/*! @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_HPP #define NLOHMANN_JSON_HPP #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include // enable ssize_t on MinGW #ifdef __GNUC__ #ifdef __MINGW32__ #include #endif #endif /*! @brief namespace for Niels Lohmann @see https://github.com/nlohmann */ namespace nlohmann { // Helper to determine whether there's a key_type for T. // http://stackoverflow.com/a/7728728/266378 template struct has_mapped_type { private: template static char test(typename C::mapped_type*); template static int test(...); public: enum { value = sizeof(test(0)) == sizeof(char) }; }; /*! @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) @tparam AllocatorType type of the allocator to use (@c std::allocator by default) @note ObjectType trick from http://stackoverflow.com/a/9860911 @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, template class AllocatorType = std::allocator > class basic_json { public: ///////////////////// // container types // ///////////////////// /// the type of elements in a basic_json container using value_type = basic_json; /// the type of an element reference using reference = value_type&; /// the type of an element const reference using const_reference = const value_type&; /// 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; /// the allocator type using allocator_type = AllocatorType; /// the type of an element pointer using pointer = typename std::allocator_traits::pointer; /// the type of an element const pointer using const_pointer = typename std::allocator_traits::const_pointer; /// an iterator for a basic_json container class iterator; /// a const iterator for a basic_json container class const_iterator; /// a reverse iterator for a basic_json container class reverse_iterator; /// a const reverse iterator for a basic_json container class const_reverse_iterator; /// returns the allocator associated with the container static allocator_type get_allocator() { return allocator_type(); } /////////////////////////// // JSON value data types // /////////////////////////// /// a type for an object using object_t = ObjectType, AllocatorType>>; /// 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 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) discarded ///< (internal) indicates the parser callback chose not to keep the value }; //////////////////////// // 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; /// boolean 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() noexcept = default; /// constructor for booleans json_value(boolean_t v) noexcept : boolean(v) {} /// constructor for numbers (integer) json_value(number_integer_t v) noexcept : number_integer(v) {} /// constructor for numbers (floating-point) json_value(number_float_t v) noexcept : number_float(v) {} /// constructor for empty values of a given type json_value(value_t t) { switch (t) { case (value_t::null): case (value_t::discarded): { break; } case (value_t::object): { AllocatorType alloc; object = alloc.allocate(1); alloc.construct(object); break; } case (value_t::array): { AllocatorType alloc; array = alloc.allocate(1); alloc.construct(array); break; } case (value_t::string): { AllocatorType alloc; string = alloc.allocate(1); alloc.construct(string, ""); break; } case (value_t::boolean): { boolean = boolean_t(false); break; } case (value_t::number_integer): { number_integer = number_integer_t(0); break; } case (value_t::number_float): { number_float = number_float_t(0.0); break; } } } /// constructor for strings json_value(const string_t& value) { AllocatorType alloc; string = alloc.allocate(1); alloc.construct(string, value); } /// constructor for objects json_value(const object_t& value) { AllocatorType alloc; object = alloc.allocate(1); alloc.construct(object, value); } /// constructor for arrays json_value(const array_t& value) { AllocatorType alloc; array = alloc.allocate(1); alloc.construct(array, value); } }; ////////////////////////// // JSON parser callback // ////////////////////////// /// JSON callback event enumeration enum class parse_event_t : uint8_t { object_start, ///< start an object scope (found a '{' token) object_end, ///< end of an object scope (found '}' token) array_start, ///< start of an array scope (found '[' token) array_end, ///< end of an array scope (found ']' token) key, ///< found an object key within an object scope value ///< a value in an appropriate context (i.e., following a tag in an object scope) }; /// per-element parser callback type using parser_callback_t = std::function; /*! @brief comparison operator for JSON value types Returns an ordering that is similar to Python: - order: null < boolean < number < object < array < string - furthermore, each type is not smaller than itself */ friend bool operator<(const value_t lhs, const value_t rhs) { static constexpr std::array order = {{ 0, // null 3, // object 4, // array 5, // string 1, // boolean 2, // integer 2 // float } }; // discarded values are not comparable if (lhs == value_t::discarded or rhs == value_t::discarded) { return false; } return order[static_cast(lhs)] < order[static_cast(rhs)]; } ////////////////// // constructors // ////////////////// /*! @brief create an empty value with a given type @param value the type to create an value of @exception std::bad_alloc if allocation for object, array, or string fails. */ basic_json(const value_t value) : m_type(value), m_value(value) {} /*! @brief create a null object (implicitly) @ingroup container */ basic_json() noexcept = default; /// create a null object (explicitly) basic_json(std::nullptr_t) noexcept : basic_json(value_t::null) {} /// create an object (explicit) basic_json(const object_t& value) : m_type(value_t::object), m_value(value) {} /// create an object (implicit) template ::value and std::is_constructible::value, int>::type = 0> basic_json(const V& value) : m_type(value_t::object) { AllocatorType alloc; m_value.object = alloc.allocate(1); using std::begin; using std::end; alloc.construct(m_value.object, begin(value), end(value)); } /// create an array (explicit) basic_json(const array_t& value) : m_type(value_t::array), m_value(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> basic_json(const V& value) : m_type(value_t::array) { AllocatorType alloc; m_value.array = alloc.allocate(1); using std::begin; using std::end; alloc.construct(m_value.array, begin(value), end(value)); } /// create a string (explicit) basic_json(const string_t& value) : m_type(value_t::string), m_value(value) {} /// create a string (explicit) basic_json(const typename string_t::value_type* value) : basic_json(string_t(value)) {} /// create a string (implicit) template ::value, int>::type = 0> basic_json(const V& value) : basic_json(string_t(value)) {} /// create a boolean (explicit) basic_json(boolean_t value) : m_type(value_t::boolean), m_value(value) {} /*! @brief create an integer number (explicit) @tparam T helper type to compare number_integer_t and int @param value an integer to create a JSON number from This constructor takes care about explicitly passed values of type number_integer_t. However, this constructor would have the same signature as the existing one for const int values, so we need to switch this one off in case number_integer_t is the same as int. */ template::value) and std::is_same::value , int>::type = 0> basic_json(const number_integer_t