/*! @file @copyright The code is licensed under the MIT License , Copyright (c) 2013-2015 Niels Lohmann. @author Niels Lohmann @see https://github.com/nlohmann/json */ #ifndef _NLOHMANN_JSON #define _NLOHMANN_JSON #include #include #include #include #include #include #include #include #include #include #include #include #include /* - ObjectType trick from http://stackoverflow.com/a/9860911 */ /*! @see https://github.com/nlohmann */ namespace nlohmann { /*! @brief JSON @tparam ObjectType type for JSON objects (@c std::map by default) @tparam ArrayType type for JSON arrays (@c std::vector by default) @tparam StringType type for JSON strings and object keys (@c std::string by default) @tparam BooleanType type for JSON booleans (@c bool by default) @tparam NumberIntegerType type for JSON integer numbers (@c int64_t by default) @tparam NumberFloatType type for JSON floating-point numbers (@c double by default) @see RFC 7159 @see ECMA 404 */ template < template class ObjectType = std::map, template class ArrayType = std::vector, class StringType = std::string, class BooleanType = bool, class NumberIntegerType = int64_t, class NumberFloatType = double > class basic_json { public: ///////////////////// // container types // ///////////////////// // forward declarations class iterator; class const_iterator; /// the type of elements in a basic_json container using value_type = basic_json; /// the type of an element reference using reference = basic_json&; /// the type of an element const reference using const_reference = const basic_json&; /// the type of an element pointer using pointer = basic_json*; /// the type of an element const pointer using const_pointer = const basic_json*; /// a type to represent differences between iterators using difference_type = std::ptrdiff_t; /// a type to represent container sizes using size_type = std::size_t; /// an iterator for a basic_json container using iterator = basic_json::iterator; /// a const iterator for a basic_json container using const_iterator = basic_json::const_iterator; /// a reverse iterator for a basic_json container using reverse_iterator = std::reverse_iterator; /// a const reverse iterator for a basic_json container using const_reverse_iterator = std::reverse_iterator; /////////////////////////// // JSON value data types // /////////////////////////// /// a type for an object using object_t = ObjectType; /// a type for an array using array_t = ArrayType; /// a type for a string using string_t = StringType; /// a type for a boolean using boolean_t = BooleanType; /// a type for a number (integer) using number_integer_t = NumberIntegerType; /// a type for a number (floating-point) using number_float_t = NumberFloatType; /// a type for list initialization using list_init_t = std::initializer_list; //////////////////////// // JSON value storage // //////////////////////// /// a JSON value union json_value { /// object (stored with pointer to save storage) object_t* object; /// array (stored with pointer to save storage) array_t* array; /// string (stored with pointer to save storage) string_t* string; /// bolean boolean_t boolean; /// number (integer) number_integer_t number_integer; /// number (floating-point) number_float_t number_float; /// default constructor (for null values) json_value() = default; /// constructor for objects json_value(object_t* v) : object(v) {} /// constructor for arrays json_value(array_t* v) : array(v) {} /// constructor for strings json_value(string_t* v) : string(v) {} /// constructor for booleans json_value(boolean_t v) : boolean(v) {} /// constructor for numbers (integer) json_value(number_integer_t v) : number_integer(v) {} /// constructor for numbers (floating-point) json_value(number_float_t v) : number_float(v) {} }; ///////////////////////////////// // JSON value type enumeration // ///////////////////////////////// /// JSON value type enumeration enum class value_t : uint8_t { null, ///< null value object, ///< object (unordered set of name/value pairs) array, ///< array (ordered collection of values) string, ///< string value boolean, ///< boolean value number_integer, ///< number value (integer) number_float ///< number value (floating-point) }; ////////////////// // constructors // ////////////////// /// create an empty value with a given type inline basic_json(const value_t value) : m_type(value) { switch (m_type) { case (value_t::null): { break; } case (value_t::object): { m_value.object = new object_t(); break; } case (value_t::array): { m_value.array = new array_t(); break; } case (value_t::string): { m_value.string = new string_t(""); break; } case (value_t::boolean): { m_value.boolean = boolean_t(false); break; } case (value_t::number_integer): { m_value.number_integer = number_integer_t(0); break; } case (value_t::number_float): { m_value.number_float = number_float_t(0.0); break; } } } /// create a null object (implicitly) inline basic_json() noexcept : m_type(value_t::null) {} /// create a null object (explicitly) inline basic_json(std::nullptr_t) noexcept : m_type(value_t::null) {} /// create an object (explicit) inline basic_json(const object_t& value) : m_type(value_t::object), m_value(new object_t(value)) {} /// create an object (implicit) template ::value and std::is_constructible::value, int>::type = 0> inline basic_json(const V& value) : m_type(value_t::object), m_value(new object_t(value.begin(), value.end())) {} /// create an array (explicit) inline basic_json(const array_t& value) : m_type(value_t::array), m_value(new array_t(value)) {} /// create an array (implicit) template ::value and not std::is_same::value and not std::is_same::value and not std::is_same::value and not std::is_same::value and not std::is_same::value and std::is_constructible::value, int>::type = 0> inline basic_json(const V& value) : m_type(value_t::array), m_value(new array_t(value.begin(), value.end())) {} /// create a string (explicit) inline basic_json(const string_t& value) : m_type(value_t::string), m_value(new string_t(value)) {} /// create a string (explicit) inline basic_json(const typename string_t::value_type* value) : m_type(value_t::string), m_value(new string_t(value)) {} /// create a string (implicit) template ::value, int>::type = 0> inline basic_json(const V& value) : basic_json(string_t(value)) {} /// create a boolean (explicit) inline basic_json(boolean_t value) : m_type(value_t::boolean), m_value(value) {} /// create an integer number (explicit) inline basic_json(const number_integer_t& value) : m_type(value_t::number_integer), m_value(value) {} /// create an integer number (implicit) template::value and std::numeric_limits::is_integer, T>::type = 0> inline basic_json(const T value) noexcept : m_type(value_t::number_integer), m_value(number_integer_t(value)) {} /// create a floating-point number (explicit) inline basic_json(const number_float_t& value) : m_type(value_t::number_float), m_value(value) {} /// create a floating-point number (implicit) template::value and std::is_floating_point::value>::type > inline basic_json(const T value) noexcept : m_type(value_t::number_float), m_value(number_float_t(value)) {} /// create a container (array or object) from an initializer list inline basic_json(list_init_t l, bool type_deduction = true, value_t manual_type = value_t::array) { // the initializer list could describe an object bool is_object = true; // check if each element is an array with two elements whose first element // is a string for (const auto& element : l) { if ((element.m_final and element.m_type == value_t::array) or (element.m_type != value_t::array or element.size() != 2 or element[0].m_type != value_t::string)) { // we found an element that makes it impossible to use the // initializer list as object is_object = false; break; } } // adjust type if type deduction is not wanted if (not type_deduction) { // mark this object's type as final m_final = true; // if array is wanted, do not create an object though possible if (manual_type == value_t::array) { is_object = false; } // if object is wanted but impossible, throw an exception if (manual_type == value_t::object and not is_object) { throw std::logic_error("cannot create JSON object from initializer list"); } } if (is_object) { // the initializer list is a list of pairs -> create object m_type = value_t::object; m_value = new object_t(); for (auto& element : l) { m_value.object->emplace(std::move(*(element[0].m_value.string)), std::move(element[1])); } } else { // the initializer list describes an array -> create array m_type = value_t::array; m_value = new array_t(std::move(l)); } } /// explicitly create an array from an initializer list inline static basic_json array(list_init_t l = list_init_t()) { return basic_json(l, false, value_t::array); } /// explicitly create an object from an initializer list inline static basic_json object(list_init_t l = list_init_t()) { return basic_json(l, false, value_t::object); } /////////////////////////////////////// // other constructors and destructor // /////////////////////////////////////// /// copy constructor inline basic_json(const basic_json& other) : m_type(other.m_type) { switch (m_type) { case (value_t::null): { break; } case (value_t::object): { m_value.object = new object_t(*other.m_value.object); break; } case (value_t::array): { m_value.array = new array_t(*other.m_value.array); break; } case (value_t::string): { m_value.string = new string_t(*other.m_value.string); break; } case (value_t::boolean): { m_value.boolean = other.m_value.boolean; break; } case (value_t::number_integer): { m_value.number_integer = other.m_value.number_integer; break; } case (value_t::number_float): { m_value.number_float = other.m_value.number_float; break; } } } /// move constructor inline basic_json(basic_json&& other) noexcept : m_type(std::move(other.m_type)), m_value(std::move(other.m_value)) { // invalidate payload other.m_type = value_t::null; other.m_value = {}; } /// copy assignment inline reference& operator=(basic_json other) noexcept ( std::is_nothrow_move_constructible::value and std::is_nothrow_move_assignable::value and std::is_nothrow_move_constructible::value and std::is_nothrow_move_assignable::value ) { std::swap(m_type, other.m_type); std::swap(m_value, other.m_value); return *this; } /// destructor inline ~basic_json() noexcept { switch (m_type) { case (value_t::object): { delete m_value.object; m_value.object = nullptr; break; } case (value_t::array): { delete m_value.array; m_value.array = nullptr; break; } case (value_t::string): { delete m_value.string; m_value.string = nullptr; break; } default: { // all other types need no specific destructor break; } } } public: /////////////////////// // object inspection // /////////////////////// /*! @brief serialization Serialization function for JSON objects. The function tries to mimick Python's @p json.dumps() function, and currently supports its @p indent parameter. @param indent if indent is nonnegative, then array elements and object members will be pretty-printed with that indent level. An indent level of 0 will only insert newlines. -1 (the default) selects the most compact representation @see https://docs.python.org/2/library/json.html#json.dump */ inline string_t dump(const int indent = -1) const noexcept { if (indent >= 0) { return dump(true, static_cast(indent)); } else { return dump(false, 0); } } /// return the type of the object (explicit) inline value_t type() const noexcept { return m_type; } /// return the type of the object (implicit) operator value_t() const noexcept { return m_type; } ////////////////////// // value conversion // ////////////////////// /// get an object (explicit) template ::value and std::is_constructible::value, int>::type = 0> inline T get() const { switch (m_type) { case (value_t::object): return T(m_value.object->begin(), m_value.object->end()); default: throw std::logic_error("cannot cast " + type_name() + " to " + typeid(T).name()); } } /// get an array (explicit) template ::value and std::is_constructible::value, int>::type = 0> inline T get() const { switch (m_type) { case (value_t::array): return T(m_value.array->begin(), m_value.array->end()); default: throw std::logic_error("cannot cast " + type_name() + " to " + typeid(T).name()); } } /// get a string (explicit) template ::value, int>::type = 0> inline T get() const { switch (m_type) { case (value_t::string): return *m_value.string; default: throw std::logic_error("cannot cast " + type_name() + " to " + typeid(T).name()); } } /// get a boolean (explicit) template ::value, int>::type = 0> inline T get() const { switch (m_type) { case (value_t::boolean): return m_value.boolean; default: throw std::logic_error("cannot cast " + type_name() + " to " + typeid(T).name()); } } /// get a number (explicit) template::value and std::is_arithmetic::value, int>::type = 0> inline T get() const { switch (m_type) { case (value_t::number_integer): return static_cast(m_value.number_integer); case (value_t::number_float): return static_cast(m_value.number_float); default: throw std::logic_error("cannot cast " + type_name() + " to " + typeid(T).name()); } } /// get a value (implicit) template inline operator T() const { return get(); } //////////////////// // element access // //////////////////// /// access specified element with bounds checking inline reference at(size_type pos) { // at only works for arrays if (m_type != value_t::array) { throw std::runtime_error("cannot use at with " + type_name()); } return m_value.array->at(pos); } /// access specified element with bounds checking inline const_reference at(size_type pos) const { // at only works for arrays if (m_type != value_t::array) { throw std::runtime_error("cannot use at with " + type_name()); } return m_value.array->at(pos); } /// access specified element inline reference operator[](size_type pos) { // at only works for arrays if (m_type != value_t::array) { throw std::runtime_error("cannot use [] with " + type_name()); } return m_value.array->operator[](pos); } /// access specified element inline const_reference operator[](size_type pos) const { // at only works for arrays if (m_type != value_t::array) { throw std::runtime_error("cannot use [] with " + type_name()); } return m_value.array->operator[](pos); } /// access specified element with bounds checking inline reference at(const typename object_t::key_type& key) { // at only works for objects if (m_type != value_t::object) { throw std::runtime_error("cannot use at with " + type_name()); } return m_value.object->at(key); } /// access specified element with bounds checking inline const_reference at(const typename object_t::key_type& key) const { // at only works for objects if (m_type != value_t::object) { throw std::runtime_error("cannot use at with " + type_name()); } return m_value.object->at(key); } /// access specified element inline reference operator[](const typename object_t::key_type& key) { // at only works for objects if (m_type != value_t::object) { throw std::runtime_error("cannot use [] with " + type_name()); } return m_value.object->operator[](key); } /// access specified element inline const_reference operator[](const typename object_t::key_type& key) const { // at only works for objects if (m_type != value_t::object) { throw std::runtime_error("cannot use [] with " + type_name()); } return m_value.object->operator[](key); } /// access specified element (needed for clang) template inline reference operator[](const T (&key)[n]) { // at only works for objects if (m_type != value_t::object) { throw std::runtime_error("cannot use [] with " + type_name()); } return m_value.object->operator[](key); } /// access specified element (needed for clang) template inline const_reference operator[](const T (&key)[n]) const { // at only works for objects if (m_type != value_t::object) { throw std::runtime_error("cannot use [] with " + type_name()); } return m_value.object->operator[](key); } /// find an element in an object inline iterator find(typename object_t::key_type key) { auto result = end(); if (m_type == value_t::object) { result.m_it.object_iterator = m_value.object->find(key); } return result; } /// find an element in an object inline const_iterator find(typename object_t::key_type key) const { auto result = cend(); if (m_type == value_t::object) { result.m_it.object_iterator = m_value.object->find(key); } return result; } /////////////// // iterators // /////////////// /// returns an iterator to the beginning of the container inline iterator begin() noexcept { iterator result(this); result.set_begin(); return result; } /// returns a const iterator to the beginning of the container inline const_iterator begin() const noexcept { const_iterator result(this); result.set_begin(); return result; } /// returns a const iterator to the beginning of the container inline const_iterator cbegin() const noexcept { const_iterator result(this); result.set_begin(); return result; } /// returns an iterator to the end of the container inline iterator end() noexcept { iterator result(this); result.set_end(); return result; } /// returns a const iterator to the end of the container inline const_iterator end() const noexcept { const_iterator result(this); result.set_end(); return result; } /// returns a const iterator to the end of the container inline const_iterator cend() const noexcept { const_iterator result(this); result.set_end(); return result; } /// returns a reverse iterator to the beginning inline reverse_iterator rbegin() noexcept { return reverse_iterator(end()); } /// returns a reverse iterator to the beginning inline const_reverse_iterator rbegin() const noexcept { return const_reverse_iterator(end()); } /// returns a reverse iterator to the end inline reverse_iterator rend() noexcept { return reverse_iterator(begin()); } /// returns a reverse iterator to the end inline const_reverse_iterator rend() const noexcept { return const_reverse_iterator(begin()); } /// returns a reverse iterator to the beginning inline const_reverse_iterator crbegin() const noexcept { return const_reverse_iterator(cend()); } /// returns a reverse iterator to the end inline const_reverse_iterator crend() const noexcept { return const_reverse_iterator(cbegin()); } ////////////// // capacity // ////////////// /// checks whether the container is empty inline bool empty() const noexcept { switch (m_type) { case (value_t::null): { return true; } case (value_t::array): { return m_value.array->empty(); } case (value_t::object): { return m_value.object->empty(); } default: { // all other types are nonempty return false; } } } /// returns the number of elements inline size_type size() const noexcept { switch (m_type) { case (value_t::null): { return 0; } case (value_t::array): { return m_value.array->size(); } case (value_t::object): { return m_value.object->size(); } default: { // all other types have size 1 return 1; } } } /// returns the maximum possible number of elements inline size_type max_size() const noexcept { switch (m_type) { case (value_t::null): { return 0; } case (value_t::array): { return m_value.array->max_size(); } case (value_t::object): { return m_value.object->max_size(); } default: { // all other types have max_size 1 return 1; } } } /////////////// // modifiers // /////////////// /// clears the contents inline void clear() noexcept { switch (m_type) { case (value_t::null): { break; } case (value_t::number_integer): { m_value.number_integer = 0; break; } case (value_t::number_float): { m_value.number_float = 0.0; break; } case (value_t::boolean): { m_value.boolean = false; break; } case (value_t::string): { m_value.string->clear(); break; } case (value_t::array): { m_value.array->clear(); break; } case (value_t::object): { m_value.object->clear(); break; } } } /// add an object to an array inline void push_back(basic_json&& value) { // push_back only works for null objects or arrays if (not(m_type == value_t::null or m_type == value_t::array)) { throw std::runtime_error("cannot add element to " + type_name()); } // transform null object into an array if (m_type == value_t::null) { m_type = value_t::array; m_value.array = new array_t; } // add element to array (move semantics) m_value.array->push_back(std::move(value)); // invalidate object value.m_type = value_t::null; } /// add an object to an array inline reference operator+=(basic_json&& value) { push_back(std::move(value)); return *this; } /// add an object to an array inline void push_back(const basic_json& value) { // push_back only works for null objects or arrays if (not(m_type == value_t::null or m_type == value_t::array)) { throw std::runtime_error("cannot add element to " + type_name()); } // transform null object into an array if (m_type == value_t::null) { m_type = value_t::array; m_value.array = new array_t; } // add element to array m_value.array->push_back(value); } /// add an object to an array inline reference operator+=(const basic_json& value) { push_back(value); return *this; } /// add an object to an object inline void push_back(const typename object_t::value_type& value) { // push_back only works for null objects or objects if (not(m_type == value_t::null or m_type == value_t::object)) { throw std::runtime_error("cannot add element to " + type_name()); } // transform null object into an object if (m_type == value_t::null) { m_type = value_t::object; m_value.object = new object_t; } // add element to array m_value.object->insert(value); } /// add an object to an object inline reference operator+=(const typename object_t::value_type& value) { push_back(value); return operator[](value.first); } /// swaps the contents inline void swap(reference other) noexcept ( std::is_nothrow_move_constructible::value and std::is_nothrow_move_assignable::value and std::is_nothrow_move_constructible::value and std::is_nothrow_move_assignable::value ) { std::swap(m_type, other.m_type); std::swap(m_value, other.m_value); } /// swaps the contents inline void swap(array_t& other) { // swap only works for arrays if (m_type != value_t::array) { throw std::runtime_error("cannot use swap with " + type_name()); } // swap arrays std::swap(*(m_value.array), other); } /// swaps the contents inline void swap(object_t& other) { // swap only works for objects if (m_type != value_t::object) { throw std::runtime_error("cannot use swap with " + type_name()); } // swap arrays std::swap(*(m_value.object), other); } /// swaps the contents inline void swap(string_t& other) { // swap only works for strings if (m_type != value_t::string) { throw std::runtime_error("cannot use swap with " + type_name()); } // swap arrays std::swap(*(m_value.string), other); } ////////////////////////////////////////// // lexicographical comparison operators // ////////////////////////////////////////// /// comparison: equal friend bool operator==(const_reference lhs, const_reference rhs) { switch (lhs.type()) { case (value_t::array): { if (rhs.type() == value_t::array) { return *lhs.m_value.array == *rhs.m_value.array; } break; } case (value_t::object): { if (rhs.type() == value_t::object) { return *lhs.m_value.object == *rhs.m_value.object; } break; } case (value_t::null): { if (rhs.type() == value_t::null) { return true; } break; } case (value_t::string): { if (rhs.type() == value_t::string) { return *lhs.m_value.string == *rhs.m_value.string; } break; } case (value_t::boolean): { if (rhs.type() == value_t::boolean) { return lhs.m_value.boolean == rhs.m_value.boolean; } break; } case (value_t::number_integer): { if (rhs.type() == value_t::number_integer) { return lhs.m_value.number_integer == rhs.m_value.number_integer; } if (rhs.type() == value_t::number_float) { return lhs.m_value.number_integer == static_cast(rhs.m_value.number_float); } break; } case (value_t::number_float): { if (rhs.type() == value_t::number_integer) { return lhs.m_value.number_float == static_cast(rhs.m_value.number_integer); } if (rhs.type() == value_t::number_float) { return lhs.m_value.number_float == rhs.m_value.number_float; } break; } } return false; } /// comparison: not equal friend bool operator!