohmann_json::nlohmann_json)

# thirdparty/CMakeLists.txt...if(FOO_USE_EXTERNAL_JSON)  find_package(nlohmann_json 3.2.0 REQUIRED)else()  set(JSON_BuildTestsOFFCACHEINTERNAL"")  add_subdirectory(nlohmann_json)endif()...

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

🍺 If you are using OS X andHomebrew, just typebrew install nlohmann-json and you're set. If you want the bleeding edge rather than the latest release, usebrew install nlohmann-json --HEAD. Seenlohmann-json for more information.

If you are using theMeson Build System, add this source tree as ameson subproject. You may also use theinclude.zip published in this project'sReleases to reduce the size of the vendored source tree. Alternatively, you can get a wrap file by downloading it fromMeson WrapDB, or simply usemeson wrap install nlohmann_json. Please see the meson project for any issues regarding the packaging.

The providedmeson.build can also be used as an alternative to cmake for installingnlohmann_json system-wide in which case a pkg-config file is installed. To use it, simply have your build system require thenlohmann_json pkg-config dependency. In Meson, it is preferred to use thedependency() object with a subproject fallback, rather than using the subproject directly.

If you are usingConan to manage your dependencies, merely addnlohmann_json/x.y.z to yourconanfile's requires, wherex.y.z is the release version you want to use. Please file issueshere if you experience problems with the packages.

If you are usingSpack to manage your dependencies, you can use thenlohmann-json package. Please see thespack project for any issues regarding the packaging.

If you are usinghunter on your project for external dependencies, then you can use thenlohmann_json package. Please see the hunter project for any issues regarding the packaging.

If you are usingBuckaroo, you can install this library's module withbuckaroo add github.com/buckaroo-pm/nlohmann-json. Please file issueshere. There is a demo repohere.

If you are usingvcpkg on your project for external dependencies, then you can install thenlohmann-json package withvcpkg install nlohmann-json and follow the then displayed descriptions. Please see the vcpkg project for any issues regarding the packaging.

If you are usingcget, you can install the latest development version withcget install nlohmann/json. A specific version can be installed withcget 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 usingCocoaPods, you can use the library by adding pod"nlohmann_json", '~>3.1.2' to your podfile (seean example). Please file issueshere.

If you are usingNuGet, you can use the packagenlohmann.json. Please checkthis extensive description on how to use the package. Please file issueshere.

If you are usingconda, you can use the packagenlohmann_json fromconda-forge executingconda install -c conda-forge nlohmann_json. Please file issueshere.

If you are usingMSYS2, you can use themingw-w64-nlohmann-json package, just typepacman -S mingw-w64-i686-nlohmann-json orpacman -S mingw-w64-x86_64-nlohmann-json for installation. Please file issueshere if you experience problems with the packages.

If you are usingMacPorts, executesudo port install nlohmann-json to install thenlohmann-json package.

If you are usingbuild2, you can use thenlohmann-json package from the public repositoryhttps://cppget.org or directly from thepackage's sources repository. In your project'smanifest file, just adddepends: nlohmann-json (probably with someversion constraints). If you are not familiar with using dependencies inbuild2,please read this introduction.Please file issueshere if you experience problems with the packages.

If you are usingwsjcpp, you can use the commandwsjcpp install "https://github.com/nlohmann/json:develop" to get the latest version. Note you can change the branch ":develop" to an existing tag or another branch.

If you are usingCPM.cmake, you can check thisexample. Afteradding CPM script to your project, implement the following snippet to your CMake:

CPMAddPackage(NAME nlohmann_json    GITHUB_REPOSITORY nlohmann/jsonVERSION 3.9.1)

If you are using bare Makefiles, you can usepkg-config to generate the include flags that point to where the library is installed:

pkg-config nlohmann_json --cflags

Users of the Meson build system will also be able to use a system-wide library, which will be found bypkg-config:

json=dependency('nlohmann_json',required:true)

The class is licensed under theMIT License:

Copyright © 2013-2022Niels Lohmann

Permission is hereby granted, free of charge, to any person obtaining a copy of this software and associated documentation files (the “Software”), to deal in the Software without restriction, including without limitation the rights to use, copy, modify, merge, publish, distribute, sublicense, and/or sell copies of the Software, and to permit persons to whom the Software is furnished to do so, subject to the following conditions:

The above copyright notice and this permission notice shall be included in all copies or substantial portions of the Software.

