gplib/include/gp/enveloppe/cbor.hpp

#pragma once

#include <gp/algorithm/repeat.hpp>
#include <gp/bitops.hpp>
#include <gp/optional.hpp>
#include <gp/pair.hpp>
#include <gp/variant.hpp>
#include <gp/vector.hpp>


namespace gp {
	enum class cbor_type : uint8_t {
		uint = 0,
		nint = 1,
		bstr = 2,
		tstr = 3,
		list = 4,
		hmap = 5,
		tags = 6,
		oths = 7
	};

	enum class cbor_oths : uint8_t {
		value_false = 20,
		value_true = 21,
		value_null = 22,
		value_undefined = 23,
		byte = 24,
		word = 25,
		dword = 26,
		qword = 27,
		terminator = 31
	};

	enum class cbor_tags {
		datetime = 0,
		unix_time = 1,
		ubignum = 2,
		nbignum = 3,
		decimal = 4,
		bigfloat = 5,
		cose_encrypt0 = 16,
		cose_mac0 = 17,
		cose_sign1 = 18,
		expected_base64url = 21,
		expected_base64 = 22,
		expected_base16 = 23,
		encoded_cbor = 24,
		url = 32,
		base64url = 33,
		base64 = 34,
		regexp = 35,
		mime = 36,
		cose_encrypt = 96,
		cose_mac = 97,
		cose_sign = 98,
		signature = 55799
	};

	struct cbor_number final {
		bool sign;
		uint64_t value;

		bool is_negative() {
			return sign;
		}

		cbor_number(int64_t v) 
		: sign{v < 0}
		, value{uint64_t((sign ? -1 : 1) * v)}
		{}

		cbor_number(bool s, uint64_t v) 
		: sign{s}
		, value{v}
		{}
	};

	struct ieee754_hf final {
		uint16_t sign : 1;
		uint16_t exponent : 5;
		uint16_t mantissa : 10;

		// TODO: support for denormalized values and NaNs
		operator float() {
			auto a = (uint32_t)((sign<<16) | ((exponent+0x1C000)<<13) | (mantissa<<13));
			return *(float*)&a;
		}

		operator double() {
			return (float)*this;
		}
	};

	struct cbor_floating_point final {
		using backing = gp::fixed_variant<
			ieee754_hf,
			float,
			double
		>;
		backing contents;
		
		cbor_floating_point(backing& p)
		: contents{p}{}
		
		cbor_floating_point(ieee754_hf p)
		: contents{p}{}
		
		cbor_floating_point(float p)
		: contents{p}{}
		
		cbor_floating_point(double p)
		: contents{p}{}

		operator ieee754_hf() {
			return contents.value<ieee754_hf>();
		}

		operator float() {
			switch(contents.type()) {
				case backing::alt<ieee754_hf>():
					return contents.value<ieee754_hf>();
				case backing::alt<float>():
					return contents.value<float>();
				default:
					gp_config::assertion(false,"this code should never be reached");
					return std::numeric_limits<float>::quiet_NaN();
			}
		}

		operator double() {
			switch(contents.type()) {
				case backing::alt<ieee754_hf>():
					return contents.value<ieee754_hf>();
				case backing::alt<float>():
					return contents.value<float>();
				case backing::alt<double>():
					return contents.value<double>();
				default:
					gp_config::assertion(false,"this code should never be reached");
					return std::numeric_limits<double>::quiet_NaN();
			}
		}
	};

	struct undefined_t final {};

	template<typename T>
	using cbor_composite = gp::fixed_variant<
		gp::undefined_t,
		cbor_number,
		gp::vector<std::byte>,
		bool,
		gp::vector<T>,
		gp::vector<gp::pair<T, T>>,
		gp::nullopt_t,
		cbor_floating_point
	>;

	class cbor_value {

		cbor_composite<cbor_value> contents;
		gp::reference_wrapper<allocator> alloc;
	public:

		cbor_value(allocator& alloc_v)
		: contents()
		, alloc(alloc_v)
		{}

		cbor_value(cbor_number number, allocator& alloc_v)
		: contents(number)
		, alloc(alloc_v)
		{}

		cbor_value(cbor_composite<cbor_value> val, allocator& alloc_v)
		: contents(val)
		, alloc(alloc_v)
		{}

