jsoncons

C++ library for parsing and serializing JSON text

View the Project on GitHub danielaparker/jsoncons

jsoncons: a C++ library for json construction

Preliminaries

Reading JSON text from a file

Constructing json values in C++

Converting to and from standard library containers

Converting CSV files to json

Pretty print

Filters

JSONPath

About jsoncons::json

Wide character support

ojson and wojson

Convert json to/from user defined type

Preliminaries

jsoncons is a C++, header-only library for constructing JSON and JSON-like data formats such as CBOR. It supports

  • Parsing JSON-like text or binary data into an unpacked representation of variant type that defines an interface for accessing and modifying that data (covers bignum and byte string values.)

  • Serializing the unpacked representation into different JSON-like text or binary data.

  • Converting from JSON-like text or binary data to C++ objects and back via json_type_traits.

  • Streaming JSON read and write events, somewhat analogously to SAX (push parsing) and StAX (pull parsing) in the XML world.

Compared to other JSON libraries, jsoncons has been designed to handle very large JSON texts. At its heart are SAX style parsers and serializers. Its json parser is an incremental parser that can be fed its input in chunks, and does not require an entire file to be loaded in memory at one time. Its unpacked in-memory representation of JSON is more compact than most, and can be made more compact still with a user-supplied allocator. It also supports memory efficient parsing of very large JSON texts with a pull parser, built on top of its incremental parser.

The jsoncons library is header-only: it consists solely of header files containing templates and inline functions, and requires no separately-compiled library binaries when linking. It has no dependence on other libraries.

To install the librray, download the latest release and unpack the zip file. Copy the directory include/jsoncons to your include directory. If you wish to use extensions, copy include/jsoncons_ext as well.

Or, download the latest code on master.

The jsoncons classes and functions are in namespace jsoncons. You need to include the header file

#include <jsoncons/json.hpp>

and, for convenience,

using jsoncons::json;

Reading JSON text from a file

Example file (books.json):

[
    {
        "title" : "Kafka on the Shore",
        "author" : "Haruki Murakami",
        "price" : 25.17
    },
    {
        "title" : "Women: A Novel",
        "author" : "Charles Bukowski",
        "price" : 12.0
    },
    {
        "title" : "Cutter's Way",
        "author" : "Ivan Passer"
    }
]

It consists of an array of book elements, each element is an object with members title, author, and price.

Read the JSON text into a json value,

std::ifstream is("books.json");
json books = json::parse(is);

Loop through the book array elements, using a range-based for loop

for (const auto& book : books.array_range())
{
    std::string author = book["author"].as<std::string>();
    std::string title = book["title"].as<std::string>();
    std::cout << author << ", " << title << std::endl;
}

or begin-end iterators

for (auto it = books.array_range().begin(); 
     it != books.array_range().end();
     ++it)
{
    std::string author = (*it)["author"].as<std::string>();
    std::string title = (*it)["title"].as<std::string>();
    std::cout << author << ", " << title << std::endl;
} 

or a traditional for loop

for (size_t i = 0; i < books.size(); ++i)
{
    json& book = books[i];
    std::string author = book["author"].as<std::string>();
    std::string title = book["title"].as<std::string>();
    std::cout << author << ", " << title << std::endl;
}

Output:

Haruki Murakami, Kafka on the Shore
Charles Bukowski, Women: A Novel
Ivan Passer, Cutter's Way

Loop through the members of the third book element, using a range-based for loop

for (const auto& member : books[2].object_range())
{
    std::cout << member.key() << "=" 
              << member.value() << std::endl;
}

or begin-end iterators:

for (auto it = books[2].object_range().begin(); 
     it != books[2].object_range().end();
     ++it)
{
    std::cout << (*it).key() << "=" 
              << (*it).value() << std::endl;
} 

Output:

author=Ivan Passer
title=Cutter's Way

Note that the third book, Cutter’s Way, is missing a price.

You have a choice of object member accessors:

  • book["price"] will throw std::out_of_range if there is no price
  • book.get_with_default("price",std::string("n/a")) will return the price converted to the default’s data type, std::string, or "n/a" if there is no price.