THE SOFTWARE IS PROVIDED “AS IS”, WITHOUT WARRANTY OF ANY KIND, EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.

The class contains the UTF-8 Decoder from Bjoern Hoehrmann which is licensed under theMIT License (see above). Copyright © 2008-2009Björn Hoehrmannbjoern@hoehrmann.de

The class contains a slightly modified version of the Grisu2 algorithm from Florian Loitsch which is licensed under theMIT License (see above). Copyright © 2009Florian Loitsch

The class contains a copy ofHedley from Evan Nemerson which is licensed asCC0-1.0.

The class contains parts ofGoogle Abseil which is licensed under theApache 2.0 License.

If you have questions regarding the library, I would like to invite you toopen an issue at GitHub. Please describe your request, problem, or question as detailed as possible, and also mention the version of the library you are using as well as the version of your compiler and operating system. Opening an issue at GitHub allows other users and contributors to this library to collaborate. For instance, I have little experience with MSVC, and most issues in this regard have been solved by a growing community. If you have a look at theclosed issues, you will see that we react quite timely in most cases.

Only if your request would contain confidential information, pleasesend me an email. For encrypted messages, please usethis key.

Commits by Niels Lohmann andreleases are signed with thisPGP Key.

I deeply appreciate the help of the following people.

1. Teemperor implemented CMake support and lcov integration, realized escape and Unicode handling in the string parser, and fixed the JSON serialization.
2. elliotgoodrich fixed an issue with double deletion in the iterator classes.
3. kirkshoop made the iterators of the class composable to other libraries.
4. wancw fixed a bug that hindered the class to compile with Clang.
5. Tomas Åblad found a bug in the iterator implementation.
6. Joshua C. Randall fixed a bug in the floating-point serialization.
7. Aaron Burghardt implemented code to parse streams incrementally. Furthermore, he greatly improved the parser class by allowing the definition of a filter function to discard undesired elements while parsing.
8. Daniel Kopeček fixed a bug in the compilation with GCC 5.0.
9. Florian Weber fixed a bug in and improved the performance of the comparison operators.
10. Eric Cornelius pointed out a bug in the handling with NaN and infinity values. He also improved the performance of the string escaping.
11. 易思龙 implemented a conversion from anonymous enums.
12. kepkin patiently pushed forward the support for Microsoft Visual studio.
13. gregmarr simplified the implementation of reverse iterators and helped with numerous hints and improvements. In particular, he pushed forward the implementation of user-defined types.
14. Caio Luppi fixed a bug in the Unicode handling.
15. dariomt fixed some typos in the examples.
16. Daniel Frey cleaned up some pointers and implemented exception-safe memory allocation.
17. Colin Hirsch took care of a small namespace issue.
18. Huu Nguyen correct a variable name in the documentation.
19. Silverweed overloadedparse() to accept an rvalue reference.
20. dariomt fixed a subtlety in MSVC type support and implemented theget_ref() function to get a reference to stored values.
21. ZahlGraf added a workaround that allows compilation using Android NDK.
22. whackashoe replaced a function that was marked as unsafe by Visual Studio.
23. 406345 fixed two small warnings.
24. Glen Fernandes noted a potential portability problem in thehas_mapped_type function.
25. Corbin Hughes fixed some typos in the contribution guidelines.
26. twelsby fixed the array subscript operator, an issue that failed the MSVC build, and floating-point parsing/dumping. He further added support for unsigned integer numbers and implemented better roundtrip support for parsed numbers.
27. Volker Diels-Grabsch fixed a link in the README file.
28. msm- added support for American Fuzzy Lop.
29. Annihil fixed an example in the README file.
30. Themercee noted a wrong URL in the README file.
31. Lv Zheng fixed a namespace issue withint64_t anduint64_t.
32. abc100m analyzed the issues with GCC 4.8 and proposed apartial solution.
33. zewt added useful notes to the README file about Android.
34. Róbert Márki added a fix to use move iterators and improved the integration via CMake.
35. Chris Kitching cleaned up the CMake files.
36. Tom Needham fixed a subtle bug with MSVC 2015 which was also proposed byMichael K..
37. Mário Feroldi fixed a small typo.
38. duncanwerner found a really embarrassing performance regression in the 2.0.0 release.
39. Damien fixed one of the last conversion warnings.
40. Thomas Braun fixed a warning in a test case and adjusted MSVC calls in the CI.
41. Théo DELRIEU patiently and constructively oversaw the long way towarditerator-range parsing. He also implemented the magic behind the serialization/deserialization of user-defined types and split the single header file into smaller chunks.
42. Stefan fixed a minor issue in the documentation.
43. Vasil Dimov fixed the documentation regarding conversions fromstd::multiset.
44. ChristophJud overworked the CMake files to ease project inclusion.
45. Vladimir Petrigo made a SFINAE hack more readable and added Visual Studio 17 to the build matrix.
46. Denis Andrejew fixed a grammar issue in the README file.