value) : m_type(value_t::number_integer), m_value(value) {} /*! @brief create an int number to support enum type (implicit) @param value an integer to create a JSON number from This constructor allows to pass enums directly to a constructor. As C++ has no way of specifying the type of an anonymous enum explicitly, we can only rely on the fact that such values implicitly convert to int. As int may already be the same type of number_integer_t, we may need to switch off that constructor, which is done above. */ basic_json(const int value) : m_type(value_t::number_integer), m_value(static_cast(value)) {} /// create an integer number (implicit) template::value and std::numeric_limits::is_integer, T>::type = 0> basic_json(const T value) noexcept : m_type(value_t::number_integer), m_value(static_cast(value)) {} /// create a floating-point number (explicit) basic_json(const number_float_t value) : m_type(value_t::number_float), m_value(value) { // replace infinity and NAN by null if (not std::isfinite(value)) { m_type = value_t::null; m_value = json_value(); } } /// create a floating-point number (implicit) template::value and std::is_floating_point::value>::type > basic_json(const T value) noexcept : basic_json(number_float_t(value)) {} /// create a container (array or object) from an initializer list basic_json(list_init_t init, 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 : init) { if (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) { // 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::domain_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 = value_t::object; for (auto& element : init) { 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; AllocatorType alloc; m_value.array = alloc.allocate(1); alloc.construct(m_value.array, std::move(init)); } } /// explicitly create an array from an initializer list static basic_json array(list_init_t init = list_init_t()) { return basic_json(init, false, value_t::array); } /// explicitly create an object from an initializer list static basic_json object(list_init_t init = list_init_t()) { return basic_json(init, false, value_t::object); } /// construct an array with count copies of given value basic_json(size_type count, const basic_json& other) : m_type(value_t::array) { AllocatorType alloc; m_value.array = alloc.allocate(1); alloc.construct(m_value.array, count, other); } /// construct a JSON container given an iterator range template ::value or std::is_same::value , int>::type = 0> basic_json(T first, T last) { // make sure iterator fits the current value if (first.m_object != last.m_object or first.m_object->m_type != last.m_object->m_type) { throw std::domain_error("iterators are not compatible"); } // set the type m_type = first.m_object->m_type; // check if iterator range is complete for non-compound values switch (m_type) { case value_t::number_integer: case value_t::number_float: case value_t::boolean: case value_t::string: { if (first.m_it.generic_iterator != 0 or last.m_it.generic_iterator != 1) { throw std::out_of_range("iterators out of range"); } break; } default: { break; } } switch (m_type) { case value_t::number_integer: { m_value.number_integer = first.m_object->m_value.number_integer; break; } case value_t::number_float: { m_value.number_float = first.m_object->m_value.number_float; break; } case value_t::boolean: { m_value.boolean = first.m_object->m_value.boolean; break; } case value_t::string: { m_value = *first.m_object->m_value.string; break; } case value_t::object: { AllocatorType alloc; m_value.object = alloc.allocate(1); alloc.construct(m_value.object, first.m_it.object_iterator, last.m_it.object_iterator); break; } case value_t::array: { AllocatorType alloc; m_value.array = alloc.allocate(1); alloc.construct(m_value.array, first.m_it.array_iterator, last.m_it.array_iterator); break; } default: { throw std::domain_error("cannot use construct with iterators from " + first.m_object->type_name()); } } } /////////////////////////////////////// // other constructors and destructor // /////////////////////////////////////// /*! @brief copy constructor @exception std::bad_alloc if allocation for object, array, or string fails. @ingroup container */ basic_json(const basic_json& other) : m_type(other.m_type) { switch (m_type) { case (value_t::null): case (value_t::discarded): { break; } case (value_t::object): { m_value = *other.m_value.object; break; } case (value_t::array): { m_value = *other.m_value.array; break; } case (value_t::string): { m_value = *other.m_value.string; break; } case (value_t::boolean): { m_value = other.m_value.boolean; break; } case (value_t::number_integer): { m_value = other.m_value.number_integer; break; } case (value_t::number_float): { m_value = other.m_value.number_float; break; } } } /// move constructor 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 = {}; } /*! @brief copy assignment @ingroup container */ 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 ) { using std::swap; std::swap(m_type, other.m_type); std::swap(m_value, other.m_value); return *this; } /*! @brief destructor @ingroup container */ ~basic_json() noexcept { switch (m_type) { case (value_t::object): { AllocatorType alloc; alloc.destroy(m_value.object); alloc.deallocate(m_value.object, 1); m_value.object = nullptr; break; } case (value_t::array): { AllocatorType alloc; alloc.destroy(m_value.array); alloc.deallocate(m_value.array, 1); m_value.array = nullptr; break; } case (value_t::string): { AllocatorType alloc; alloc.destroy(m_value.string); alloc.deallocate(m_value.string, 1); 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 */ string_t dump(const int indent = -1) const noexcept { std::stringstream ss; if (indent >= 0) { dump(ss, true, static_cast(indent)); } else { dump(ss, false, 0); } return ss.str(); } /// return the type of the object (explicit) value_t type() const noexcept { return m_type; } // return whether value is null bool is_null() const noexcept { return m_type == value_t::null; } // return whether value is boolean bool is_boolean() const noexcept { return m_type == value_t::boolean; } // return whether value is number bool is_number() const noexcept { return (m_type == value_t::number_integer) or (m_type == value_t::number_float); } // return whether value an integer is number bool is_number_integer() const noexcept { return m_type == value_t::number_integer; } // return whether value is a floating-point number bool is_number_float() const noexcept { return m_type == value_t::number_float; } // return whether value is object bool is_object() const noexcept { return m_type == value_t::object; } // return whether value is array bool is_array() const noexcept { return m_type == value_t::array; } // return whether value is string bool is_string() const noexcept { return m_type == value_t::string; } // return whether value is discarded bool is_discarded() const noexcept { return m_type == value_t::discarded; } /// return the type of the object (implicit) operator value_t() const noexcept { return m_type; } private: ////////////////////// // value conversion // ////////////////////// /// get an object (explicit) template ::value and std::is_convertible::value , int>::type = 0> T get_impl(T*) const { switch (m_type) { case (value_t::object): { return T(m_value.object->begin(), m_value.object->end()); } default: { throw std::domain_error("cannot