		cbor_value(const cbor_value& oth)
		: contents(oth.contents)
		, alloc(oth.alloc)
		{}

		cbor_value(cbor_value&& oth)
		: contents(gp::move(oth.contents))
		, alloc(gp::move(oth.alloc))
		{}

		cbor_value& operator=(cbor_value& value) {
			contents = value.contents;
			alloc = value.alloc;
			return *this;
		}

		cbor_value& operator=(cbor_value&& value) {
			gp::swap(contents, value.contents);
			gp::swap(alloc, value.alloc);
			return *this;
		}

		cbor_value& operator=(cbor_composite<cbor_value>& value) {
			contents = value;
			return *this;
		}

		template<typename T>
		cbor_value& operator=(T& value) {
			contents = value;
			return *this;
		}

		template<typename T>
		cbor_value& operator=(T&& value) {
			contents = gp::move(value);
			return *this;
		}

		auto new_array() {
			return gp::vector<cbor_value>{alloc};
		}

		auto new_object() {
			return gp::vector<gp::pair<cbor_value, cbor_value>>{alloc};
		}

		static gp::pair<uint64_t, gp::buffer<std::byte>::associated_iterator> decode_integer(gp::buffer<std::byte> src) {
			#define ERROR return {0, src.begin()}
			auto local = (uint8_t)0b00011111 & (uint8_t)src[0];
			if(local <= 23) {
				return {local, src.begin()+1};
			} else {
				switch((cbor_oths)local) {
					case cbor_oths::byte: {
						if(src.size()<2) ERROR;
						return {
							(uint8_t)*(src.begin()+1), 
							src.begin()+2
						};
					}
					case cbor_oths::word: {
						if(src.size()<3) ERROR;
						return {
							uint16_t(*(src.slice_start(3).slice_end(2).cast<gp::endian_wrapper<uint16_t, endian::big>>().begin().data)), 
							src.begin()+3
						};
					}
					case cbor_oths::dword: {
						if(src.size()<5) ERROR;
						return {
							uint32_t(*(src.slice_start(5).slice_end(4).cast<gp::endian_wrapper<uint32_t, endian::big>>().begin().data)), 
							src.begin()+5
						};
					}
					case cbor_oths::qword: {
						if(src.size()<9) ERROR;
						return {
							uint64_t(*(src.slice_start(9).slice_end(8).cast<gp::endian_wrapper<uint64_t, endian::big>>().begin().data)), 
							src.begin()+9
						};
					}
					default: {
						ERROR;
					}
				}
			}
			#undef ERROR
		}

		template<typename U>
		bool is_a() {
			if constexpr (
				std::is_same_v<U, uint8_t>
				|| std::is_same_v<U, uint16_t>
				|| std::is_same_v<U, uint32_t>
				|| std::is_same_v<U, uint64_t>
				|| std::is_same_v<U, int8_t>
				|| std::is_same_v<U, int16_t>
				|| std::is_same_v<U, int32_t>
				|| std::is_same_v<U, int64_t>
			) {
				if(contents.is_a<cbor_number>())
				{
					auto& v = contents.value<cbor_number>();
					if(!std::is_signed_v<U> && v.is_negative()) return false;
					if(
						std::numeric_limits<int64_t>::max()<v.value
						&& (
							!std::is_same_v<U, uint64_t>
							|| v.is_negative()
						)
					) return false;
					int64_t signed_v = (v.is_negative() ? -1 : 1 ) * v.value;
					if(
						std::numeric_limits<U>::min() <= signed_v
						&& std::numeric_limits<U>::max() >= signed_v
					) return true;
					return false;
				} else return false;
			} else if constexpr (
				std::is_same_v<U, ieee754_hf>
				|| std::is_same_v<U, float>
				|| std::is_same_v<U, double>
			) {
				auto& v = contents.value<cbor_floating_point>();
				switch(v.contents.type()) {
					case cbor_floating_point::backing::alt<ieee754_hf>():
						return std::is_same_v<U, ieee754_hf>;
					case cbor_floating_point::backing::alt<float>():
						return std::is_same_v<U, float> || std::is_same_v<U, ieee754_hf>;
					case cbor_floating_point::backing::alt<double>():
						return std::is_same_v<U, double> || std::is_same_v<U, float> || std::is_same_v<U, ieee754_hf>;
					default:
						return false;
				}
			} else {
				return contents.is_a<U>();
			}
		}