47. Pierre-Antoine Lacaze found a subtle bug in thedump() function.
48. TurpentineDistillery pointed tostd::locale::classic() to avoid too much locale joggling, found some nice performance improvements in the parser, improved the benchmarking code, and realized locale-independent number parsing and printing.
49. cgzones had an idea how to fix the Coverity scan.
50. Jared Grubb silenced a nasty documentation warning.
51. Yixin Zhang fixed an integer overflow check.
52. Bosswestfalen merged two iterator classes into a smaller one.
53. Daniel599 helped to get Travis execute the tests with Clang's sanitizers.
54. Jonathan Lee fixed an example in the README file.
55. gnzlbg supported the implementation of user-defined types.
56. Alexej Harm helped to get the user-defined types working with Visual Studio.
57. Jared Grubb supported the implementation of user-defined types.
58. EnricoBilla noted a typo in an example.
59. Martin Hořeňovský found a way for a 2x speedup for the compilation time of the test suite.
60. ukhegg found proposed an improvement for the examples section.
61. rswanson-ihi noted a typo in the README.
62. Mihai Stan fixed a bug in the comparison withnullptrs.
63. Tushar Maheshwari addedcotire support to speed up the compilation.
64. TedLyngmo noted a typo in the README, removed unnecessary bit arithmetic, and fixed some-Weffc++ warnings.
65. Krzysztof Woś made exceptions more visible.
66. ftillier fixed a compiler warning.
67. tinloaf made sure all pushed warnings are properly popped.
68. Fytch found a bug in the documentation.
69. Jay Sistar implemented a Meson build description.
70. Henry Lee fixed a warning in ICC and improved the iterator implementation.
71. Vincent Thiery maintains a package for the Conan package manager.
72. Steffen fixed a potential issue with MSVC andstd::min.
73. Mike Tzou fixed some typos.
74. amrcode noted a misleading documentation about comparison of floats.
75. Oleg Endo reduced the memory consumption by replacing<iostream> with<iosfwd>.
76. dan-42 cleaned up the CMake files to simplify including/reusing of the library.
77. Nikita Ofitserov allowed for moving values from initializer lists.
78. Greg Hurrell fixed a typo.
79. Dmitry Kukovinets fixed a typo.
80. kbthomp1 fixed an issue related to the Intel OSX compiler.
81. Markus Werle fixed a typo.
82. WebProdPP fixed a subtle error in a precondition check.
83. Alex noted an error in a code sample.
84. Tom de Geus reported some warnings with ICC and helped to fix them.
85. Perry Kundert simplified reading from input streams.
86. Sonu Lohani fixed a small compilation error.
87. Jamie Seward fixed all MSVC warnings.
88. Nate Vargas added a Doxygen tag file.
89. pvleuven helped to fix a warning in ICC.
90. Pavel helped to fix some warnings in MSVC.
91. Jamie Seward avoided unnecessary string copies infind() andcount().
92. Mitja fixed some typos.
93. Jorrit Wronski updated the Hunter package links.
94. Matthias Möller added a.natvis for the MSVC debug view.
95. bogemic fixed some C++17 deprecation warnings.
96. Eren Okka fixed some MSVC warnings.
97. abolz integrated the Grisu2 algorithm for proper floating-point formatting, allowing more roundtrip checks to succeed.
98. Vadim Evard fixed a Markdown issue in the README.
99. zerodefect fixed a compiler warning.
100. Kert allowed to template the string type in the serialization and added the possibility to override the exceptional behavior.
101. mark-99 helped fixing an ICC error.
102. Patrik Huber fixed links in the README file.
103. johnfb found a bug in the implementation of CBOR's indefinite length strings.
104. Paul Fultz II added a note on the cget package manager.
105. Wilson Lin made the integration section of the README more concise.
106. RalfBielig detected and fixed a memory leak in the parser callback.
107. agrianius allowed to dump JSON to an alternative string type.
108. Kevin Tonon overworked the C++11 compiler checks in CMake.
109. Axel Huebl simplified a CMake check and added support for theSpack package manager.
110. Carlos O'Ryan fixed a typo.
111. James Upjohn fixed a version number in the compilers section.
112. Chuck Atkins adjusted the CMake files to the CMake packaging guidelines and provided documentation for the CMake integration.
113. Jan Schöppach fixed a typo.
114. martin-mfg fixed a typo.
115. Matthias Möller removed the dependency fromstd::stringstream.
116. agrianius added code to use alternative string implementations.
117. Daniel599 allowed to use more algorithms with theitems() function.
118. Julius Rakow fixed the Meson include directory and fixed the links tocppreference.com.
119. Sonu Lohani fixed the compilation with MSVC 2015 in debug mode.
120. grembo fixed the test suite and re-enabled several test cases.
121. Hyeon Kim introduced the macroJSON_INTERNAL_CATCH to control the exception handling inside the library.
122. thyu fixed a compiler warning.
123. David Guthrie fixed a subtle compilation error with Clang 3.4.2.
124. Dennis Fischer allowed to callfind_package without installing the library.
125. Hyeon Kim fixed an issue with a double macro definition.
126. Ben Berman made some error messages more understandable.
127. zakalibit fixed a compilation problem with the Intel C++ compiler.
128. mandreyel fixed a compilation problem.
129. Kostiantyn Ponomarenko added version and license information to the Meson build file.
130. Henry Schreiner added support for GCC 4.8.