cast " + type_name() + " to " + typeid(T).name()); } } } /// get an object (explicit) object_t get_impl(object_t*) const { switch (m_type) { case (value_t::object): { return *(m_value.object); } default: { throw std::domain_error("cannot cast " + type_name() + " to object"); } } } /// get an array (explicit) template ::value and not std::is_same::value and not std::is_arithmetic::value and not std::is_convertible::value and not has_mapped_type::value , int>::type = 0> T get_impl(T*) const { switch (m_type) { case (value_t::array): { T to_vector; std::transform(m_value.array->begin(), m_value.array->end(), std::inserter(to_vector, to_vector.end()), [](basic_json i) { return i.get(); }); return to_vector; } default: { throw std::domain_error("cannot cast " + type_name() + " to " + typeid(T).name()); } } } /// get an array (explicit) template ::value and not std::is_same::value , int>::type = 0> std::vector get_impl(std::vector*) const { switch (m_type) { case (value_t::array): { std::vector to_vector; to_vector.reserve(m_value.array->size()); std::transform(m_value.array->begin(), m_value.array->end(), std::inserter(to_vector, to_vector.end()), [](basic_json i) { return i.get(); }); return to_vector; } default: { throw std::domain_error("cannot cast " + type_name() + " to " + typeid(T).name()); } } } /// get an array (explicit) template ::value and not has_mapped_type::value , int>::type = 0> T get_impl(T*) const { switch (m_type) { case (value_t::array): { return T(m_value.array->begin(), m_value.array->end()); } default: { throw std::domain_error("cannot cast " + type_name() + " to " + typeid(T).name()); } } } array_t get_impl(array_t*) const { switch (m_type) { case (value_t::array): { return *(m_value.array); } default: { throw std::domain_error("cannot cast " + type_name() + " to array"); } } } /// get a string (explicit) template ::value , int>::type = 0> T get_impl(T*) const { switch (m_type) { case (value_t::string): { return *m_value.string; } default: { throw std::domain_error("cannot cast " + type_name() + " to " + typeid(T).name()); } } } /// get a number (explicit) template::value , int>::type = 0> T get_impl(T*) 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::domain_error("cannot cast " + type_name() + " to " + typeid(T).name()); } } } /// get a boolean (explicit) boolean_t get_impl(boolean_t*) const { switch (m_type) { case (value_t::boolean): { return m_value.boolean; } default: { throw std::domain_error("cannot cast " + type_name() + " to " + typeid(boolean_t).name()); } } } public: /// get a value (explicit) // template T get() const { return get_impl(static_cast(nullptr)); } /// get a value (implicit) template operator T() const { return get(); } //////////////////// // element access // //////////////////// /// access specified element with bounds checking reference at(size_type idx) { // at only works for arrays if (m_type != value_t::array) { throw std::domain_error("cannot use at with " + type_name()); } return m_value.array->at(idx); } /// access specified element with bounds checking const_reference at(size_type idx) const { // at only works for arrays if (m_type != value_t::array) { throw std::domain_error("cannot use at with " + type_name()); } return m_value.array->at(idx); } /// access specified element with bounds checking reference at(const typename object_t::key_type& key) { // at only works for objects if (m_type != value_t::object) { throw std::domain_error("cannot use at with " + type_name()); } return m_value.object->at(key); } /// access specified element with bounds checking const_reference at(const typename object_t::key_type& key) const { // at only works for objects if (m_type != value_t::object) { throw std::domain_error("cannot use at with " + type_name()); } return m_value.object->at(key); } /// access specified element reference operator[](size_type idx) { // implicitly convert null to object if (m_type == value_t::null) { m_type = value_t::array; AllocatorType alloc; m_value.array = alloc.allocate(1); alloc.construct(m_value.array); } // [] only works for arrays if (m_type != value_t::array) { throw std::domain_error("cannot use [] with " + type_name()); } for (size_t i = m_value.array->size(); i <= idx; ++i) { m_value.array->push_back(basic_json()); } return m_value.array->operator[](idx); } /// access specified element const_reference operator[](size_type idx) const { // at only works for arrays if (m_type != value_t::array) { throw std::domain_error("cannot use [] with " + type_name()); } return m_value.array->operator[](idx); } /// access specified element reference operator[](const typename object_t::key_type& key) { // implicitly convert null to object if (m_type == value_t::null) { m_type = value_t::object; AllocatorType alloc; m_value.object = alloc.allocate(1); alloc.construct(m_value.object); } // [] only works for objects if (m_type != value_t::object) { throw std::domain_error("cannot use [] with " + type_name()); } return m_value.object->operator[](key); } /// access specified element const_reference operator[](const typename object_t::key_type& key) const { // at only works for objects if (m_type != value_t::object) { throw std::domain_error("cannot use [] with " + type_name()); } return m_value.object->operator[](key); } /// access specified element (needed for clang) template reference operator[](const T (&key)[n]) { // implicitly convert null to object if (m_type == value_t::null) { m_type = value_t::object; m_value = value_t::object; } // at only works for objects if (m_type != value_t::object) { throw std::domain_error("cannot use [] with " + type_name()); } return m_value.object->operator[](key); } /// access specified element (needed for clang) template const_reference operator[](const T (&key)[n]) const { // at only works for objects if (m_type != value_t::object) { throw std::domain_error("cannot use [] with " + type_name()); } return m_value.object->operator[](key); } /// access the first element reference front() { return *begin(); } /// access the first element const_reference front() const { return *cbegin(); } /// access the last element reference back() { auto tmp = end(); --tmp; return *tmp; } /// access the last element const_reference back() const { auto tmp = cend(); --tmp; return *tmp; } /// remove element given an iterator template ::value or std::is_same::value , int>::type = 0> T erase(T pos) { // make sure iterator fits the current value if (this != pos.m_object or m_type != pos.m_object->m_type) { throw std::domain_error("iterator does not fit current value"); } T result = end(); switch (m_type) { case value_t::number_integer: case value_t::number_float: case value_t::boolean: case value_t::string: { if (pos.m_it.generic_iterator != 0) { throw std::out_of_range("iterator out of range"); } if (m_type == value_t::string) { delete m_value.string; m_value.string = nullptr; } m_type = value_t::null; break; } case value_t::object: { result.m_it.object_iterator = m_value.object->erase(pos.m_it.object_iterator); break; } case value_t::array: { result.m_it.array_iterator = m_value.array->erase(pos.m_it.array_iterator); break; } default: { throw std::domain_error("cannot use erase with " + type_name()); } } return result; } /// remove elements given an iterator range template ::value or std::is_same::value , int>::type = 0> T erase(T first, T last) { // make