=(const_reference lhs, const_reference rhs) { return not (lhs == rhs); } /// comparison: less than friend bool operator<(const_reference lhs, const_reference rhs) { switch (lhs.type()) { case (value_t::array): { if (rhs.type() == value_t::array) { return *lhs.m_value.array < *rhs.m_value.array; } break; } case (value_t::object): { if (rhs.type() == value_t::object) { return *lhs.m_value.object < *rhs.m_value.object; } break; } case (value_t::null): { if (rhs.type() == value_t::null) { return false; } break; } case (value_t::string): { if (rhs.type() == value_t::string) { return *lhs.m_value.string < *rhs.m_value.string; } break; } case (value_t::boolean): { if (rhs.type() == value_t::boolean) { return lhs.m_value.boolean < rhs.m_value.boolean; } break; } case (value_t::number_integer): { if (rhs.type() == value_t::number_integer) { return lhs.m_value.number_integer < rhs.m_value.number_integer; } if (rhs.type() == value_t::number_float) { return lhs.m_value.number_integer < static_cast(rhs.m_value.number_float); } break; } case (value_t::number_float): { if (rhs.type() == value_t::number_integer) { return lhs.m_value.number_float < static_cast(rhs.m_value.number_integer); } if (rhs.type() == value_t::number_float) { return lhs.m_value.number_float < rhs.m_value.number_float; } break; } } return false; } /// comparison: less than or equal friend bool operator<=(const_reference lhs, const_reference rhs) { return not (rhs < lhs); } /// comparison: greater than friend bool operator>(const_reference lhs, const_reference rhs) { return not (lhs <= rhs); } /// comparison: greater than or equal friend bool operator>=(const_reference lhs, const_reference rhs) { return not (lhs < rhs); } /////////////////// // serialization // /////////////////// /// serialize to stream friend std::ostream& operator<<(std::ostream& o, const basic_json& j) { // read width member and use it as indentation parameter if nonzero const int indentation = (o.width() == 0) ? -1 : o.width(); o << j.dump(indentation); return o; } /// serialize to stream friend std::ostream& operator>>(const basic_json& j, std::ostream& o) { // read width member and use it as indentation parameter if nonzero const int indentation = (o.width() == 0) ? -1 : o.width(); o << j.dump(indentation); return o; } ///////////////////// // deserialization // ///////////////////// /// deserialize from string static basic_json parse(const string_t& s) { return parser(s).parse(); } /// deserialize from stream friend std::istream& operator>>(std::istream& i, basic_json& j) { j = parser(i).parse(); return i; } /// deserialize from stream friend std::istream& operator<<(basic_json& j, std::istream& i) { j = parser(i).parse(); return i; } private: /////////////////////////// // convenience functions // /////////////////////////// /// return the type as string inline string_t type_name() const noexcept { switch (m_type) { case (value_t::null): { return "null"; } case (value_t::object): { return "object"; } case (value_t::array): { return "array"; } case (value_t::string): { return "string"; } case (value_t::boolean): { return "boolean"; } default: { return "number"; } } } /*! @brief escape a string Escape a string by replacing certain special characters by a sequence of an escape character (backslash) and another character and other control characters by a sequence of "\u" followed by a four-digit hex representation. @param s the string to escape @return escaped string */ static string_t escape_string(const string_t& s) noexcept { // create a result string of at least the size than s string_t result; result.reserve(s.size()); for (const auto c : s) { switch (c) { // quotation mark (0x22) case '"': { result += "\\\""; break; } // reverse solidus (0x5c) case '\\': { result += "\\\\"; break; } // backspace (0x08) case '\b': { result += "\\b"; break; } // formfeed (0x0c) case '\f': { result += "\\f"; break; } // newline (0x0a) case '\n': { result += "\\n"; break; } // carriage return (0x0d) case '\r': { result += "\\r"; break; } // horizontal tab (0x09) case '\t': { result += "\\t"; break; } default: { if (c <= 0x1f) { // control characters (everything between 0x00 and 0x1f) // -> create four-digit hex representation std::stringstream ss; ss << "\\u" << std::hex << std::setw(4) << std::setfill('0') << int(c); result += ss.str(); } else { // all other characters are added as-is result.append(1, c); } break; } } } return result; } /*! @brief internal implementation of the serialization function This function is called by the public member function dump and organizes the serializaion internally. The indentation level is propagated as additional parameter. In case of arrays and objects, the function is called recursively. Note that - strings and object keys are escaped using escape_string() - numbers are converted to a string before output using std::to_string() @param prettyPrint whether the output shall be pretty-printed @param indentStep the indent level @param currentIndent the current indent level (only used internally) */ inline string_t dump(const bool prettyPrint, const unsigned int indentStep, const unsigned int currentIndent = 0) const noexcept { // variable to hold indentation for recursive calls auto new_indent = currentIndent; // helper function to return whitespace as indentation const auto indent = [prettyPrint, &new_indent]() { return prettyPrint ? string_t(new_indent, ' ') : string_t(); }; switch (m_type) { case (value_t::object): { if (m_value.object->empty()) { return "{}"; } string_t result = "{"; // increase indentation if (prettyPrint) { new_indent += indentStep; result += "\n"; } for (auto i = m_value.object->cbegin(); i != m_value.object->cend(); ++i) { if (i != m_value.object->cbegin()) { result += prettyPrint ? ",\n" : ","; } result += indent() + "\"" + escape_string(i->first) + "\":" + (prettyPrint ? " " : "") + i->second.dump(prettyPrint, indentStep, new_indent); } // decrease indentation if (prettyPrint) { new_indent -= indentStep; result += "\n"; } return result + indent() + "}"; } case (value_t::array): { if (m_value.array->empty()) { return "[]"; } string_t result = "["; // increase indentation if (prettyPrint) { new_indent += indentStep; result += "\n"; } for (auto i = m_value.array->cbegin(); i != m_value.array->cend(); ++i) { if (i != m_value.array->cbegin()) { result += prettyPrint ? ",\n" : ","; } result += indent() + i->dump(prettyPrint, indentStep, new_indent); } // decrease indentation if (prettyPrint) { new_indent -= indentStep; result += "\n"; } return result + indent() + "]"; } case (value_t::string): { return string_t("\"") + escape_string(*m_value.string) + "\""; } case (value_t::boolean): { return m_value.boolean ? "true" : "false"; } case (value_t::number_integer): { return std::to_string(m_value.number_integer); } case (value_t::number_float): { return std::to_string(m_value.number_float); } default: { return "null"; } } } private: ////////////////////// // member variables // ////////////////////// /// the type of the current element value_t m_type = value_t::null; /// whether the type of JSON object may change later bool m_final = false; /// the value of the current element json_value m_value = {}; private: /////////////// // iterators // /////////////// /// values of a generic iterator type of non-container JSON values enum class generic_iterator_value { /// the iterator was not initialized uninitialized, /// the iterator points to the only value begin, /// the iterator points past the only value end, /// the iterator points to an invalid value invalid }; /// an iterator value template union internal_iterator { /// iterator for JSON objects object_iterator_t object_iterator; /// iterator for JSON arrays array_iterator_t array_iterator; /// generic iteraotr for all other value types generic_iterator_value generic_iterator; /// default constructor internal_iterator() : generic_iterator(generic_iterator_value::uninitialized) {} /// constructor for object iterators internal_iterator(object_iterator_t v) : object_iterator(v) {} /// constructor for array iterators internal_iterator(array_iterator_t v) : array_iterator(v) {} /// constructor for generic iterators internal_iterator(generic_iterator_value v) : generic_iterator(v) {} }; public: /// a bidirectional iterator for the basic_json class class iterator : public std::iterator { public: /// the type of the values when the iterator is dereferenced using value_type = basic_json::value_type; /// a type to represent differences between iterators using difference_type = basic_json::difference_type; /// defines a pointer to the type iterated over (value_type) using pointer = basic_json::pointer; /// defines a reference to the type iterated over (value_type) using reference = basic_json::reference; /// the category of the iterator using iterator_category = std::bidirectional_iterator_tag; /// default constructor inline iterator() = default; /// constructor for a given JSON instance inline iterator(pointer object) : m_object(object) { switch (m_object->m_type) { case (basic_json::value_t::object): { m_it.object_iterator = typename object_t::iterator(); break; } case (basic_json::value_t::array): { m_it.array_iterator = typename array_t::iterator(); break; } default: { m_it.generic_iterator = generic_iterator_value::uninitialized; break; } } } /// copy