rippled/src/libxrpl/protocol/Serializer.cpp

#include <xrpl/protocol/Serializer.h>

#include <xrpl/basics/Buffer.h>
#include <xrpl/basics/Slice.h>
#include <xrpl/basics/base_uint.h>
#include <xrpl/basics/contract.h>
#include <xrpl/basics/safe_cast.h>
#include <xrpl/beast/utility/instrumentation.h>
#include <xrpl/protocol/HashPrefix.h>
#include <xrpl/protocol/digest.h>

#include <boost/endian/conversion.hpp>

#include <array>
#include <cstddef>
#include <cstdint>
#include <cstring>
#include <stdexcept>
#include <string>
#include <type_traits>

namespace xrpl {

int
Serializer::add16(std::uint16_t i)
{
    int const ret = data_.size();
    data_.push_back(static_cast<unsigned char>(i >> 8));
    data_.push_back(static_cast<unsigned char>(i & 0xff));
    return ret;
}

int
Serializer::add32(HashPrefix p)
{
    // This should never trigger; the size & type of a hash prefix are
    // integral parts of the protocol and unlikely to ever change.
    static_assert(std::is_same_v<std::uint32_t, std::underlying_type_t<decltype(p)>>);

    return add32(safeCast<std::uint32_t>(p));
}

template <>
int
Serializer::addInteger(unsigned char i)
{
    return add8(i);
}
template <>
int
Serializer::addInteger(std::uint16_t i)
{
    return add16(i);
}
template <>
int
Serializer::addInteger(std::uint32_t i)
{
    return add32(i);
}
template <>
int
Serializer::addInteger(std::uint64_t i)
{
    return add64(i);
}
template <>
int
Serializer::addInteger(std::int32_t i)
{
    return add32(i);
}

int
Serializer::addRaw(Blob const& vector)
{
    int const ret = data_.size();
    data_.insert(data_.end(), vector.begin(), vector.end());
    return ret;
}

int
Serializer::addRaw(Slice slice)
{
    int const ret = data_.size();
    data_.insert(data_.end(), slice.begin(), slice.end());
    return ret;
}

int
Serializer::addRaw(Serializer const& s)
{
    int const ret = data_.size();
    data_.insert(data_.end(), s.begin(), s.end());
    return ret;
}

int
Serializer::addRaw(void const* ptr, int len)
{
    int const ret = data_.size();
    data_.insert(data_.end(), static_cast<char const*>(ptr), static_cast<char const*>(ptr) + len);
    return ret;
}

int
Serializer::addFieldID(int type, int name)
{
    int const ret = data_.size();
    XRPL_ASSERT(
        (type > 0) && (type < 256) && (name > 0) && (name < 256),
        "xrpl::Serializer::addFieldID : inputs inside range");

    if (type < 16)
    {
        if (name < 16)
        {
            // common type, common name
            data_.push_back(static_cast<unsigned char>((type << 4) | name));
        }
        else
        {
            // common type, uncommon name
            data_.push_back(static_cast<unsigned char>(type << 4));
            data_.push_back(static_cast<unsigned char>(name));
        }
    }
    else if (name < 16)
    {
        // uncommon type, common name
        data_.push_back(static_cast<unsigned char>(name));
        data_.push_back(static_cast<unsigned char>(type));
    }
    else
    {
        // uncommon type, uncommon name
        data_.push_back(static_cast<unsigned char>(0));
        data_.push_back(static_cast<unsigned char>(type));
        data_.push_back(static_cast<unsigned char>(name));
    }

    return ret;
}

int
Serializer::add8(unsigned char byteValue)
{
    int const ret = data_.size();
    data_.push_back(byteValue);
    return ret;
}

bool
Serializer::get8(int& byte, int offset) const
{
    if (offset >= data_.size())
        return false;

    byte = data_[offset];
    return true;
}

bool
Serializer::chop(int bytes)
{
    if (bytes > data_.size())
        return false;

    data_.resize(data_.size() - bytes);
    return true;
}

uint256
Serializer::getSHA512Half() const
{
    return sha512Half(makeSlice(data_));
}

int
Serializer::addVL(Blob const& vector)
{
    int const ret = addEncoded(vector.size());
    addRaw(vector);
    XRPL_ASSERT(
        data_.size() == (ret + vector.size() + encodeLengthLength(vector.size())),
        "xrpl::Serializer::addVL : size matches expected");
    return ret;
}

int
Serializer::addVL(Slice const& slice)
{
    int const ret = addEncoded(slice.size());
    if (!slice.empty())
        addRaw(slice.data(), slice.size());
    return ret;
}

int
Serializer::addVL(void const* ptr, int len)
{
    int const ret = addEncoded(len);

    if (len != 0)
        addRaw(ptr, len);

    return ret;
}

int
Serializer::addEncoded(int length)
{
    // Without this, a negative length would fall into the 1 byte case below and
    // be cast to a first byte no header uses. A size too big for int arrives
    // here negative as well, since callers pass sizes through this parameter.
    if (length < kMinValueOfLengthFor1ByteHeader)
        Throw<std::overflow_error>("addEncoded: length is negative or did not fit in an int");

    std::array<std::byte, kMaxNumberOfBytesInHeader> bytes{};
    int numBytes = 0;