sure iterator fits the current value if (this != first.m_object or this != last.m_object or m_type != first.m_object->m_type or m_type != last.m_object->m_type) { throw std::domain_error("iterators do not fit current value"); } T result = end(); switch (m_type) { case value_t::number_integer: case value_t::number_float: case value_t::boolean: case value_t::string: { if (first.m_it.generic_iterator != 0 or last.m_it.generic_iterator != 1) { throw std::out_of_range("iterators out of range"); } if (m_type == value_t::string) { delete m_value.string; m_value.string = nullptr; } m_type = value_t::null; break; } case value_t::object: { result.m_it.object_iterator = m_value.object->erase(first.m_it.object_iterator, last.m_it.object_iterator); break; } case value_t::array: { result.m_it.array_iterator = m_value.array->erase(first.m_it.array_iterator, last.m_it.array_iterator); break; } default: { throw std::domain_error("cannot use erase with " + type_name()); } } return result; } /// remove element from an object given a key size_type erase(const typename object_t::key_type& key) { // this erase only works for objects if (m_type != value_t::object) { throw std::domain_error("cannot use erase with " + type_name()); } return m_value.object->erase(key); } /// remove element from an array given an index void erase(const size_type idx) { // this erase only works for arrays if (m_type != value_t::array) { throw std::domain_error("cannot use erase with " + type_name()); } if (idx >= size()) { throw std::out_of_range("index out of range"); } m_value.array->erase(m_value.array->begin() + static_cast(idx)); } /// find an element in an object 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 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; } /// returns the number of occurrences of a key in an object size_type count(typename object_t::key_type key) const { // return 0 for all nonobject types return (m_type == value_t::object) ? m_value.object->count(key) : 0; } /////////////// // iterators // /////////////// /*! @brief returns an iterator to the first element @ingroup container */ iterator begin() noexcept { iterator result(this); result.set_begin(); return result; } /*! @brief returns a const iterator to the first element @ingroup container */ const_iterator begin() const noexcept { return cbegin(); } /*! @brief returns a const iterator to the first element @ingroup container */ const_iterator cbegin() const noexcept { const_iterator result(this); result.set_begin(); return result; } /*! @brief returns an iterator to one past the last element @ingroup container */ iterator end() noexcept { iterator result(this); result.set_end(); return result; } /*! @brief returns a const iterator to one past the last element @ingroup container */ const_iterator end() const noexcept { return cend(); } /*! @brief returns a const iterator to one past the last element @ingroup container */ const_iterator cend() const noexcept { const_iterator result(this); result.set_end(); return result; } /*! @brief returns a reverse iterator to the first element @ingroup reversiblecontainer */ reverse_iterator rbegin() noexcept { return reverse_iterator(end()); } /*! @brief returns a const reverse iterator to the first element @ingroup reversiblecontainer */ const_reverse_iterator rbegin() const noexcept { return crbegin(); } /*! @brief returns a reverse iterator to one past the last element @ingroup reversiblecontainer */ reverse_iterator rend() noexcept { return reverse_iterator(begin()); } /*! @brief returns a const reverse iterator to one past the last element @ingroup reversiblecontainer */ const_reverse_iterator rend() const noexcept { return crend(); } /*! @brief returns a const reverse iterator to the first element @ingroup reversiblecontainer */ const_reverse_iterator crbegin() const noexcept { return const_reverse_iterator(cend()); } /*! @brief returns a const reverse iterator to one past the last element @ingroup reversiblecontainer */ const_reverse_iterator crend() const noexcept { return const_reverse_iterator(cbegin()); } ////////////// // capacity // ////////////// /*! @brief checks whether the container is empty @ingroup container */ 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; } } } /*! @brief returns the number of elements @ingroup container */ 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; } } } /*! @brief returns the maximum possible number of elements @ingroup container */ 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 void clear() noexcept { switch (m_type) { case (value_t::null): case (value_t::discarded): { 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 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::domain_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 = value_t::array; } // 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 reference operator+=(basic_json&& value) { push_back(std::move(value)); return *this; } /// add an object to an array 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::domain_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 = value_t::array; } // add element to array m_value.array->push_back(value); } /// add an object to an array reference operator+=(const basic_json& value) { push_back(value); return *this; } /// add an object to an object 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::domain_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 = value_t::object; } // add element to array m_value.object->insert(value); } /// add an object to an object reference operator+=(const typename object_t::value_type& value) { push_back(value); return operator[](value.first); } /*! @brief exchanges the values @ingroup container */ 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 void swap(array_t& other) { // swap only works for arrays if (m_type != value_t::array) { throw std::domain_error("cannot use swap with " + type_name()); } // swap arrays std::swap(*(m_value.array), other); } /// swaps the contents void swap(object_t& other) { // swap only works for objects if (m_type != value_t::object) { throw std::domain_error("cannot use swap with " + type_name()); } // swap arrays std::swap(*(m_value.object), other); } /// swaps the contents void swap(string_t& other) { // swap only works for strings if (m_type != value_t::string) { throw std::domain_error("cannot use swap with " + type_name()); } // swap arrays std::swap(*(m_value.string), other); } ////////////////////////////////////////// // lexicographical comparison operators // ////////////////////////////////////////// /*! @brief comparison: equal @ingroup container */ friend bool operator==(const_reference lhs, const_reference rhs) noexcept { const auto lhs_type = lhs.type(); const auto rhs_type = rhs.type(); if (lhs_type == rhs_type) { switch (lhs_type) { case (value_t::array): return *lhs.m_value.array == *rhs.m_value.array; case (value_t::object): return *lhs.m_value.object == *rhs.m_value.object; case (value_t::null): return true; case (value_t::string): return *lhs.m_value.string == *rhs.m_value.string; case (value_t::boolean): return lhs.m_value.boolean == rhs.m_value.boolean; case (value_t::number_integer): return lhs.m_value.number_integer == rhs.m_value.number_integer; case (value_t::number_float): return approx(lhs.m_value.number_float, rhs.m_value.number_float); case (value_t::discarded): return false; } } else if (lhs_type == value_t::number_integer and rhs_type == value_t::number_float) { return approx(static_cast(lhs.m_value.number_integer), rhs.m_value.number_float); } else if (lhs_type == value_t::number_float and rhs_type == value_t::number_integer) { return approx(lhs.m_value.number_float, static_cast(rhs.m_value.number_integer)); } return false; } /*! @brief comparison: not equal @ingroup container */ friend bool operator!