assignment inline iterator& operator=(const iterator& other) noexcept { m_object = other.m_object; m_it = other.m_it; return *this; } /// set the iterator to the first value inline void set_begin() noexcept { switch (m_object->m_type) { case (basic_json::value_t::object): { m_it.object_iterator = m_object->m_value.object->begin(); break; } case (basic_json::value_t::array): { m_it.array_iterator = m_object->m_value.array->begin(); break; } case (basic_json::value_t::null): { // set to end so begin()==end() is true: null is empty m_it.generic_iterator = generic_iterator_value::end; break; } default: { m_it.generic_iterator = generic_iterator_value::begin; break; } } } /// set the iterator past the last value inline void set_end() noexcept { switch (m_object->m_type) { case (basic_json::value_t::object): { m_it.object_iterator = m_object->m_value.object->end(); break; } case (basic_json::value_t::array): { m_it.array_iterator = m_object->m_value.array->end(); break; } default: { m_it.generic_iterator = generic_iterator_value::end; break; } } } /// return a reference to the value pointed to by the iterator inline reference operator*() const { switch (m_object->m_type) { case (basic_json::value_t::object): { return m_it.object_iterator->second; } case (basic_json::value_t::array): { return *m_it.array_iterator; } case (basic_json::value_t::null): { throw std::out_of_range("cannot get value"); } default: { if (m_it.generic_iterator == generic_iterator_value::begin) { return *m_object; } else { throw std::out_of_range("cannot get value"); } } } } /// dereference the iterator inline pointer operator->() const { switch (m_object->m_type) { case (basic_json::value_t::object): { return &(m_it.object_iterator->second); } case (basic_json::value_t::array): { return &*m_it.array_iterator; } default: { if (m_it.generic_iterator == generic_iterator_value::begin) { return m_object; } else { throw std::out_of_range("cannot get value"); } } } } /// post-increment (it++) inline iterator operator++(int) { iterator result = *this; switch (m_object->m_type) { case (basic_json::value_t::object): { m_it.object_iterator++; break; } case (basic_json::value_t::array): { m_it.array_iterator++; break; } default: { if (m_it.generic_iterator == generic_iterator_value::begin) { m_it.generic_iterator = generic_iterator_value::end; } else { m_it.generic_iterator = generic_iterator_value::invalid; } break; } } return result; } /// pre-increment (++it) inline iterator& operator++() { switch (m_object->m_type) { case (basic_json::value_t::object): { ++m_it.object_iterator; break; } case (basic_json::value_t::array): { ++m_it.array_iterator; break; } default: { if (m_it.generic_iterator == generic_iterator_value::begin) { m_it.generic_iterator = generic_iterator_value::end; } else { m_it.generic_iterator = generic_iterator_value::invalid; } break; } } return *this; } /// post-decrement (it--) inline iterator operator--(int) { iterator result = *this; switch (m_object->m_type) { case (basic_json::value_t::object): { m_it.object_iterator--; break; } case (basic_json::value_t::array): { m_it.array_iterator--; break; } case (basic_json::value_t::null): { m_it.generic_iterator = generic_iterator_value::invalid; break; } default: { if (m_it.generic_iterator == generic_iterator_value::end) { m_it.generic_iterator = generic_iterator_value::begin; } else { m_it.generic_iterator = generic_iterator_value::invalid; } break; } } return result; } /// pre-decrement (--it) inline iterator& operator--() { switch (m_object->m_type) { case (basic_json::value_t::object): { --m_it.object_iterator; break; } case (basic_json::value_t::array): { --m_it.array_iterator; break; } case (basic_json::value_t::null): { m_it.generic_iterator = generic_iterator_value::invalid; break; } default: { if (m_it.generic_iterator == generic_iterator_value::end) { m_it.generic_iterator = generic_iterator_value::begin; } else { m_it.generic_iterator = generic_iterator_value::invalid; } break; } } return *this; } /// comparison: equal inline bool operator==(const iterator& other) const { if (m_object != other.m_object or m_object->m_type != other.m_object->m_type) { return false; } switch (m_object->m_type) { case (basic_json::value_t::object): { return (m_it.object_iterator == other.m_it.object_iterator); } case (basic_json::value_t::array): { return (m_it.array_iterator == other.m_it.array_iterator); } default: { return (m_it.generic_iterator == other.m_it.generic_iterator); } } } /// comparison: not equal inline bool operator!=(const iterator& other) const { return not operator==(other); } private: /// associated JSON instance pointer m_object = nullptr; /// the actual iterator of the associated instance internal_iterator m_it; }; /// a const bidirectional iterator for the basic_json class class const_iterator : public std::iterator { public: /// the type of the values when the iterator is dereferenced using value_type = basic_json::value_type; /// a type to represent differences between iterators using difference_type = basic_json::difference_type; /// defines a pointer to the type iterated over (value_type) using pointer = basic_json::const_pointer; /// defines a reference to the type iterated over (value_type) using reference = basic_json::const_reference; /// the category of the iterator using iterator_category = std::bidirectional_iterator_tag; /// default constructor inline const_iterator() = default; /// constructor for a given JSON instance inline const_iterator(pointer object) : m_object(object) { switch (m_object->m_type) { case (basic_json::value_t::object): { m_it.object_iterator = typename object_t::const_iterator(); break; } case (basic_json::value_t::array): { m_it.array_iterator = typename array_t::const_iterator(); break; } default: { m_it.generic_iterator = generic_iterator_value::uninitialized; break; } } } /// copy constructor given a nonconst iterator inline const_iterator(const iterator& other) : m_object(other.m_object) { switch (m_object->m_type) { case (basic_json::value_t::object): { m_it.object_iterator = other.m_it.object_iterator; break; } case (basic_json::value_t::array): { m_it.array_iterator = other.m_it.array_iterator; break; } default: { m_it.generic_iterator = other.m_it.generic_iterator; break; } } } /// copy assignment inline const_iterator& operator=(const const_iterator& other) noexcept { m_object = other.m_object; m_it = other.m_it; return *this; } /// set the iterator to the first value inline void set_begin() noexcept { switch (m_object->m_type) { case (basic_json::value_t::object): { m_it.object_iterator = m_object->m_value.object->cbegin(); break; } case (basic_json::value_t::array): { m_it.array_iterator = m_object->m_value.array->cbegin(); break; } case (basic_json::value_t::null): { // set to end so begin()==end() is true: null is empty m_it.generic_iterator = generic_iterator_value::end; break; } default: { m_it.generic_iterator = generic_iterator_value::begin; break; } } } /// set the iterator past the last value inline void set_end() noexcept { switch (m_object->m_type) { case (basic_json::value_t::object): { m_it.object_iterator = m_object->m_value.object->cend(); break; } case (basic_json::value_t::array): { m_it.array_iterator = m_object->m_value.array->cend(); break; } default: { m_it.generic_iterator = generic_iterator_value::end; break; } } } /// return a reference to the value pointed to by the iterator inline reference operator*() const { switch (m_object->m_type) { case (basic_json::value_t::object): { return m_it.object_iterator->second; } case (basic_json::value_t::array): { return *m_it.array_iterator; } case (basic_json::value_t::null): { throw std::out_of_range("cannot get value"); } default: { if (m_it.generic_iterator == generic_iterator_value::begin) { return *m_object; } else { throw std::out_of_range("cannot get value"); } } } } /// dereference the iterator inline pointer operator->() const { switch (m_object->m_type) { case (basic_json::value_t::object): { return &(m_it.object_iterator->second); } case (basic_json::value_t::array): { return &*m_it.array_iterator; } default: { if (m_it.generic_iterator == generic_iterator_value::begin) { return m_object; } else { throw std::out_of_range("cannot get value"); } } } } /// post-increment (it++) inline const_iterator