		template<typename U>
		U get_value() {
			gp_config::assertion(is_a<U>(), "can't convert value legally");
			if constexpr (
				std::is_same_v<U, uint8_t>
				|| std::is_same_v<U, uint16_t>
				|| std::is_same_v<U, uint32_t>
				|| std::is_same_v<U, uint64_t>
				|| std::is_same_v<U, int8_t>
				|| std::is_same_v<U, int16_t>
				|| std::is_same_v<U, int32_t>
				|| std::is_same_v<U, int64_t>
			) {
				auto& v = contents.value<cbor_number>();
				return (v.is_negative() ? -1 : 1 ) * v.value;
			} else if constexpr (
				std::is_same_v<U, ieee754_hf>
				|| std::is_same_v<U, float>
				|| std::is_same_v<U, double>
			) {
				auto& v = contents.value<cbor_floating_point>();
				return v;
			} else {
				return contents.value<U>();
			}
		}

		static gp::pair<cbor_value, gp::buffer<std::byte>::associated_iterator> decode(allocator& alloc, gp::buffer<std::byte> src) {
			#define ERROR return {cbor_value{alloc}, src.begin()}
			if(src.size()==0) ERROR;
			auto discriminant = (cbor_type)(((uint8_t)*src.begin()) >> 5);
			auto local = uint8_t(((uint8_t)*src.begin()) & 0b00011111);
			switch(discriminant) {
				case cbor_type::uint: 
				case cbor_type::nint: {
					auto nb = decode_integer(src);
					return {cbor_value{cbor_number{(bool)discriminant, nb.first}, alloc}, nb.second};
				}
				case cbor_type::bstr: {
					gp::vector<std::byte> str{alloc};
					if(local == 31) {
						auto sub = src.slice_end(src.size() - 1);
						if(!sub.size()) ERROR;
						while((uint8_t)sub[0] && 0b00011111 != 31) {
							auto len = decode_integer(sub);
							if(len.second == sub.begin()) ERROR;
							str.reserve(str.size() + len.first);
							auto end = (len.second + len.first);
							if(len.first + (len.second - src.begin()) > src.size()) ERROR;
							for(auto it = len.second; it != end; it++) {
								str.push_back(*it);
							}
							sub = sub.slice_end(sub.size() - (len.first + (len.second - sub.begin())));
						}
						return {cbor_value{str, alloc}, sub.begin()+1};
					} else if(auto len = decode_integer(src); len.second != src.begin()) {
						str.reserve(len.first);
						auto end = (len.second + len.first);
						if(len.first + (len.second - src.begin()) > src.size()) ERROR;
						for(auto it = len.second; it != end; it++) {
							str.push_back(*it);
						}
						return {cbor_value{str, alloc}, end};
					}
					ERROR;
				}
				case cbor_type::tstr: {
					ERROR;
				}
				case cbor_type::list: {
					gp::vector<cbor_value> list{alloc};
					if(local == 31) {
						// TODO: Add the indefinite length version
					} else if(auto len = decode_integer(src); len.second != src.begin()) {
						auto sub = src.slice_end(src.size()-(len.second - src.begin()));
						while(len.first--) {
							auto tmp = decode(alloc, sub);
							if(tmp.second == sub.begin()) ERROR;
							list.push_back(tmp.first);
							sub = sub.slice_end(sub.size() - (tmp.second - sub.begin()));
						}
						return {cbor_value{list, alloc}, sub.begin()};
					}
					ERROR;
				}
				case cbor_type::hmap: {
					gp::vector<gp::pair<cbor_value, cbor_value>> list{alloc};
					if(local == 31) {
						// TODO: Add the indefinite length version
					} else if(auto len = decode_integer(src); len.second != src.begin()) {
						auto sub = src.slice_end(src.size()-(len.second - src.begin()));
						while(len.first--) {
							auto tmpl = decode(alloc, sub);
							if(tmpl.second == sub.begin()) ERROR;
							sub = sub.slice_end(sub.size() - (tmpl.second - sub.begin()));
							auto tmpr = decode(alloc, sub);
							if(tmpr.second == sub.begin()) ERROR;
							list.push_back(gp::make_pair(
								tmpl.first,
								tmpr.first
							));
							sub = sub.slice_end(sub.size() - (tmpr.second - sub.begin()));
						}
						return {cbor_value{list, alloc}, sub.begin()};
					}
					ERROR;
				}
				case cbor_type::tags: {
					// TODO: add tag decoding methods
				}
				case cbor_type::oths: {
					switch((cbor_oths)local) {
						case cbor_oths::value_false: {
							return {
								cbor_value{
									cbor_composite<cbor_value>(false), 
									alloc
								}, 
								src.begin()+1
							};
						}
						case cbor_oths::value_true: {
							return {
								cbor_value{
									cbor_composite<cbor_value>(true), 
									alloc
								}, 
								src.begin()+1
							};
						}
						case cbor_oths::value_null: {
							return {
								cbor_value{
									cbor_composite<cbor_value>(nullopt), 
									alloc
								}, 
								src.begin()+1
							};
						}
						case cbor_oths::value_undefined: {
							return {
								cbor_value{
									cbor_composite<cbor_value>(undefined_t{}), 
									alloc
								}, 
								src.begin()+1
							};
						}
						case cbor_oths::word: {
							if(src.size()<3) ERROR;
							return {
								cbor_value{
									cbor_floating_point{ieee754_hf(*(src.slice_start(3).slice_end(2).cast<gp::endian_wrapper<ieee754_hf, endian::big>>().begin().data))}, 
									alloc
								}, 
								src.begin()+3
							};
						}
						case cbor_oths::dword: {
							if(src.size()<5) ERROR;
							return {
								cbor_value{
									cbor_floating_point{float(*(src.slice_start(5).slice_end(4).cast<gp::endian_wrapper<float, endian::big>>().begin().data))}, 
									alloc
								}, 
								src.begin()+5
							};
						}
						case cbor_oths::qword: {
							if(src.size()<9) ERROR;
							return {
								cbor_value{
									cbor_floating_point{double(*(src.slice_start(9).slice_end(8).cast<gp::endian_wrapper<double, endian::big>>().begin().data))}, 
									alloc
								}, 
								src.begin()+9
							};
						}
						default: {
							ERROR;
						}
					}	
				}
			}
			ERROR;
			#undef ERROR
		}