=(const_reference lhs, const_reference rhs) noexcept { return not (lhs == rhs); } /// comparison: less than friend bool operator<(const_reference lhs, const_reference rhs) noexcept { const auto lhs_type = lhs.type(); const auto rhs_type = rhs.type(); if (lhs_type == rhs_type) { switch (lhs_type) { case (value_t::array): return *lhs.m_value.array < *rhs.m_value.array; case (value_t::object): return *lhs.m_value.object < *rhs.m_value.object; case (value_t::null): return false; case (value_t::string): return *lhs.m_value.string < *rhs.m_value.string; case (value_t::boolean): return lhs.m_value.boolean < rhs.m_value.boolean; case (value_t::number_integer): return lhs.m_value.number_integer < rhs.m_value.number_integer; case (value_t::number_float): return lhs.m_value.number_float < rhs.m_value.number_float; case (value_t::discarded): return false; } } else if (lhs_type == value_t::number_integer and rhs_type == value_t::number_float) { return static_cast(lhs.m_value.number_integer) < rhs.m_value.number_float; } else if (lhs_type == value_t::number_float and rhs_type == value_t::number_integer) { return lhs.m_value.number_float < static_cast(rhs.m_value.number_integer); } // We only reach this line if we cannot compare values. In that case, // we compare types. return lhs_type < rhs_type; } /// comparison: less than or equal friend bool operator<=(const_reference lhs, const_reference rhs) noexcept { return not (rhs < lhs); } /// comparison: greater than friend bool operator>(const_reference lhs, const_reference rhs) noexcept { return not (lhs <= rhs); } /// comparison: greater than or equal friend bool operator>=(const_reference lhs, const_reference rhs) noexcept { 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 bool pretty_print = (o.width() > 0); const auto indentation = (pretty_print ? o.width() : 0); // reset width to 0 for subsequent calls to this stream o.width(0); // do the actual serialization j.dump(o, pretty_print, static_cast(indentation)); return o; } /// serialize to stream friend std::ostream& operator>>(const basic_json& j, std::ostream& o) { return o << j; } ///////////////////// // deserialization // ///////////////////// /// deserialize from string static basic_json parse(const string_t& s, parser_callback_t cb = nullptr) { return parser(s, cb).parse(); } /// deserialize from stream static basic_json parse(std::istream& i, parser_callback_t cb = nullptr) { return parser(i, cb).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 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"; } case (value_t::discarded): { return "discarded"; } 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 o the stream to write the escaped string to @param s the string to escape */ static void escape_string(std::ostream& o, const string_t& s) noexcept { for (const auto c : s) { switch (c) { // quotation mark (0x22) case '"': { o << "\\\""; break; } // reverse solidus (0x5c) case '\\': { o << "\\\\"; break; } // backspace (0x08) case '\b': { o << "\\b"; break; } // formfeed (0x0c) case '\f': { o << "\\f"; break; } // newline (0x0a) case '\n': { o << "\\n"; break; } // carriage return (0x0d) case '\r': { o << "\\r"; break; } // horizontal tab (0x09) case '\t': { o << "\\t"; break; } default: { if (c >= 0 and c <= 0x1f) { // control characters (everything between 0x00 and 0x1f) // -> create four-digit hex representation o << "\\u" << std::hex << std::setw(4) << std::setfill('0') << int(c); } else { // all other characters are added as-is o << c; } break; } } } } /*! @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() - integer numbers are converted implictly via operator<< - floating-point numbers are converted to a string using "%g" format @param o stream to write to @param pretty_print whether the output shall be pretty-printed @param indent_step the indent level @param current_indent the current indent level (only used internally) */ void dump(std::ostream& o, const bool pretty_print, const unsigned int indent_step, const unsigned int current_indent = 0) const noexcept { // variable to hold indentation for recursive calls unsigned int new_indent = current_indent; switch (m_type) { case (value_t::object): { if (m_value.object->empty()) { o << "{}"; return; } o << "{"; // increase indentation if (pretty_print) { new_indent += indent_step; o << "\n"; } for (auto i = m_value.object->cbegin(); i != m_value.object->cend(); ++i) { if (i != m_value.object->cbegin()) { o << (pretty_print ? ",\n" : ","); } o << string_t(new_indent, ' ') << "\""; escape_string(o, i->first); o << "\":" << (pretty_print ? " " : ""); i->second.dump(o, pretty_print, indent_step, new_indent); } // decrease indentation if (pretty_print) { new_indent -= indent_step; o << "\n"; } o << string_t(new_indent, ' ') + "}"; return; } case (value_t::array): { if (m_value.array->empty()) { o << "[]"; return; } o << "["; // increase indentation if (pretty_print) { new_indent += indent_step; o << "\n"; } for (auto i = m_value.array->cbegin(); i != m_value.array->cend(); ++i) { if (i != m_value.array->cbegin()) { o << (pretty_print ? ",\n" : ","); } o << string_t(new_indent, ' '); i->dump(o, pretty_print, indent_step, new_indent); } // decrease indentation if (pretty_print) { new_indent -= indent_step; o << "\n"; } o << string_t(new_indent, ' ') << "]"; return; } case (value_t::string): { o << string_t("\""); escape_string(o, *m_value.string); o << "\""; return; } case (value_t::boolean): { o << (m_value.boolean ? "true" : "false"); return; } case (value_t::number_integer): { o << m_value.number_integer; return; } case (value_t::number_float): { // 15 digits of precision allows round-trip IEEE 754 // string->double->string; to be safe, we read this value from // std::numeric_limits::digits10 o << std::setprecision(std::numeric_limits::digits10) << m_value.number_float; return; } case (value_t::discarded): { o << ""; return; } default: { o << "null"; return; } } } /// "equality" comparison for floating point numbers template static bool approx(const T a, const T b) { return not (a > b or a < b); } private: ////////////////////// // member variables // ////////////////////// /// the type of the current element value_t m_type = value_t::null; /// the value of the current element json_value m_value = {}; private: /////////////// // iterators // /////////////// /// an iterator value union internal_iterator { /// iterator for JSON objects typename object_t::iterator object_iterator; /// iterator for JSON arrays typename array_t::iterator array_iterator; /// generic iterator for all other value types difference_type generic_iterator; /// default constructor internal_iterator() : generic_iterator(-1) {} }; public: /// a random access iterator for the basic_json class class iterator : public std::iterator { // allow basic_json class to access m_it friend class basic_json; public: /// the type of the values when the iterator is dereferenced using value_type = typename