operator++(int) { const_iterator result = *this; switch (m_object->m_type) { case (basic_json::value_t::object): { m_it.object_iterator++; break; } case (basic_json::value_t::array): { m_it.array_iterator++; break; } default: { if (m_it.generic_iterator == generic_iterator_value::begin) { m_it.generic_iterator = generic_iterator_value::end; } else { m_it.generic_iterator = generic_iterator_value::invalid; } break; } } return result; } /// pre-increment (++it) inline const_iterator& operator++() { switch (m_object->m_type) { case (basic_json::value_t::object): { ++m_it.object_iterator; break; } case (basic_json::value_t::array): { ++m_it.array_iterator; break; } default: { if (m_it.generic_iterator == generic_iterator_value::begin) { m_it.generic_iterator = generic_iterator_value::end; } else { m_it.generic_iterator = generic_iterator_value::invalid; } break; } } return *this; } /// post-decrement (it--) inline const_iterator operator--(int) { const_iterator result = *this; switch (m_object->m_type) { case (basic_json::value_t::object): { m_it.object_iterator--; break; } case (basic_json::value_t::array): { m_it.array_iterator--; break; } case (basic_json::value_t::null): { m_it.generic_iterator = generic_iterator_value::invalid; break; } default: { if (m_it.generic_iterator == generic_iterator_value::end) { m_it.generic_iterator = generic_iterator_value::begin; } else { m_it.generic_iterator = generic_iterator_value::invalid; } break; } } return result; } /// pre-decrement (--it) inline const_iterator& operator--() { switch (m_object->m_type) { case (basic_json::value_t::object): { --m_it.object_iterator; break; } case (basic_json::value_t::array): { --m_it.array_iterator; break; } case (basic_json::value_t::null): { m_it.generic_iterator = generic_iterator_value::invalid; break; } default: { if (m_it.generic_iterator == generic_iterator_value::end) { m_it.generic_iterator = generic_iterator_value::begin; } else { m_it.generic_iterator = generic_iterator_value::invalid; } break; } } return *this; } /// comparison: equal inline bool operator==(const const_iterator& other) const { if (m_object != other.m_object or m_object->m_type != other.m_object->m_type) { return false; } switch (m_object->m_type) { case (basic_json::value_t::object): { return (m_it.object_iterator == other.m_it.object_iterator); } case (basic_json::value_t::array): { return (m_it.array_iterator == other.m_it.array_iterator); } default: { return (m_it.generic_iterator == other.m_it.generic_iterator); } } } /// comparison: not equal inline bool operator!=(const const_iterator& other) const { return not operator==(other); } private: /// associated JSON instance pointer m_object = nullptr; /// the actual iterator of the associated instance internal_iterator m_it; }; private: ////////////////////// // lexer and parser // ////////////////////// /*! @brief lexical analysis This class organizes the lexical analysis during JSON deserialization. The core of it is a scanner generated by re2c that processes a buffer and recognizes tokens according to RFC 7159 and ECMA-404. */ class lexer { public: /// token types for the parser enum class token_type { uninitialized, ///< indicating the scanner is uninitialized literal_true, ///< the "true" literal literal_false, ///< the "false" literal literal_null, ///< the "null" literal value_string, ///< a string - use get_string() for actual value value_number, ///< a number - use get_number() for actual value begin_array, ///< the character for array begin "[" begin_object, ///< the character for object begin "{" end_array, ///< the character for array end "]" end_object, ///< the character for object end "}" name_separator, ///< the name separator ":" value_separator, ///< the value separator "," parse_error, ///< indicating a parse error end_of_input ///< indicating the end of the input buffer }; /// the char type to use in the lexer using lexer_char_t = typename string_t::value_type; /// constructor with a given buffer inline lexer(const string_t& s) noexcept : m_content(s.c_str()) { m_start = m_cursor = m_content; m_limit = m_content + s.size(); } /// default constructor inline lexer() = default; /*! @brief create a string from a Unicode code point @param codepoint the code point (must be in [0x0, 0x10ffff] @return string representation of the code point @exception std::out_of_range if code point is >0x10ffff @exception std::invalid_argument if the low surrogate is invalid @see */ inline static string_t to_unicode(const size_t codepoint1, size_t codepoint2 = 0) { string_t result; // calculate the codepoint from the given code points size_t codepoint = codepoint1; if (codepoint1 >= 0xD800 and codepoint1 <= 0xDBFF) { if (codepoint2 >= 0xDC00 and codepoint2 <= 0xDFFF) { codepoint = // high surrogate occupies the most significant 22 bits (codepoint1 << 10) // low surrogate occupies the least significant 15 bits + codepoint2 // there is still the 0xD800, 0xDC00 and 0x10000 noise // in the result so we have to substract with: // (0xD800 << 10) + DC00 - 0x10000 = 0x35FDC00 - 0x35FDC00; } else { throw std::invalid_argument("missing or wrong low surrogate"); } } if (codepoint <= 0x7f) { // 1-byte characters: 0xxxxxxx (ASCI) result.append(1, static_cast(codepoint)); } else if (codepoint <= 0x7ff) { // 2-byte characters: 110xxxxx 10xxxxxx result.append(1, static_cast(0xC0 | ((codepoint >> 6) & 0x1F))); result.append(1, static_cast(0x80 | (codepoint & 0x3F))); } else if (codepoint <= 0xffff) { // 3-byte characters: 1110xxxx 10xxxxxx 10xxxxxx result.append(1, static_cast(0xE0 | ((codepoint >> 12) & 0x0F))); result.append(1, static_cast(0x80 | ((codepoint >> 6) & 0x3F))); result.append(1, static_cast(0x80 | (codepoint & 0x3F))); } else if (codepoint <= 0x10ffff) { // 4-byte characters: 11110xxx 10xxxxxx 10xxxxxx 10xxxxxx result.append(1, static_cast(0xF0 | ((codepoint >> 18) & 0x07))); result.append(1, static_cast(0x80 | ((codepoint >> 12) & 0x3F))); result.append(1, static_cast(0x80 | ((codepoint >> 6) & 0x3F))); result.append(1, static_cast(0x80 | (codepoint & 0x3F))); } else { throw std::out_of_range("code points above 0x10FFFF are invalid"); } return result; } /// return name of values of type token_type inline static std::string token_type_name(token_type t) noexcept { switch (t) { case (token_type::uninitialized): return ""; case (token_type::literal_true): return "true literal"; case (token_type::literal_false): return "false literal"; case (token_type::literal_null): return "null literal"; case (token_type::value_string): return "string literal"; case (token_type::value_number): return "number literal"; case (token_type::begin_array): return "["; case (token_type::begin_object): return "{"; case (token_type::end_array): return "]"; case (token_type::end_object): return "}"; case (token_type::name_separator): return ":"; case (token_type::value_separator): return ","; case (token_type::end_of_input): return ""; default: return ""; } } /*! This function implements a scanner for JSON. It is specified using regular expressions that try to follow RFC 7159 and ECMA-404 as close as possible. These regular expressions are then translated into a deterministic finite automaton (DFA) by the tool re2c . As a result, the translated code for this function consists of a large block of code with goto jumps. @return the class of the next token read from the buffer */ inline token_type scan() noexcept { // pointer for backtracking information const lexer_char_t* m_marker = nullptr; // remember the begin of the token m_start = m_cursor; { lexer_char_t yych; unsigned int yyaccept = 0; static const unsigned char yybm[] = { 0, 64, 64, 64, 64, 64, 64, 64, 64, 96, 96, 64, 64, 96, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 96, 64, 0, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 192, 192, 192, 192, 192, 192, 192, 192, 192, 192, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 0, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, }; yych = *m_cursor; if (yych <= '9') { if (yych <= ' ') { if (yych <= '\n') { if (yych <= 0x00) { goto basic_json_parser_27; } if (yych <= 0x08) { goto basic_json_parser_29; } if (yych >= '\n') { goto basic_json_parser_4; } } else { if (yych == '\r') { goto basic_json_parser_2; } if (yych <= 0x1F) { goto basic_json_parser_29; } } } else { if (yych <= ',') { if (yych == '"') { goto