		static auto encode_float(buffer<std::byte> dest, cbor_floating_point& value) {
			switch(value.contents.type()) {
				case cbor_floating_point::backing::alt<ieee754_hf>():{
					if(dest.size() < 3) return dest.begin();
					dest[0] = std::byte(((uint8_t)cbor_type::oths << 5u) + (uint8_t)cbor_oths::word);
					(dest.slice_start(3).slice_end(2).cast<gp::endian_wrapper<ieee754_hf, gp::endian::big>>())[0] = value.contents.value<ieee754_hf>();
					return dest.begin()+3;
				}
				case cbor_floating_point::backing::alt<float>():{
					if(dest.size() < 5) return dest.begin();
					dest[0] = std::byte(((uint8_t)cbor_type::oths << 5u) + (uint8_t)cbor_oths::dword);
					(dest.slice_start(5).slice_end(4).cast<gp::endian_wrapper<float, gp::endian::big>>())[0] = value.contents.value<float>();
					return dest.begin()+5;
				}
				case cbor_floating_point::backing::alt<double>():{
					if(dest.size() < 9) return dest.begin();
					dest[0] = std::byte(((uint8_t)cbor_type::oths << 5u) + (uint8_t)cbor_oths::qword);
					(dest.slice_start(9).slice_end(8).cast<gp::endian_wrapper<double, gp::endian::big>>())[0] = value.contents.value<double>();
					return dest.begin()+9;
				}
				default: return dest.begin();
			}
		}