131. knilch made sure the test suite does not stall when run in the wrong directory.
132. Antonio Borondo fixed an MSVC 2017 warning.
133. Dan Gendreau implemented theNLOHMANN_JSON_SERIALIZE_ENUM macro to quickly define an enum/JSON mapping.
134. efp added line and column information to parse errors.
135. julian-becker added BSON support.
136. Pratik Chowdhury added support for structured bindings.
137. David Avedissian added support for Clang 5.0.1 (PS4 version).
138. Jonathan Dumaresq implemented an input adapter to read fromFILE*.
139. kjpus fixed a link in the documentation.
140. Manvendra Singh fixed a typo in the documentation.
141. ziggurat29 fixed an MSVC warning.
142. Sylvain Corlay added code to avoid an issue with MSVC.
143. mefyl fixed a bug when JSON was parsed from an input stream.
144. Millian Poquet allowed to install the library via Meson.
145. Michael Behrns-Miller found an issue with a missing namespace.
146. Nasztanovics Ferenc fixed a compilation issue with libc 2.12.
147. Andreas Schwab fixed the endian conversion.
148. Mark-Dunning fixed a warning in MSVC.
149. Gareth Sylvester-Bradley addedoperator/ for JSON Pointers.
150. John-Mark noted a missing header.
151. Vitaly Zaitsev fixed compilation with GCC 9.0.
152. Laurent Stacul fixed compilation with GCC 9.0.
153. Ivor Wanders helped to reduce the CMake requirement to version 3.1.
154. njlr updated the Buckaroo instructions.
155. Lion fixed a compilation issue with GCC 7 on CentOS.
156. Isaac Nickaein improved the integer serialization performance and implemented thecontains() function.
157. past-due suppressed an unfixable warning.
158. Elvis Oric improved Meson support.
159. Matěj Plch fixed an example in the README.
160. Mark Beckwith fixed a typo.
161. scinart fixed bug in the serializer.
162. Patrick Boettcher implementedpush_back() andpop_back() for JSON Pointers.
163. Bruno Oliveira added support for Conda.
164. Michele Caini fixed links in the README.
165. Hani documented how to install the library with NuGet.
166. Mark Beckwith fixed a typo.
167. yann-morin-1998 helped to reduce the CMake requirement to version 3.1.
168. Konstantin Podsvirov maintains a package for the MSYS2 software distro.
169. remyabel added GNUInstallDirs to the CMake files.
170. Taylor Howard fixed a unit test.
171. Gabe Ron implemented theto_string method.
172. Watal M. Iwasaki fixed a Clang warning.
173. Viktor Kirilov switched the unit tests fromCatch todoctest
174. Juncheng E fixed a typo.
175. tete17 fixed a bug in thecontains function.
176. Xav83 fixed some cppcheck warnings.
177. 0xflotus fixed some typos.
178. Christian Deneke added a const version ofjson_pointer::back.
179. Julien Hamaide made theitems() function work with custom string types.
180. Evan Nemerson updated fixed a bug in Hedley and updated this library accordingly.
181. Florian Pigorsch fixed a lot of typos.
182. Camille Bégué fixed an issue in the conversion fromstd::pair andstd::tuple tojson.
183. Anthony VH fixed a compile error in an enum deserialization.
184. Yuriy Vountesmery noted a subtle bug in a preprocessor check.
185. Chen fixed numerous issues in the library.
186. Antony Kellermann added a CI step for GCC 10.1.
187. Alex fixed an MSVC warning.
188. Rainer proposed an improvement in the floating-point serialization in CBOR.
189. Francois Chabot made performance improvements in the input adapters.
190. Arthur Sonzogni documented how the library can be included viaFetchContent.
191. Rimas Misevičius fixed an error message.
192. Alexander Myasnikov fixed some examples and a link in the README.
193. Hubert Chathi made CMake's version config file architecture-independent.
194. OmnipotentEntity implemented the binary values for CBOR, MessagePack, BSON, and UBJSON.
195. ArtemSarmini fixed a compilation issue with GCC 10 and fixed a leak.
196. Evgenii Sopov integrated the library to the wsjcpp package manager.
197. Sergey Linev fixed a compiler warning.
198. Miguel Magalhães fixed the year in the copyright.
199. Gareth Sylvester-Bradley fixed a compilation issue with MSVC.
200. Alexander “weej” Jones fixed an example in the README.
201. Antoine Cœur fixed some typos in the documentation.
202. jothepro updated links to the Hunter package.
203. Dave Lee fixed link in the README.
204. Joël Lamotte added instruction for using Build2's package manager.
205. Paul Jurczak fixed an example in the README.
206. Sonu Lohani fixed a warning.
207. Carlos Gomes Martinho updated the Conan package source.
208. Konstantin Podsvirov fixed the MSYS2 package documentation.
209. Tridacnid improved the CMake tests.
210. Michael fixed MSVC warnings.
211. Quentin Barbarat fixed an example in the documentation.
212. XyFreak fixed a compiler warning.
213. TotalCaesar659 fixed links in the README.
214. Tanuj Garg improved the fuzzer coverage for UBSAN input.
215. AODQ fixed a compiler warning.
216. jwittbrodt madeNLOHMANN_DEFINE_TYPE_NON_INTRUSIVE inline.
217. pfeatherstone improved the upper bound of arguments of theNLOHMANN_DEFINE_TYPE_NON_INTRUSIVE/NLOHMANN_DEFINE_TYPE_INTRUSIVE macros.
218. Jan Procházka fixed a bug in the CBOR parser for binary and string values.
219. T0b1-iOS fixed a bug in the new hash implementation.
220. Matthew Bauer adjusted the CBOR writer to create tags for binary subtypes.
221. gatopeich implemented an ordered map container fornlohmann::ordered_json.
222. Érico Nogueira Rolim added support for pkg-config.