basic_json::value_type; /// a type to represent differences between iterators using difference_type = typename basic_json::difference_type; /// defines a pointer to the type iterated over (value_type) using pointer = typename basic_json::pointer; /// defines a reference to the type iterated over (value_type) using reference = typename basic_json::reference; /// the category of the iterator using iterator_category = std::bidirectional_iterator_tag; /// default constructor iterator() = default; /// constructor for a given JSON instance iterator(pointer object) noexcept : 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 = -1; break; } } } /// copy constructor iterator(const iterator& other) noexcept : m_object(other.m_object), m_it(other.m_it) {} /// copy assignment iterator& operator=(iterator 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_object, other.m_object); std::swap(m_it, other.m_it); return *this; } private: /// set the iterator to the first value 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 = 1; break; } default: { m_it.generic_iterator = 0; break; } } } /// set the iterator past the last value 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 = 1; break; } } } public: /// return a reference to the value pointed to by the iterator reference operator*() { 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 == 0) { return *m_object; } else { throw std::out_of_range("cannot get value"); } } } } /// dereference the iterator pointer operator->() { 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 == 0) { return m_object; } else { throw std::out_of_range("cannot get value"); } } } } /// post-increment (it++) iterator operator++(int) { auto 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: { m_it.generic_iterator++; break; } } return result; } /// pre-increment (++it) 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: { ++m_it.generic_iterator; break; } } return *this; } /// post-decrement (it--) iterator operator--(int) { auto 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: { m_it.generic_iterator--; break; } } return result; } /// pre-decrement (--it) 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: { --m_it.generic_iterator; break; } } return *this; } /// comparison: equal bool operator==(const iterator& other) const { // if objects are not the same, the comparison is undefined if (m_object != other.m_object) { throw std::domain_error("cannot compare iterators of different containers"); } 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 bool operator!=(const iterator& other) const { return not operator==(other); } /// comparison: smaller bool operator<(const iterator& other) const { // if objects are not the same, the comparison is undefined if (m_object != other.m_object) { throw std::domain_error("cannot compare iterators of different containers"); } switch (m_object->m_type) { case (basic_json::value_t::object): { throw std::domain_error("cannot use operator< for object iterators"); } 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: less than or equal bool operator<=(const iterator& other) const { return not other.operator < (*this); } /// comparison: greater than bool operator>(const iterator& other) const { return not operator<=(other); } /// comparison: greater than or equal bool operator>=(const iterator& other) const { return not operator<(other); } /// add to iterator iterator& operator+=(difference_type i) { switch (m_object->m_type) { case (basic_json::value_t::object): { throw std::domain_error("cannot use operator+= for object iterators"); } case (basic_json::value_t::array): { m_it.array_iterator += i; break; } default: { m_it.generic_iterator += i; break; } } return *this; } /// subtract from iterator iterator& operator-=(difference_type i) { return operator+=(-i); } /// add to iterator iterator operator+(difference_type i) { auto result = *this; result += i; return result; } /// subtract from iterator iterator operator-(difference_type i) { auto result = *this; result -= i; return result; } /// return difference difference_type operator-(const iterator& other) const { switch (m_object->m_type) { case (basic_json::value_t::object): { throw std::domain_error("cannot use operator- for object iterators"); return 0; } 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; } } } /// access to successor reference operator[](difference_type n) { switch (m_object->m_type) { case (basic_json::value_t::object): { throw std::domain_error("cannot use operator[] for object iterators"); } case (basic_json::value_t::array): { return *(m_it.array_iterator + n); } case (basic_json::value_t::null): { throw std::out_of_range("cannot get value"); } default: { if (m_it.generic_iterator == -n) { return *m_object; } else { throw std::out_of_range("cannot get value"); } } } } /// return the key of an object iterator typename object_t::key_type key() const { switch (m_object->m_type) { case (basic_json::value_t::object): { return m_it.object_iterator->first; } default: { throw std::domain_error("cannot use key() for non-object iterators"); } } } /// return the key of an iterator reference value() { return operator*(); } private: /// associated JSON instance pointer m_object = nullptr; /// the actual iterator of the associated instance internal_iterator m_it; }; /// a const random access iterator for the basic_json class class const_iterator : public std::iterator { // allow basic_json class to access m_it friend class basic_json; public: /// the type of the values when the iterator is dereferenced using value_type = typename basic_json::value_type; /// a type to represent differences between iterators using difference_type = typename basic_json::difference_type; /// defines a pointer to the type iterated over (value_type) using pointer = typename basic_json::const_pointer; /// defines a reference to the type iterated over (value_type) using reference = typename basic_json::const_reference; /// the category of the iterator using iterator_category = std::bidirectional_iterator_tag; /// default constructor const_iterator() = default; /// constructor for a given JSON instance const_iterator(pointer object) noexcept : 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 = -1; break; } } } /// copy constructor given a nonconst iterator const_iterator(const iterator& other) noexcept : 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 constructor const_iterator(const const_iterator& other) noexcept : m_object(other.m_object), m_it(other.m_it) {} /// copy assignment const_iterator& operator=(const_iterator 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_object, other.m_object); std::swap(m_it, other.m_it); return *this; } private: /// set the iterator to the first value 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 = 1; break; } default: { m_it.generic_iterator = 0; break; } } } /// set the iterator past the last value 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 = 1; break; } } } public: /// return a reference to the value pointed to by the iterator 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 == 0) { return *m_object; } else { throw std::out_of_range("cannot get value"); } } } } /// dereference the iterator 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 == 0) { return m_object; } else { throw std::out_of_range("cannot get value"); } } } } /// post-increment (it++) const_iterator operator++(int) { auto 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: { m_it.generic_iterator++; break; } } return result; } /// pre-increment (++it) 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: { ++m_it.generic_iterator; break; } } return *this; } /// post-decrement (it--) const_iterator operator--(int) { auto 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: { m_it.generic_iterator--; break; } } return result; } /// pre-decrement (--it) 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: { --m_it.generic_iterator; break; } } return *this; } /// comparison: equal bool operator==(const const_iterator& other) const { // if objects are not the same, the comparison is undefined if (m_object != other.m_object) { throw std::domain_error("cannot compare iterators of different containers"); } 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 bool operator!