basic_json_parser_26; } if (yych <= '+') { goto basic_json_parser_29; } goto basic_json_parser_14; } else { if (yych <= '-') { goto basic_json_parser_22; } if (yych <= '/') { goto basic_json_parser_29; } if (yych <= '0') { goto basic_json_parser_23; } goto basic_json_parser_25; } } } else { if (yych <= 'm') { if (yych <= '\\') { if (yych <= ':') { goto basic_json_parser_16; } if (yych == '[') { goto basic_json_parser_6; } goto basic_json_parser_29; } else { if (yych <= ']') { goto basic_json_parser_8; } if (yych == 'f') { goto basic_json_parser_21; } goto basic_json_parser_29; } } else { if (yych <= 'z') { if (yych <= 'n') { goto basic_json_parser_18; } if (yych == 't') { goto basic_json_parser_20; } goto basic_json_parser_29; } else { if (yych <= '{') { goto basic_json_parser_10; } if (yych == '}') { goto basic_json_parser_12; } goto basic_json_parser_29; } } } basic_json_parser_2: ++m_cursor; yych = *m_cursor; goto basic_json_parser_5; basic_json_parser_3: { return scan(); } basic_json_parser_4: ++m_cursor; yych = *m_cursor; basic_json_parser_5: if (yybm[0 + yych] & 32) { goto basic_json_parser_4; } goto basic_json_parser_3; basic_json_parser_6: ++m_cursor; { return token_type::begin_array; } basic_json_parser_8: ++m_cursor; { return token_type::end_array; } basic_json_parser_10: ++m_cursor; { return token_type::begin_object; } basic_json_parser_12: ++m_cursor; { return token_type::end_object; } basic_json_parser_14: ++m_cursor; { return token_type::value_separator; } basic_json_parser_16: ++m_cursor; { return token_type::name_separator; } basic_json_parser_18: yyaccept = 0; yych = *(m_marker = ++m_cursor); if (yych == 'u') { goto basic_json_parser_59; } basic_json_parser_19: { return token_type::parse_error; } basic_json_parser_20: yyaccept = 0; yych = *(m_marker = ++m_cursor); if (yych == 'r') { goto basic_json_parser_55; } goto basic_json_parser_19; basic_json_parser_21: yyaccept = 0; yych = *(m_marker = ++m_cursor); if (yych == 'a') { goto basic_json_parser_50; } goto basic_json_parser_19; basic_json_parser_22: yych = *++m_cursor; if (yych <= '/') { goto basic_json_parser_19; } if (yych <= '0') { goto basic_json_parser_49; } if (yych <= '9') { goto basic_json_parser_40; } goto basic_json_parser_19; basic_json_parser_23: yyaccept = 1; yych = *(m_marker = ++m_cursor); if (yych <= 'D') { if (yych == '.') { goto basic_json_parser_42; } } else { if (yych <= 'E') { goto basic_json_parser_43; } if (yych == 'e') { goto basic_json_parser_43; } } basic_json_parser_24: { return token_type::value_number; } basic_json_parser_25: yyaccept = 1; yych = *(m_marker = ++m_cursor); goto basic_json_parser_41; basic_json_parser_26: yyaccept = 0; yych = *(m_marker = ++m_cursor); if (yych <= 0x00) { goto basic_json_parser_19; } goto basic_json_parser_31; basic_json_parser_27: ++m_cursor; { return token_type::end_of_input; } basic_json_parser_29: yych = *++m_cursor; goto basic_json_parser_19; basic_json_parser_30: ++m_cursor; yych = *m_cursor; basic_json_parser_31: if (yybm[0 + yych] & 64) { goto basic_json_parser_30; } if (yych <= 0x00) { goto basic_json_parser_32; } if (yych <= '"') { goto basic_json_parser_34; } goto basic_json_parser_33; basic_json_parser_32: m_cursor = m_marker; if (yyaccept == 0) { goto basic_json_parser_19; } else { goto basic_json_parser_24; } basic_json_parser_33: ++m_cursor; yych = *m_cursor; if (yych <= 'e') { if (yych <= '/') { if (yych == '"') { goto basic_json_parser_30; } if (yych <= '.') { goto basic_json_parser_32; } goto basic_json_parser_30; } else { if (yych <= '\\') { if (yych <= '[') { goto basic_json_parser_32; } goto basic_json_parser_30; } else { if (yych == 'b') { goto basic_json_parser_30; } goto basic_json_parser_32; } } } else { if (yych <= 'q') { if (yych <= 'f') { goto basic_json_parser_30; } if (yych == 'n') { goto basic_json_parser_30; } goto basic_json_parser_32; } else { if (yych <= 's') { if (yych <= 'r') { goto basic_json_parser_30; } goto basic_json_parser_32; } else { if (yych <= 't') { goto basic_json_parser_30; } if (yych <= 'u') { goto basic_json_parser_36; } goto basic_json_parser_32; } } } basic_json_parser_34: ++m_cursor; { return token_type::value_string; } basic_json_parser_36: ++m_cursor; yych = *m_cursor; if (yych <= '@') { if (yych <= '/') { goto basic_json_parser_32; } if (yych >= ':') { goto basic_json_parser_32; } } else { if (yych <= 'F') { goto basic_json_parser_37; } if (yych <= '`') { goto basic_json_parser_32; } if (yych >= 'g') { goto basic_json_parser_32; } } basic_json_parser_37: ++m_cursor; yych = *m_cursor; if (yych <= '@') { if (yych <= '/') { goto basic_json_parser_32; } if (yych >= ':') { goto basic_json_parser_32; } } else { if (yych <= 'F') { goto basic_json_parser_38; } if (yych <= '`') { goto basic_json_parser_32; } if (yych >= 'g') { goto basic_json_parser_32; } } basic_json_parser_38: ++m_cursor; yych = *m_cursor; if (yych <= '@') { if (yych <= '/') { goto basic_json_parser_32; } if (yych >= ':') { goto basic_json_parser_32; } } else { if (yych <= 'F') { goto basic_json_parser_39; } if (yych <= '`') { goto basic_json_parser_32; } if (yych >= 'g') { goto basic_json_parser_32; } } basic_json_parser_39: ++m_cursor; yych = *m_cursor; if (yych <= '@') { if (yych <= '/') { goto basic_json_parser_32; } if (yych <= '9') { goto basic_json_parser_30; } goto basic_json_parser_32; } else { if (yych <= 'F') { goto basic_json_parser_30; } if (yych <= '`') { goto basic_json_parser_32; } if (yych <= 'f') { goto basic_json_parser_30; } goto basic_json_parser_32; } basic_json_parser_40: yyaccept = 1; m_marker = ++m_cursor; yych = *m_cursor; basic_json_parser_41: if (yybm[0 + yych] & 128) { goto basic_json_parser_40; } if (yych <= 'D') { if (yych != '.') { goto basic_json_parser_24; } } else { if (yych <= 'E') { goto basic_json_parser_43; } if (yych == 'e') { goto basic_json_parser_43; } goto basic_json_parser_24; } basic_json_parser_42: yych = *++m_cursor; if (yych <= '/') { goto basic_json_parser_32; } if (yych <= '9') { goto basic_json_parser_47; } goto basic_json_parser_32; basic_json_parser_43: yych = *++m_cursor; if (yych <= ',') { if (yych != '+') { goto basic_json_parser_32; } } else { if (yych <= '-') { goto basic_json_parser_44; } if (yych <= '/') { goto basic_json_parser_32; } if (yych <= '9') { goto basic_json_parser_45; } goto basic_json_parser_32; } basic_json_parser_44: yych = *++m_cursor; if (yych <= '/') { goto basic_json_parser_32; } if (yych >= ':') { goto basic_json_parser_32; } basic_json_parser_45: ++m_cursor; yych = *m_cursor; if (yych <= '/') { goto basic_json_parser_24; } if (yych <= '9') { goto basic_json_parser_45; } goto basic_json_parser_24; basic_json_parser_47: yyaccept = 1; m_marker = ++m_cursor; yych = *m_cursor; if (yych <= 'D') { if (yych <= '/') { goto basic_json_parser_24; } if (yych <= '9') { goto basic_json_parser_47; } goto basic_json_parser_24; } else { if (yych <= 'E') { goto basic_json_parser_43; } if (yych == 'e') { goto basic_json_parser_43; } goto basic_json_parser_24; } basic_json_parser_49: yyaccept = 1; yych = *(m_marker = ++m_cursor); if (yych <= 'D') { if (yych == '.') { goto basic_json_parser_42; } goto basic_json_parser_24; } else { if (yych <= 'E') { goto basic_json_parser_43; } if (yych == 'e') { goto basic_json_parser_43; } goto basic_json_parser_24; } basic_json_parser_50: yych = *++m_cursor; if (yych != 'l') { goto basic_json_parser_32; } yych = *++m_cursor; if (yych != 's') { goto basic_json_parser_32; } yych = *++m_cursor; if (yych != 'e') { goto basic_json_parser_32; } ++m_cursor; { return token_type::literal_false; } basic_json_parser_55: yych = *++m_cursor; if (yych != 'u') { goto basic_json_parser_32; } yych = *++m_cursor; if (yych != 'e') { goto basic_json_parser_32; } ++m_cursor; { return token_type::literal_true; } basic_json_parser_59: yych = *++m_cursor; if (yych != 'l') { goto basic_json_parser_32; } yych = *++m_cursor; if (yych != 'l') { goto basic_json_parser_32; } ++m_cursor; { return token_type::literal_null; } } } /// return string representation of last read token inline string_t get_token() const noexcept { return string_t(m_start, static_cast(m_cursor - m_start)); } /*! @brief return string value for string tokens The function iterates the characters between the opening and closing quotes of the string value. The complete string is the range [m_start,m_cursor). Consequently, we iterate from m_start+1 to m_cursor-1. We differentiate two cases: 1. Escaped characters. In this case, a new character is constructed according to the nature of the escape. Some escapes create new characters (e.g., @c "\\n" is replaced by @c "\n"), some are copied as is (e.g., @c "\\\\"). Furthermore, Unicode escapes of the shape @c "\\uxxxx" need special care. In this case, to_unicode takes care of the construction of the values. 