223. KonanM proposed an implementation for theNLOHMANN_DEFINE_TYPE_NON_INTRUSIVE/NLOHMANN_DEFINE_TYPE_INTRUSIVE macros.
224. Guillaume Racicot implementedstring_view support and allowed C++20 support.
225. Alex Reinking improved CMake support forFetchContent.
226. Hannes Domani provided a GDB pretty printer.
227. Lars Wirzenius reviewed the README file.
228. Jun Jie fixed a compiler path in the CMake scripts.
229. Ronak Buch fixed typos in the documentation.
230. Alexander Karzhenkov fixed a move constructor and the Travis builds.
231. Leonardo Lima added CPM.Cmake support.
232. Joseph Blackman fixed a warning.
233. Yaroslav updated doctest and implemented unit tests.
234. Martin Stump fixed a bug in the CMake files.
235. Jaakko Moisio fixed a bug in the input adapters.
236. bl-ue fixed some Markdown issues in the README file.
237. William A. Wieselquist fixed an example from the README.
238. abbaswasim fixed an example from the README.
239. Remy Jette fixed a warning.
240. Fraser fixed the documentation.
241. Ben Beasley updated doctest.
242. Doron Behar fixed pkg-config.pc.
243. raduteo fixed a warning.
244. David Pfahler added the possibility to compile the library without I/O support.
245. Morten Fyhn Amundsen fixed a typo.
246. jpl-mac allowed to treat the library as a system header in CMake.
247. Jason Dsouza fixed the indentation of the CMake file.
248. offa added a link to Conan Center to the documentation.
249. TotalCaesar659 updated the links in the documentation to use HTTPS.
250. Rafail Giavrimis fixed the Google Benchmark default branch.
251. Louis Dionne fixed a conversion operator.
252. justanotheranonymoususer made the examples in the README more consistent.
253. Finkman suppressed some-Wfloat-equal warnings.
254. Ferry Huberts fixed-Wswitch-enum warnings.
255. Arseniy Terekhin made the GDB pretty-printer robust against unset variable names.
256. Amir Masoud Abdol updated the Homebrew command as nlohmann/json is now in homebrew-core.
257. Hallot fixed some-Wextra-semi-stmt warnings.
258. Giovanni Cerretani fixed-Wunused warnings onJSON_DIAGNOSTICS.
259. Bogdan Popescu hosts thedocset for offline documentation viewers.
260. Carl Smedstad fixed an assertion error when usingJSON_DIAGNOSTICS.
261. miikka75 provided an important fix to compile C++17 code with Clang 9.
262. Maarten Becker fixed a warning for shadowed variables.
263. Cristi Vîjdea fixed typos in theoperator[] documentation.
264. Alex Beregszaszi fixed spelling mistakes in comments.
265. Dirk Stolle fixed typos in documentation.
266. Daniel Albuschat corrected the parameter name in theparse documentation.
267. Prince Mendiratta fixed a link to the FAQ.
268. Florian Albrechtskirchinger implementedstd::string_view support for object keys and made dozens of other improvements.
269. Qianqian Fang implemented the Binary JData (BJData) format.
270. pketelsen added macrosNLOHMANN_DEFINE_TYPE_INTRUSIVE_WITH_DEFAULT andNLOHMANN_DEFINE_TYPE_NON_INTRUSIVE_WITH_DEFAULT.
271. DarkZeros adjusted to code to not clash with Arduino defines.
272. flagarde fixed the output ofmeta() for MSVC.
273. Giovanni Cerretani fixed a check forstd::filesystem.
274. Dimitris Apostolou fixed a typo.
275. Ferry Huberts fixed a typo.
276. Michael Nosthoff fixed a typo.
277. JungHoon Lee fixed a typo.
278. Faruk D. fixed the CITATION.CFF file.
279. Andrea Cocito added a clarification on macro usage to the documentation.
280. Krzysiek Karbowiak refactored the tests to useCHECK_THROWS_WITH_AS.
281. Chaoqi Zhang fixed a typo.
282. ivanovmp fixed a whitespace error.
283. KsaNL fixed a build error when including<windows.h>.
284. Andrea Pappacoda moved.pc and.cmake files toshare directory.
285. Wolf Vollprecht added thepatch_inplace function.
286. Jake Zimmerman highlighted common usage patterns in the README file.
287. NN added the Visual Studio output directory to.gitignore.
288. Romain Reignier improved the performance the vector output adapter.
289. Mike fixed thestd::iterator_traits.
290. Richard Hozák added macroJSON_NO_ENUM to disable default enum conversions.
291. vakokako fixed tests when compiling with C++20.
292. Alexander “weej” Jones fixed an example in the README.
293. Eli Schwartz added more files to theinclude.zip archive.
294. Kevin Lu fixed a compilation issue when typedefs with certain names were present.
295. Trevor Hickey improved the description of an example.
296. Jef LeCompte updated the year in the README file.
297. Alexandre Hamez fixed a warning.
298. Maninderpal Badhan fixed a typo.
299. kevin-- added a note to an example in the README file.
300. I fixed a typo.
301. Gregorio Litenstein fixed the Clang detection.
302. Andreas Smas added a Doozer badge.
303. WanCW fixed the string conversion with Clang.
304. zhaohuaxishi fixed a Doxygen error.
305. emvivre removed an invalid parameter from CMake.
306. Tobias Hermann fixed a link in the README file.
307. Michael fixed a warning.
308. Ryan Mulder addedensure_ascii to thedump function.
309. Muri Nicanor fixed thesed discovery in the Makefile.
310. David Avedissian implemented SFINAE-friendlyiterator_traits.
311. AQNOUCH Mohammed fixed a typo in the README.
312. Gareth Sylvester-Bradley addedoperator/= andoperator/ to construct JSON pointers.
313. Michael Macnair added support for afl-fuzz testing.
314. Berkus Decker fixed a typo in the README.
315. Illia Polishchuk improved the CMake testing.
316. Ikko Ashimine fixed a typo.