So if you want to show “n/a” for the missing price, you can use this accessor

std::string price = book.get_with_default("price","n/a");

Or you can check if book has a member “price” with the method contains, and output accordingly,

if (book.contains("price"))
{
    double price = book["price"].as<double>();
    std::cout << price;
}
else
{
    std::cout << "n/a";
}

Constructing json values in C++

The default json constructor produces an empty json object. For example

json image_sizing;
std::cout << image_sizing << std::endl;

produces

{}

To construct a json object with members, take an empty json object and set some name-value pairs

image_sizing.insert_or_assign("Resize To Fit",true);  // a boolean 
image_sizing.insert_or_assign("Resize Unit", "pixels");  // a string
image_sizing.insert_or_assign("Resize What", "long_edge");  // a string
image_sizing.insert_or_assign("Dimension 1",9.84);  // a double
image_sizing.insert_or_assign("Dimension 2",json::null());  // a null value

Or, use an object initializer-list:

json file_settings = json::object{
    {"Image Format", "JPEG"},
    {"Color Space", "sRGB"},
    {"Limit File Size", true},
    {"Limit File Size To", 10000}
};

To construct a json array, initialize with the array type

json color_spaces = json::array();

and add some elements

color_spaces.push_back("sRGB");
color_spaces.push_back("AdobeRGB");
color_spaces.push_back("ProPhoto RGB");

Or, use an array initializer-list:

json image_formats = json::array{"JPEG","PSD","TIFF","DNG"};

The operator[] provides another way for setting name-value pairs.

json file_export;
file_export["File Format Options"]["Color Spaces"] = 
    std::move(color_spaces);
file_export["File Format Options"]["Image Formats"] = 
    std::move(image_formats);
file_export["File Settings"] = std::move(file_settings);
file_export["Image Sizing"] = std::move(image_sizing);

Note that if file_export["File Format Options"] doesn’t exist,

file_export["File Format Options"]["Color Spaces"] = 
    std::move(color_spaces)

creates "File Format Options" as an object and puts "Color Spaces" in it.

Serializing

std::cout << pretty_print(file_export) << std::endl;

produces

{
    "File Format Options": {
        "Color Spaces": ["sRGB","AdobeRGB","ProPhoto RGB"],
        "Image Formats": ["JPEG","PSD","TIFF","DNG"]
    },
    "File Settings": {
        "Color Space": "sRGB",
        "Image Format": "JPEG",
        "Limit File Size": true,
        "Limit File Size To": 10000
    },
    "Image Sizing": {
        "Dimension 1": 9.84,
        "Dimension 2": null,
        "Resize To Fit": true,
        "Resize Unit": "pixels",
        "Resize What": "long_edge"
    }
}

Converting to and from standard library containers

The jsoncons library supports converting to and from the standard library sequence and associative containers.

std::vector<int> v = {1,2,3,4};
json j(v);
std::cout << j << std::endl;

Output:

[1,2,3,4]
json j = json::array{1,true,"last"};
auto d = j.as<std::deque<std::string>>();
for (auto x : d)
{
    std::cout << x << std::endl;
}

Output:

1
true
last

std::map<std::string,int> m = {{"one",1},{"two",2},{"three",3}};
json j(m);
std::cout << j << std::endl;

Output:

{"one":1,"three":3,"two":2}
json j;
j["one"] = 1;
j["two"] = 2;
j["three"] = 3;

auto um = j.as<std::unordered_map<std::string,int>>();
for (const auto& x : um)
{
    std::cout << x.first << "=" << x.second << std::endl;
}

Output:

one=1
three=3
two=2

Converting CSV files to json

Example CSV file (tasks.csv):

project_id, task_name, task_start, task_finish
4001,task1,01/01/2003,01/31/2003
4001,task2,02/01/2003,02/28/2003
4001,task3,03/01/2003,03/31/2003
4002,task1,04/01/2003,04/30/2003
4002,task2,05/01/2003,

You can read the CSV file into a json value with the decode_csv function.

#include <fstream>
#include <jsoncons/json.hpp>
#include <jsoncons_ext/csv/csv_reader.hpp>
#include <jsoncons_ext/csv/csv_serializer.hpp>

using namespace jsoncons;
using namespace jsoncons::csv;

int main()
{
    std::ifstream is("input/tasks.csv");

    csv_options options;
    params.assume_header(true)
          .trim(true)
          .ignore_empty_values(true) 
          .column_types("integer,string,string,string");
    ojson tasks = decode_csv<ojson>(is, options);

    std::cout << "(1)\n" << pretty_print(tasks) << "\n\n";

    std::cout << "(2)\n";
    encode_csv(tasks, std::cout);
}

Output:

(1)
[
    {
        "project_id": 4001,
        "task_name": "task1",
        "task_start": "01/01/2003",
        "task_finish": "01/31/2003"
    },
    {
        "project_id": 4001,
        "task_name": "task2",
        "task_start": "02/01/2003",
        "task_finish": "02/28/2003"
    },
    {
        "project_id": 4001,
        "task_name": "task3",
        "task_start": "03/01/2003",
        "task_finish": "03/31/2003"
    },
    {
        "project_id": 4002,
        "task_name": "task1",
        "task_start": "04/01/2003",
        "task_finish": "04/30/2003"
    },
    {
        "project_id": 4002,
        "task_name": "task2",
        "task_start": "05/01/2003"
    }
]

There are a few things to note about the effect of the parameter settings.