=(const const_iterator& other) const { return not operator==(other); } /// comparison: smaller bool operator<(const const_iterator& other) const { // if objects are not the same, the comparison is undefined if (m_object != other.m_object) { throw std::domain_error("cannot compare iterators of different containers"); } switch (m_object->m_type) { case (basic_json::value_t::object): { throw std::domain_error("cannot use operator< for object iterators"); } 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: less than or equal bool operator<=(const const_iterator& other) const { return not other.operator < (*this); } /// comparison: greater than bool operator>(const const_iterator& other) const { return not operator<=(other); } /// comparison: greater than or equal bool operator>=(const const_iterator& other) const { return not operator<(other); } /// add to iterator const_iterator& operator+=(difference_type i) { switch (m_object->m_type) { case (basic_json::value_t::object): { throw std::domain_error("cannot use operator+= for object iterators"); } case (basic_json::value_t::array): { m_it.array_iterator += i; break; } default: { m_it.generic_iterator += i; break; } } return *this; } /// subtract from iterator const_iterator& operator-=(difference_type i) { return operator+=(-i); } /// add to iterator const_iterator operator+(difference_type i) { auto result = *this; result += i; return result; } /// subtract from iterator const_iterator operator-(difference_type i) { auto result = *this; result -= i; return result; } /// return difference difference_type operator-(const const_iterator& other) const { switch (m_object->m_type) { case (basic_json::value_t::object): { throw std::domain_error("cannot use operator- for object iterators"); } 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; } } } /// access to successor reference operator[](difference_type n) const { switch (m_object->m_type) { case (basic_json::value_t::object): { throw std::domain_error("cannot use operator[] for object iterators"); } case (basic_json::value_t::array): { return *(m_it.array_iterator + n); } case (basic_json::value_t::null): { throw std::out_of_range("cannot get value"); } default: { if (m_it.generic_iterator == -n) { return *m_object; } else { throw std::out_of_range("cannot get value"); } } } } /// return the key of an object iterator typename object_t::key_type key() const { switch (m_object->m_type) { case (basic_json::value_t::object): { return m_it.object_iterator->first; } default: { throw std::domain_error("cannot use key() for non-object iterators"); } } } /// return the value of an iterator reference value() const { return operator*(); } private: /// associated JSON instance pointer m_object = nullptr; /// the actual iterator of the associated instance internal_iterator m_it; }; /// a reverse random access iterator for the basic_json class class reverse_iterator : private std::reverse_iterator { public: reverse_iterator(const typename std::reverse_iterator::iterator_type& it) : std::reverse_iterator(it) {} /// return the key of an object iterator typename object_t::key_type key() const { return this->base().key(); } /// return the value of an iterator reference value() const { return this->base().operator * (); } }; /// a const reverse random access iterator for the basic_json class class const_reverse_iterator : private std::reverse_iterator { public: const_reverse_iterator(const typename std::reverse_iterator::iterator_type& it) : std::reverse_iterator(it) {} /// return the key of an object iterator typename object_t::key_type key() const { return this->base().key(); } /// return the value of an iterator const_reference value() const { return this->base().operator * (); } }; 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 = unsigned char; /// constructor with a given buffer lexer(const string_t& s) noexcept : m_stream(nullptr), m_buffer(s) { m_content = reinterpret_cast(s.c_str()); m_start = m_cursor = m_content; m_limit = m_content + s.size(); } lexer(std::istream* s) noexcept : m_stream(s) { getline(*m_stream, m_buffer); m_content = reinterpret_cast(m_buffer.c_str()); m_start = m_cursor = m_content; m_limit = m_content + m_buffer.size(); } /// default constructor lexer() = default; /*! @brief create a string from a Unicode code point @param codepoint1 the code point (can be high surrogate) @param codepoint2 the code point (can be low surrogate or 0) @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 */ static string_t to_unicode(const std::size_t codepoint1, const std::size_t codepoint2 = 0) { string_t result; // calculate the codepoint from the given code points std::size_t codepoint = codepoint1; // check if codepoint1 is a high surrogate if (codepoint1 >= 0xD800 and codepoint1 <= 0xDBFF) { // check if codepoint2 is a low surrogate 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 < 0x80) { // 1-byte characters: 0xxxxxxx (ASCII) 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 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 */ token_type scan() noexcept { // pointer for backtracking information 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, 0, 0, 0, 0, 0, 0, 0, 0, 32, 32, 0, 0, 32, 0, 0, 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, }; if ((m_limit - m_cursor) < 5) { yyfill(); // LCOV_EXCL_LINE; } 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; if (m_limit <= m_cursor) { yyfill(); // LCOV_EXCL_LINE; } 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 <= 0x0F) { 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; if (m_limit <= m_cursor) { yyfill(); // LCOV_EXCL_LINE; } yych = *m_cursor; basic_json_parser_31: if (yybm[0 + yych] & 64) { goto basic_json_parser_30; } if (yych <= 0x0F) { 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; if (m_limit <= m_cursor) { yyfill(); // LCOV_EXCL_LINE; } 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; if (m_limit <= m_cursor) { yyfill(); // LCOV_EXCL_LINE; } 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; if (m_limit <= m_cursor) { yyfill(); // LCOV_EXCL_LINE; } 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; if (m_limit <= m_cursor) { yyfill(); // LCOV_EXCL_LINE; } 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; if (m_limit <= m_cursor) { yyfill(); // LCOV_EXCL_LINE; } 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; if ((m_limit - m_cursor) < 3) { yyfill(); // LCOV_EXCL_LINE; } 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; if (m_limit <= m_cursor) { yyfill(); // LCOV_EXCL_LINE; } 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; if ((m_limit - m_cursor) < 3) { yyfill(); // LCOV_EXCL_LINE; } 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; } } } /// append data from the stream to the internal buffer void yyfill() noexcept { if (not m_stream or not * m_stream) { return; } const ssize_t offset_start = m_start - m_content; const ssize_t offset_marker = m_marker - m_start; const ssize_t offset_cursor = m_cursor - m_start; m_buffer.erase(0, static_cast(offset_start)); std::string line; std::getline(*m_stream, line); m_buffer += "\n" + line; // add line with newline symbol m_content = reinterpret_cast(m_buffer.c_str()); m_start = m_content; m_marker = m_start + offset_marker; m_cursor = m_start + offset_cursor; m_limit = m_start + m_buffer.size() - 1; } /// return string representation of last read token string_t get_token() const noexcept { return string_t(reinterpret_cast(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 */ 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 lexer_char_t* 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; } case '\\': { result += "\\"; break; } case '/': { result += "/"; break; } case '"': { result += "\""; break; } // unicode case 'u': { // get code xxxx from uxxxx auto codepoint = std::strtoul(std::string(reinterpret_cast(i + 1), 4).c_str(), nullptr, 16); // check if codepoint is a high surrogate 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(reinterpret_cast (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, static_cast(*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 */ long double get_number() const { // conversion typename string_t::value_type* endptr; const auto float_val = std::strtold(reinterpret_cast(m_start), &endptr); // return float_val if the whole number was translated and NAN // otherwise return (reinterpret_cast(endptr) == m_cursor) ? float_val : NAN; } private: /// optional input stream std::istream* m_stream; /// the buffer string_t m_buffer; /// the buffer pointer const lexer_char_t* m_content = nullptr; /// pointer to the beginning of the current symbol const lexer_char_t* m_start = nullptr; /// pointer for backtracking information const lexer_char_t* m_marker = 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; }; /*! @brief syntax analysis */ class parser { public: /// constructor for strings parser(const string_t& s, parser_callback_t cb = nullptr) : callback(cb), m_lexer(s) { // read first token get_token(); } /// a parser reading from an input stream parser(std::istream& _is, parser_callback_t cb = nullptr) : callback(cb), m_lexer(&_is) { // read first token get_token(); } /// public parser interface basic_json parse() { basic_json result = parse_internal(true); expect(lexer::token_type::end_of_input); return result; } private: /// the actual parser basic_json parse_internal(bool keep) { auto result = basic_json(value_t::discarded); switch (last_token) { case (lexer::token_type::begin_object): { if (keep and (not callback or (keep = callback(depth++, parse_event_t::object_start, result)))) { // explicitly set result to object to cope with {} result.m_type = value_t::object; result.m_value = json_value(value_t::object); } // read next token get_token(); // closing } -> we are done if (last_token == lexer::token_type::end_object) { get_token(); if (keep and callback and not callback(--depth, parse_event_t::object_end, result)) { result = basic_json(value_t::discarded); } return result; } // no comma is expected here unexpect(lexer::token_type::value_separator); // 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(); bool keep_tag = false; if (keep) { keep_tag = callback ? callback(depth, parse_event_t::key, basic_json(key)) : true; } // parse separator (:) get_token(); expect(lexer::token_type::name_separator); // parse and add value get_token(); auto value = parse_internal(keep); if (keep and keep_tag and not value.is_discarded()) { result[key] = std::move(value); } } while (last_token == lexer::token_type::value_separator); // closing } expect(lexer::token_type::end_object); get_token(); if (keep and callback and not callback(--depth, parse_event_t::object_end, result)) { result = basic_json(value_t::discarded); } return result; } case (lexer::token_type::begin_array): { if (keep and (not callback or (keep = callback(depth++, parse_event_t::array_start, result)))) { // explicitly set result to object to cope with [] result.m_type = value_t::array; result.m_value = json_value(value_t::array); } // read next token get_token(); // closing ] -> we are done if (last_token == lexer::token_type::end_array) { get_token(); if (callback and not callback(--depth, parse_event_t::array_end, result)) { result = basic_json(value_t::discarded); } return result; } // no comma is expected here unexpect(lexer::token_type::value_separator); // otherwise: parse values do { // ugly, but could be fixed with loop reorganization if (last_token == lexer::token_type::value_separator) { get_token(); } // parse value auto value = parse_internal(keep); if (keep and not value.is_discarded()) { result.push_back(std::move(value)); } } while (last_token == lexer::token_type::value_separator); // closing ] expect(lexer::token_type::end_array); get_token(); if (keep and callback and not callback(--depth, parse_event_t::array_end, result)) { result = basic_json(value_t::discarded); } return result; } case (lexer::token_type::literal_null): { get_token(); result.m_type = value_t::null; break; } case (lexer::token_type::value_string): { const auto s = m_lexer.get_string(); get_token(); result = basic_json(s); break; } case (lexer::token_type::literal_true): { get_token(); result.m_type = value_t::boolean; result.m_value = true; break; } case (lexer::token_type::literal_false): { get_token(); result.m_type = value_t::boolean; result.m_value = false; break; } 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 (approx(float_val, static_cast(int_val))) { // we basic_json not lose precision -> return int result.m_type = value_t::number_integer; result.m_value = int_val; } else { // we would lose precision -> returnfloat result.m_type = value_t::number_float; result.m_value = static_cast(float_val); } break; } default: { // the last token was unexpected unexpect(last_token); } } if (keep and callback and not callback(depth, parse_event_t::value, result)) { result = basic_json(value_t::discarded); } return result; } /// get next token from lexer typename lexer::token_type get_token() { last_token = m_lexer.scan(); return last_token; } 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); } } void unexpect(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 += "\' ("; error_msg += lexer::token_type_name(last_token) + ")"; throw std::invalid_argument(error_msg); } } private: /// levels of recursion int depth = 0; /// callback function parser_callback_t callback; /// 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 { /*! @brief exchanges the values of two JSON objects @ingroup container */ 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 std::size_t operator()(const nlohmann::json& j) const { // a naive hashing via the string representation const auto& h = hash(); return h(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(reinterpret_cast (const_cast(s))); } #endif