Thanks a lot for helping out! Pleaselet me know if I forgot someone.

The library itself consists of a single header file licensed under the MIT license. However, it is built, tested, documented, and whatnot using a lot of third-party tools and services. Thanks a lot!

• amalgamate.py - Amalgamate C source and header files to create a single header file
• American fuzzy lop for fuzz testing
• AppVeyor forcontinuous integration on Windows
• Artistic Style for automatic source code indentation
• Clang for compilation with code sanitizers
• CMake for build automation
• Codacy for furthercode analysis
• Coveralls to measurecode coverage
• Coverity Scan forstatic analysis
• cppcheck for static analysis
• doctest for the unit tests
• git-update-ghpages to upload the documentation to gh-pages
• GitHub Changelog Generator to generate theChangeLog
• Google Benchmark to implement the benchmarks
• Hedley to avoid re-inventing several compiler-agnostic feature macros
• lcov to process coverage information and create an HTML view
• libFuzzer to implement fuzz testing for OSS-Fuzz
• Material for MkDocs for the style of the documentation site
• MkDocs for the documentation site
• OSS-Fuzz for continuous fuzz testing of the library (project repository)
• Probot for automating maintainer tasks such as closing stale issues, requesting missing information, or detecting toxic comments.
• Valgrind to check for correct memory management