  • assume_header true tells the csv parser to parse the first line of the file for column names, which become object member names.
  • trim true tells the parser to trim leading and trailing whitespace, in particular, to remove the leading whitespace in the column names.
  • ignore_empty_values true causes the empty last value in the task_finish column to be omitted.
  • The column_types setting specifies that column one (“project_id”) contains integers and the remaining columns strings.

Pretty print

The pretty_print function applies stylistic formatting to JSON text. For example

    json val;

    val["verts"] = json::array{1, 2, 3};
    val["normals"] = json::array{1, 0, 1};
    val["uvs"] = json::array{0, 0, 1, 1};

    std::cout << pretty_print(val) << std::endl;

produces

{
    "normals": [1,0,1],
    "uvs": [0,0,1,1],
    "verts": [1,2,3]
}

By default, within objects, arrays of scalar values are displayed on the same line.

The pretty_print function takes an optional second parameter, json_options, that allows custom formatting of output. To display the array scalar values on a new line, set the object_array_line_splits property to line_split_kind::new_line. The code

json_options options;
format.object_array_line_splits(line_split_kind::new_line);
std::cout << pretty_print(val,options) << std::endl;

produces

{
    "normals": [
        1,0,1
    ],
    "uvs": [
        0,0,1,1
    ],
    "verts": [
        1,2,3
    ]
}

To display the elements of array values on multiple lines, set the object_array_line_splits property to line_split_kind::multi_line. The code

json_options options;
format.object_array_line_splits(line_split_kind::multi_line);
std::cout << pretty_print(val,options) << std::endl;

produces

{
    "normals": [
        1,
        0,
        1
    ],
    "uvs": [
        0,
        0,
        1,
        1
    ],
    "verts": [
        1,
        2,
        3
    ]
}

Filters

You can rename object member names with the built in filter rename_object_member_filter

#include <sstream>
#include <jsoncons/json.hpp>
#include <jsoncons/json_filter.hpp>

using namespace jsoncons;

int main()
{
    std::string s = R"({"first":1,"second":2,"fourth":3,"fifth":4})";    

    json_serializer serializer(std::cout);

    // Filters can be chained
    rename_object_member_filter filter2("fifth", "fourth", serializer);
    rename_object_member_filter filter1("fourth", "third", filter2);

    // A filter can be passed to any function that takes
    // a json_content_handler ...
    std::cout << "(1) ";
    std::istringstream is(s);
    json_reader reader(is, filter1);
    reader.read();
    std::cout << std::endl;

    // or a json_content_handler    
    std::cout << "(2) ";
    ojson j = ojson::parse(s);
    j.dump(filter1);
    std::cout << std::endl;
}

Output:

(1) {"first":1,"second":2,"third":3,"fourth":4}
(2) {"first":1,"second":2,"third":3,"fourth":4}

Or define and use your own filters. See json_filter for details.

JSONPath

Stefan Goessner’s JSONPath is an XPATH inspired query language for selecting parts of a JSON structure.

Example JSON file (booklist.json):

{ "store": {
    "book": [ 
      { "category": "reference",
        "author": "Nigel Rees",
        "title": "Sayings of the Century",
        "price": 8.95
      },
      { "category": "fiction",
        "author": "Evelyn Waugh",
        "title": "Sword of Honour",
        "price": 12.99
      },
      { "category": "fiction",
        "author": "Herman Melville",
        "title": "Moby Dick",
        "isbn": "0-553-21311-3",
        "price": 8.99
      },
      { "category": "fiction",
        "author": "J. R. R. Tolkien",
        "title": "The Lord of the Rings",
        "isbn": "0-395-19395-8",
        "price": 22.99
      }
    ]
  }
}

JSONPath examples:

#include <jsoncons_ext/jsonpath/json_query.hpp>

using jsoncons::jsonpath::json_query;

std::ifstream is("./input/booklist.json");
json booklist = json::parse(is);