		static auto encode_length(buffer<std::byte> dest, cbor_type major, uint64_t value) {
			auto num = value;
			if(value <= 23) {
				if(dest.size() < 1) return dest.begin();
				dest[0] = std::byte(((uint8_t)major << 5u) + value);
				return dest.begin()+1;
			} else if(value <= 255) {
				if(dest.size() < 2) return dest.begin();
				dest[0] = std::byte(((uint8_t)major << 5u) + (uint8_t)cbor_oths::byte);
				dest[1] = std::byte(value);
				return dest.begin() + 2;
			} else if(value <= 65535) {
				if(dest.size() < 3) return dest.begin();
				dest[0] = std::byte(((uint8_t)major << 5u) + (uint8_t)cbor_oths::word);
				(dest.slice_start(3).slice_end(2).cast<gp::endian_wrapper<uint16_t, gp::endian::big>>())[0] = num;
				return dest.begin()+3;
			} else if(value <= 4294967295) {
				if(dest.size() < 5) return dest.begin();
				dest[0] = std::byte(((uint8_t)major << 5u) + (uint8_t)cbor_oths::dword);
				(dest.slice_start(5).slice_end(4).cast<gp::endian_wrapper<uint32_t, gp::endian::big>>())[0] = num;
				return dest.begin()+5;
			} else {
				if(dest.size() < 9) return dest.begin();
				dest[0] = std::byte(((uint8_t)major << 5u) + (uint8_t)cbor_oths::qword);
				(dest.slice_start(9).slice_end(8).cast<gp::endian_wrapper<uint64_t, gp::endian::big>>())[0] = num;
				return dest.begin()+9;
			}
		}

		auto encode(buffer<std::byte> dest) {
			switch(contents.type()) {
				case cbor_composite<cbor_value>::alt<undefined_t>(): {
					if(dest.size() < 1) return dest.begin();
					dest[0] = std::byte(((uint8_t)cbor_type::oths << 5u) + (uint8_t)cbor_oths::value_undefined);
					return dest.begin()+1;
				}
				case cbor_composite<cbor_value>::alt<cbor_number>(): {
					auto& ref = contents.value<cbor_number>();
					return encode_length(
						dest, 
						ref.is_negative() ? cbor_type::nint : cbor_type::uint,
						ref.value
					);
				}
				case cbor_composite<cbor_value>::alt<gp::vector<std::byte>>(): {
					auto& ref = contents.value<gp::vector<std::byte>>();
					auto it = encode_length(
						dest,
						cbor_type::bstr,
						ref.size()
					);
					if(it == dest.begin()) return it;

					for(auto a : ref) {
						*(it++) = a;
					}
					return it;
				}
				case cbor_composite<cbor_value>::alt<bool>(): {
					if(dest.size() < 1) return dest.begin();
					if(contents.value<bool>()) 
						dest[0] = std::byte(((uint8_t)cbor_type::oths << 5u) + (uint8_t)cbor_oths::value_true);
					else 
						dest[0] = std::byte(((uint8_t)cbor_type::oths << 5u) + (uint8_t)cbor_oths::value_false);
					return dest.begin()+1;
				}
				case cbor_composite<cbor_value>::alt<gp::vector<cbor_value>>(): {
					auto& ary = contents.value<gp::vector<cbor_value>>();
					auto it_begin = encode_length(dest, cbor_type::list, ary.size());
					if(it_begin == dest.begin()) return dest.begin();
					for(auto& elem : ary) {
						auto slice = dest.slice_end(dest.size() - (it_begin - dest.begin()));
						auto it = elem.encode(slice);
						if(it == it_begin) 
							return dest.begin();
						it_begin = it;
					}
					return it_begin;
				}
				case cbor_composite<cbor_value>::alt<gp::vector<gp::pair<cbor_value, cbor_value>>>(): {
					auto& ary = contents.value<gp::vector<gp::pair<cbor_value,cbor_value>>>();
					auto it_begin = encode_length(dest, cbor_type::hmap, ary.size());
					if(it_begin == dest.begin()) return dest.begin();
					
					for(auto& elem : ary) {
						auto slice = dest.slice_end(dest.size() - (it_begin - dest.begin()));
						auto it = elem.first.encode(slice);
						if(it == it_begin) return dest.begin();
						it_begin = it;

						slice = dest.slice_end(dest.size() - (it_begin - dest.begin()));
						it = elem.second.encode(slice);
						if(it == it_begin) return dest.begin();
						it_begin = it;
					}
					return it_begin;
				}
				case cbor_composite<cbor_value>::alt<gp::nullopt_t>(): {
					if(dest.size() < 1) return dest.begin();
					dest[0] = std::byte(((uint8_t)cbor_type::oths << 5u) + (uint8_t)cbor_oths::value_null);
					return dest.begin()+1;
				}
				case cbor_composite<cbor_value>::alt<cbor_floating_point>(): {
					return encode_float(dest, contents.value<cbor_floating_point>());
				}
				default: return dest.begin();
			}
		}
	};
}