The library is currently used in Apple macOS Sierra-Monterey and iOS 10-15. I am not sure what they are using the library for, but I am happy that it runs on so many devices.

The library supportsUnicode input as follows:

• OnlyUTF-8 encoded input is supported which is the default encoding for JSON according toRFC 8259.
• std::u16string andstd::u32string can be parsed, assuming UTF-16 and UTF-32 encoding, respectively. These encodings are not supported when reading from files or other input containers.
• Other encodings such as Latin-1 or ISO 8859-1 arenot supported and will yield parse or serialization errors.
• Unicode noncharacters will not be replaced by the library.
• Invalid surrogates (e.g., incomplete pairs such as\uDEAD) will yield parse errors.
• The strings stored in the library are UTF-8 encoded. When using the default string type (std::string), note that its length/size functions return the number of stored bytes rather than the number of characters or glyphs.
• When you store strings with different encodings in the library, callingdump() may throw an exception unlessjson::error_handler_t::replace orjson::error_handler_t::ignore are used as error handlers.
• To store wide strings (e.g.,std::wstring), you need to convert them to a UTF-8 encodedstd::string before, seean example.

This library does not support comments by default. It does so for three reasons:

1. Comments are not part of theJSON specification. You may argue that// or/* */ are allowed in JavaScript, but JSON is not JavaScript.

2. This was not an oversight: Douglas Crockfordwrote on this in May 2012:

   I removed comments from JSON because I saw people were using them to hold parsing directives, a practice which would have destroyed interoperability. I know that the lack of comments makes some people sad, but it shouldn't.

   Suppose you are using JSON to keep configuration files, which you would like to annotate. Go ahead and insert all the comments you like. Then pipe it through JSMin before handing it to your JSON parser.