// The authors of books that are cheaper than $10
json result1 = json_query(booklist, "$.store.book[?(@.price < 10)].author");
std::cout << "(1) " << result1 << std::endl;

// The number of books
json result2 = json_query(booklist, "$..book.length");
std::cout << "(2) " << result2 << std::endl;

// The third book
json result3 = json_query(booklist, "$..book[2]");
std::cout << "(3)\n" << pretty_print(result3) << std::endl;

// All books whose author's name starts with Evelyn
json result4 = json_query(booklist, "$.store.book[?(@.author =~ /Evelyn.*?/)]");
std::cout << "(4)\n" << pretty_print(result4) << std::endl;

// The titles of all books that have isbn number
json result5 = json_query(booklist, "$..book[?(@.isbn)].title");
std::cout << "(5) " << result5 << std::endl;

// All authors and titles of books
json result6 = json_query(booklist, "$['store']['book']..['author','title']");
std::cout << "(6)\n" << pretty_print(result6) << std::endl;

Output:

(1) ["Nigel Rees","Herman Melville"]
(2) [4]
(3)
[
    {
        "author": "Herman Melville",
        "category": "fiction",
        "isbn": "0-553-21311-3",
        "price": 8.99,
        "title": "Moby Dick"
    }
]
(4)
[
    {
        "author": "Evelyn Waugh",
        "category": "fiction",
        "price": 12.99,
        "title": "Sword of Honour"
    }
]
(5) ["Moby Dick","The Lord of the Rings"]
(6)
[
    "Nigel Rees",
    "Sayings of the Century",
    "Evelyn Waugh",
    "Sword of Honour",
    "Herman Melville",
    "Moby Dick",
    "J. R. R. Tolkien",
    "The Lord of the Rings"
]

About jsoncons::json

The json class is an instantiation of the basic_json class template that uses char as the character type and sorts object members in alphabetically order.

typedef basic_json<char,
                   ImplementationPolicy = sorted_policy,
                   Allocator = std::allocator<char>> json;

If you prefer to retain the original insertion order, use ojson instead.

The library includes an instantiation for wide characters as well, wjson

typedef basic_json<wchar_t,
                   ImplementationPolicy = sorted_policy,
                   Allocator = std::allocator<wchar_t>> wjson;

If you prefer to retain the original insertion order, use wojson instead.

Note that the allocator type allows you to supply a custom allocator. For example, you can use the boost fast_pool_allocator:

#include <boost/pool/pool_alloc.hpp>
#include <jsoncons/json.hpp>

typedef jsoncons::basic_json<char, boost::fast_pool_allocator<char>> myjson;

myjson o;

o.insert_or_assign("FirstName","Joe");
o.insert_or_assign("LastName","Smith");

This results in a json value being constucted with all memory being allocated from the boost memory pool. (In this particular case there is no improvement in performance over std::allocator.)

Note that the underlying memory pool used by the boost::fast_pool_allocator is never freed.

Wide character support

jsoncons supports wide character strings and streams with wjson and wjson_reader. It supports UTF16 encoding if wchar_t has size 2 (Windows) and UTF32 encoding if wchar_t has size 4. You can construct a wjson value in exactly the same way as a json value, for instance:

using jsoncons::wjson;

wjson root;
root[L"field1"] = L"test";
root[L"field2"] = 3.9;
root[L"field3"] = true;

std::wcout << root << L"\n";

which prints

{"field1":"test","field2":3.9,"field3":true}

ojson and wojson

The ojson (wojson) class is an instantiation of the basic_json class template that uses char (wchar_t) as the character type and keeps object members in their original order.

ojson o = ojson::parse(R"(
{
    "street_number" : "100",
    "street_name" : "Queen St W",
    "city" : "Toronto",
    "country" : "Canada"
}
)");

std::cout << pretty_print(o) << std::endl;

Output:

{
    "street_number": "100",
    "street_name": "Queen St W",
    "city": "Toronto",
    "country": "Canada"
}

Insert “postal_code” at end

o.insert_or_assign("postal_code", "M5H 2N2");

std::cout << pretty_print(o) << std::endl;

Output:

{
    "street_number": "100",
    "street_name": "Queen St W",
    "city": "Toronto",
    "country": "Canada",
    "postal_code": "M5H 2N2"
}

Insert “province” before “country”

auto it = o.find("country");
o.insert_or_assign(it,"province","Ontario");

std::cout << pretty_print(o) << std::endl;

Output:

{
    "street_number": "100",
    "street_name": "Queen St W",
    "city": "Toronto",
    "province": "Ontario",
    "country": "Canada",
    "postal_code": "M5H 2N2"
}

Convert unpacked json value to user defined type and back

In the json class, constructors, accessors and modifiers are templated, for example,

template <class T>
json(const T& val)

template<class T, class... Args>
bool is(Args&&... args) const

template<class T, class... Args>
T as(Args&&... args) const

template <class T>
basic_json& operator=(const T& val)

template <class T>
void push_back(T&& val)

template <class T>
void insert_or_assign(const string_view_type& name, T&& val)

The implementations of these functions and operators make use of the class template json_type_traits

If you want to use the json constructor, is<T>, as<T>, operator=, push_back, insert, and insert_or_assign to access or modify with a new type, you need to show json how to interact with that type, by extending json_type_traits in the jsoncons namespace.

Note that the json::is() and json::as() functions accept template packs, which they forward to the `json_type_traits` `is` and `as` functions. This allows user defined `json_type_traits` implementations to resolve, for instance, a name into a C++ object looked up from a registry, as illustrated below.

#include <string>
#include <unordered_map>
#include <memory>
#include <jsoncons/json.hpp>

using namespace jsoncons;
using namespace jsoncons::literals;

class Employee
{
    std::string name_;
public:
    Employee(const std::string& name)
        : name_(name)
    {
    }
    virtual ~Employee() = default;
    const std::string& name() const
    {
        return name_;
    }
    virtual double calculatePay() const = 0;
};

class HourlyEmployee : public Employee
{
public:
    HourlyEmployee(const std::string& name)
        : Employee(name)
    {
    }
    double calculatePay() const override
    {
        return 10000;
    }
};

class CommissionedEmployee : public Employee
{
public:
    CommissionedEmployee(const std::string& name)
        : Employee(name)
    {
    }
    double calculatePay() const override
    {
        return 20000;
    }
};

class EmployeeRegistry
{
    typedef std::unordered_map<std::string,std::shared_ptr<Employee>> employee_map;
    employee_map employees_;
public:
    EmployeeRegistry()
    {
        employees_.try_emplace("John Smith",std::make_shared<HourlyEmployee>("John Smith"));
        employees_.try_emplace("Jane Doe",std::make_shared<CommissionedEmployee>("Jane Doe"));
    }

    bool contains(const std::string& name) const
    {
        return employees_.count(name) > 0; 
    }

    std::shared_ptr<Employee> get(const std::string& name) const
    {
        auto it = employees_.find(name);
        if (it == employees_.end())
        {
            throw std::runtime_error("Employee not found");
        }
        return it->second; 
    }
};

namespace jsoncons
{
    template<class Json>
    struct json_type_traits<Json, std::shared_ptr<Employee>>
    {
        static bool is(const Json& rhs, const EmployeeRegistry& registry) noexcept
        {
            return rhs.is_string() && registry.contains(rhs.as<std::string>());
        }
        static std::shared_ptr<Employee> as(const Json& rhs, 
                                            const EmployeeRegistry& registry)
        {
            return registry.get(rhs.as<std::string>());
        }
        static Json to_json(std::shared_ptr<Employee> val)
        {
            Json j(val->name());
            return j;
        }
    };
};

int main()
{
    json j = R"(
    {
        "EmployeeName" : "John Smith"
    }
    )"_json;

    EmployeeRegistry registry;

    std::shared_ptr<Employee> employee = j["EmployeeName"].as<std::shared_ptr<Employee>>(registry);

    std::cout << "(1) " << employee->name() << " => " 
              << employee->calculatePay() << std::endl;

    // j does not have a key "SalesRep", so get_with_default returns "Jane Doe"
    // The template parameter is explicitly specified as json, to return a json string
    // json::as is then applied to the returned json string  
    std::shared_ptr<Employee> salesRep = j.get_with_default<json>("SalesRep","Jane Doe")
                                          .as<std::shared_ptr<Employee>>(registry);

    std::cout << "(2) " << salesRep->name() << " => " 
              << salesRep->calculatePay() << std::endl;

    json j2;
    j2["EmployeeName"] = employee;
    j2["SalesRep"] = salesRep;

    std::cout << "(3)\n" << pretty_print(j2) << std::endl;
}

Output:

(1) John Smith => 10000
(2) Jane Doe => 20000
(3)
{
    "EmployeeName": "John Smith",
    "SalesRep": "Jane Doe"
}

For more information, consult the latest documentation and roadmap.