2. Unescaped characters are copied as is. @return string value of current token without opening and closing quotes @exception std::out_of_range if to_unicode fails */ inline string_t get_string() const { string_t result; result.reserve(static_cast(m_cursor - m_start - 2)); // iterate the result between the quotes for (const typename string_t::value_type* i = m_start + 1; i < m_cursor - 1; ++i) { // process escaped characters if (*i == '\\') { // read next character ++i; switch (*i) { // the default escapes case 't': { result += "\t"; break; } case 'b': { result += "\b"; break; } case 'f': { result += "\f"; break; } case 'n': { result += "\n"; break; } case 'r': { result += "\r"; break; } // characters that are not "un"escsaped case '\\': { result += "\\\\"; break; } case '/': { result += "\\/"; break; } case '"': { result += "\\\""; break; } // unicode case 'u': { // get code xxxx from uxxxx auto codepoint = std::strtoul(std::string(i + 1, 4).c_str(), nullptr, 16); if (codepoint >= 0xD800 and codepoint <= 0xDBFF) { // make sure there is a subsequent unicode if ((i + 6 >= m_limit) or * (i + 5) != '\\' or * (i + 6) != 'u') { throw std::invalid_argument("missing low surrogate"); } // get code yyyy from uxxxx\uyyyy auto codepoint2 = std::strtoul(std::string(i + 7, 4).c_str(), nullptr, 16); result += to_unicode(codepoint, codepoint2); // skip the next 11 characters (xxxx\uyyyy) i += 11; } else { // add unicode character(s) result += to_unicode(codepoint); // skip the next four characters (xxxx) i += 4; } break; } } } else { // all other characters are just copied to the end of the // string result.append(1, *i); } } return result; } /*! @brief return number value for number tokens This function translates the last token into a floating point number. The pointer m_begin points to the beginning of the parsed number. We pass this pointer to std::strtod which sets endptr to the first character past the converted number. If this pointer is not the same as m_cursor, then either more or less characters have been used during the comparison. This can happen for inputs like "01" which will be treated like number 0 followed by number 1. @return the result of the number conversion or NAN if the conversion read past the current token. The latter case needs to be treated by the caller function. @exception std::range_error if passed value is out of range */ inline number_float_t get_number() const { // conversion lexer_char_t* endptr; const auto float_val = std::strtod(reinterpret_cast(m_start), &endptr); // return float_val if the whole number was translated and NAN // otherwise return (endptr == m_cursor) ? float_val : NAN; } private: /// the buffer const lexer_char_t* m_content = nullptr; /// pointer to he beginning of the current symbol const lexer_char_t* m_start = nullptr; /// pointer to the current symbol const lexer_char_t* m_cursor = nullptr; /// pointer to the end of the buffer const lexer_char_t* m_limit = nullptr; }; class parser { public: /// constructor for strings inline parser(const string_t& s) : m_buffer(s), m_lexer(m_buffer) { // read first token get_token(); } /// a parser reading from an input stream inline parser(std::istream& _is) { while (_is) { string_t input_line; std::getline(_is, input_line); m_buffer += input_line; } // initializer lexer m_lexer = lexer(m_buffer); // read first token get_token(); } /// public parser interface inline basic_json parse() { basic_json result = parse_internal(); expect(lexer::token_type::end_of_input); return result; } private: /// the actual parser inline basic_json parse_internal() { switch (last_token) { case (lexer::token_type::begin_object): { // explicitly set result to object to cope with {} basic_json result(value_t::object); // read next token get_token(); // closing } -> we are done if (last_token == lexer::token_type::end_object) { get_token(); return result; } // otherwise: parse key-value pairs do { // ugly, but could be fixed with loop reorganization if (last_token == lexer::token_type::value_separator) { get_token(); } // store key expect(lexer::token_type::value_string); const auto key = m_lexer.get_string(); // parse separator (:) get_token(); expect(lexer::token_type::name_separator); // parse value get_token(); result[key] = parse_internal(); } while (last_token == lexer::token_type::value_separator); // closing } expect(lexer::token_type::end_object); get_token(); return result; } case (lexer::token_type::begin_array): { // explicitly set result to object to cope with [] basic_json result(value_t::array); // read next token get_token(); // closing ] -> we are done if (last_token == lexer::token_type::end_array) { get_token(); return result; } // otherwise: parse values do { // ugly, but could be fixed with loop reorganization if (last_token == lexer::token_type::value_separator) { get_token(); } // parse value result.push_back(parse_internal()); } while (last_token == lexer::token_type::value_separator); // closing ] expect(lexer::token_type::end_array); get_token(); return result; } case (lexer::token_type::literal_null): { get_token(); return basic_json(nullptr); } case (lexer::token_type::value_string): { const auto s = m_lexer.get_string(); get_token(); return basic_json(s); } case (lexer::token_type::literal_true): { get_token(); return basic_json(true); } case (lexer::token_type::literal_false): { get_token(); return basic_json(false); } case (lexer::token_type::value_number): { auto float_val = m_lexer.get_number(); // NAN is returned if token could not be translated // completely if (std::isnan(float_val)) { throw std::invalid_argument(std::string("parse error - ") + m_lexer.get_token() + " is not a number"); } get_token(); // check if conversion loses precision const auto int_val = static_cast(float_val); if (float_val == int_val) { // we basic_json not lose precision -> return int return basic_json(int_val); } else { // we would lose precision -> returnfloat return basic_json(float_val); } } default: { std::string error_msg = "parse error - unexpected \'"; error_msg += m_lexer.get_token(); error_msg += "\' ("; error_msg += lexer::token_type_name(last_token) + ")"; throw std::invalid_argument(error_msg); } } } /// get next token from lexer inline typename lexer::token_type get_token() { last_token = m_lexer.scan(); return last_token; } inline void expect(typename lexer::token_type t) const { if (t != last_token) { std::string error_msg = "parse error - unexpected \'"; error_msg += m_lexer.get_token(); error_msg += "\' (" + lexer::token_type_name(last_token); error_msg += "); expected " + lexer::token_type_name(t); throw std::invalid_argument(error_msg); } } private: /// the buffer string_t m_buffer; /// the type of the last read token typename lexer::token_type last_token = lexer::token_type::uninitialized; /// the lexer lexer m_lexer; }; }; ///////////// // presets // ///////////// /// default JSON class using json = basic_json<>; } ///////////////////////// // nonmember functions // ///////////////////////// // specialization of std::swap, and std::hash namespace std { /// swaps the values of two JSON objects template <> inline void swap(nlohmann::json& j1, nlohmann::json& j2) noexcept( is_nothrow_move_constructible::value and is_nothrow_move_assignable::value ) { j1.swap(j2); } /// hash value for JSON objects template <> struct hash { /// return a hash value for a JSON object inline size_t operator()(const nlohmann::json& j) const { // a naive hashing via the string representation return hash()(j.dump()); } }; } /*! This operator implements a user-defined string literal for JSON objects. It can be used by adding \p "_json" to a string literal and returns a JSON object if no parse error occurred. @param s a string representation of a JSON object @return a JSON object */ inline nlohmann::json operator "" _json(const char* s, std::size_t) { return nlohmann::json::parse(s); } #endif