3. It is dangerous for interoperability if some libraries would add comment support while others don't. Please checkThe Harmful Consequences of the Robustness Principle on this.

However, you can pass set parameterignore_comments to true in theparse function to ignore// or/* */ comments. Comments will then be treated as whitespace.

By default, the library does not preserve theinsertion order of object elements. This is standards-compliant, as theJSON standard defines objects as "an unordered collection of zero or more name/value pairs".

If you do want to preserve the insertion order, you can try the typenlohmann::ordered_json. Alternatively, you can use a more sophisticated ordered map liketsl::ordered_map (integration) ornlohmann::fifo_map (integration).

We checked with Valgrind and the Address Sanitizer (ASAN) that there are no memory leaks.

If you find that a parsing program with this library does not release memory, please consider the following case, and it may be unrelated to this library.

Your program is compiled with glibc. There is a tunable threshold that glibc uses to decide whether to actually return memory to the system or whether to cache it for later reuse. If in your program you make lots of small allocations and those small allocations are not a contiguous block and are presumably below the threshold, then they will not get returned to the OS.Here is a related issue#1924.

• The code contains numerous debugassertions which can be switched off by defining the preprocessor macroNDEBUG, see thedocumentation ofassert. In particular, noteoperator[] implementsunchecked access for const objects: If the given key is not present, the behavior is undefined (think of a dereferenced null pointer) and yields anassertion failure if assertions are switched on. If you are not sure whether an element in an object exists, use checked access with theat() function. Furthermore, you can defineJSON_ASSERT(x) to replace calls toassert(x).
• As the exact number type is not defined in theJSON specification, this library tries to choose the best fitting C++ number type automatically. As a result, the typedouble may be used to store numbers which may yieldfloating-point exceptions in certain rare situations if floating-point exceptions have been unmasked in the calling code. These exceptions are not caused by the library and need to be fixed in the calling code, such as by re-masking the exceptions prior to calling library functions.
• The code can be compiled without C++runtime type identification features; that is, you can use the-fno-rtti compiler flag.
• Exceptions are used widely within the library. They can, however, be switched off with either using the compiler flag-fno-exceptions or by defining the symbolJSON_NOEXCEPTION. In this case, exceptions are replaced byabort() calls. You can further control this behavior by definingJSON_THROW_USER (overridingthrow),JSON_TRY_USER (overridingtry), andJSON_CATCH_USER (overridingcatch). Note thatJSON_THROW_USER should leave the current scope (e.g., by throwing or aborting), as continuing after it may yield undefined behavior. Note the explanatorywhat() string of exceptions is not available for MSVC if exceptions are disabled, see#2824.

To compile and run the tests, you need to execute

$ mkdir build$cd build$ cmake .. -DJSON_BuildTests=On$ cmake --build.$ ctest --output-on-failure

Note that during thectest stage, several JSON test files are downloaded from anexternal repository. If policies forbid downloading artifacts during testing, you can download the files yourself and pass the directory with the test files via-DJSON_TestDataDirectory=path to CMake. Then, no Internet connectivity is required. Seeissue #2189 for more information.

If the test suite is not found, several test suites will fail like this:

===============================================================================json/tests/src/make_test_data_available.hpp:21:TEST CASE:  check test suite is downloadedjson/tests/src/make_test_data_available.hpp:23: FATAL ERROR: REQUIRE( utils::check_testsuite_downloaded() ) is NOT correct!  values: REQUIRE( false )  logged: Test data not found in 'json/cmake-build-debug/json_test_data'.          Please execute target 'download_test_data' before running this test suite.          See <https://github.com/nlohmann/json#execute-unit-tests> for more information.===============================================================================

In case you have downloaded the library rather than checked out the code via Git, testcmake_fetch_content_configure will fail. Please executectest -LE git_required to skip these tests. Seeissue #2189 for more information.

Some tests change the installed files and hence make the whole process not reproducible. Please executectest -LE not_reproducible to skip these tests. Seeissue #2324 for more information.

Note you need to callcmake -LE "not_reproducible|git_required" to exclude both labels. Seeissue #2596 for more information.

As Intel compilers use unsafe floating point optimization by default, the unit tests may fail. Use flag/fp:precise then.