    if (length <= kMaxValueOfLengthFor1ByteHeader)
    {
        bytes[0] = static_cast<std::byte>(length);
        numBytes = 1;
    }
    else if (length <= kMaxValueOfLengthFor2ByteHeader)
    {
        // Count from the smallest length a 2 byte header covers.
        int const offset = length - kMinValueOfLengthFor2ByteHeader;
        bytes[0] = static_cast<std::byte>(
            kMinValueOfFirstByteFor2ByteHeader + (offset / kNumberOfValuesInOneByte));
        bytes[1] = static_cast<std::byte>(offset % kNumberOfValuesInOneByte);
        numBytes = 2;
    }
    else if (length <= kMaxValueOfLengthFor3ByteHeader)
    {
        int const offset = length - kMinValueOfLengthFor3ByteHeader;
        bytes[0] = static_cast<std::byte>(
            kMinValueOfFirstByteFor3ByteHeader + (offset / kNumberOfValuesInTwoBytes));
        bytes[1] =
            static_cast<std::byte>((offset / kNumberOfValuesInOneByte) % kNumberOfValuesInOneByte);
        bytes[2] = static_cast<std::byte>(offset % kNumberOfValuesInOneByte);
        numBytes = 3;
    }
    else
    {
        Throw<std::overflow_error>("addEncoded: length is too large to encode");
    }

    return addRaw(bytes.data(), numBytes);
}

int
Serializer::encodeLengthLength(int length)
{
    if (length < kMinValueOfLengthFor1ByteHeader)
    {
        Throw<std::overflow_error>(
            "encodeLengthLength: length is negative or did not fit in an int");
    }

    if (length <= kMaxValueOfLengthFor1ByteHeader)
        return 1;

    if (length <= kMaxValueOfLengthFor2ByteHeader)
        return 2;

    if (length <= kMaxValueOfLengthFor3ByteHeader)
        return 3;

    Throw<std::overflow_error>("encodeLengthLength: length is too large to encode");
}

int
Serializer::decodeLengthLength(std::byte firstByte)
{
    int const firstByteValue = std::to_integer<int>(firstByte);

    if (firstByteValue <= kMaxValueOfFirstByteFor1ByteHeader)
        return 1;

    if (firstByteValue <= kMaxValueOfFirstByteFor2ByteHeader)
        return 2;

    if (firstByteValue <= kMaxValueOfFirstByteFor3ByteHeader)
        return 3;

    Throw<std::overflow_error>("decodeLengthLength: first byte does not start any header");
}

int
Serializer::decodeVLLength(std::byte firstByte)
{
    int const length = std::to_integer<int>(firstByte);

    // A bigger value means a longer header, so it is not a length by itself.
    if (length > kMaxValueOfLengthFor1ByteHeader)
        Throw<std::overflow_error>("decodeVLLength 1 byte: first byte is not a length");

    return length;
}

int
Serializer::decodeVLLength(std::byte firstByte, std::byte secondByte)
{
    int const firstByteValue = std::to_integer<int>(firstByte);

    if (firstByteValue < kMinValueOfFirstByteFor2ByteHeader)
        Throw<std::overflow_error>("decodeVLLength 2 byte: first byte is below the range");

    if (firstByteValue > kMaxValueOfFirstByteFor2ByteHeader)
        Throw<std::overflow_error>("decodeVLLength 2 byte: first byte is above the range");

    // Both bytes are bounded by their own type, and the first one is bounded to
    // the 2 byte range above, so this cannot leave the range the header covers.
    return kMinValueOfLengthFor2ByteHeader +
        ((firstByteValue - kMinValueOfFirstByteFor2ByteHeader) * kNumberOfValuesInOneByte) +
        std::to_integer<int>(secondByte);
}

int
Serializer::decodeVLLength(std::byte firstByte, std::byte secondByte, std::byte thirdByte)
{
    int const firstByteValue = std::to_integer<int>(firstByte);

    if (firstByteValue < kMinValueOfFirstByteFor3ByteHeader)
        Throw<std::overflow_error>("decodeVLLength 3 byte: first byte is below the range");

    if (firstByteValue > kMaxValueOfFirstByteFor3ByteHeader)
        Throw<std::overflow_error>("decodeVLLength 3 byte: first byte is above the range");

    int const length = kMinValueOfLengthFor3ByteHeader +
        ((firstByteValue - kMinValueOfFirstByteFor3ByteHeader) * kNumberOfValuesInTwoBytes) +
        (std::to_integer<int>(secondByte) * kNumberOfValuesInOneByte) +
        std::to_integer<int>(thirdByte);

    // A 3 byte header reaches further than kMaxValueOfLengthFor3ByteHeader, which
    // is as far as the encoder goes. Refuse the rest, so every length accepted
    // here is one that can be written back.
    if (length > kMaxValueOfLengthFor3ByteHeader)
        Throw<std::overflow_error>("decodeVLLength 3 byte: length is too large to re-encode");

    return length;
}

//------------------------------------------------------------------------------

SerialIter::SerialIter(void const* data, std::size_t size) noexcept
    : p_(reinterpret_cast<std::uint8_t const*>(data)), remain_(size)
{
}

void
SerialIter::reset() noexcept
{
    p_ -= used_;
    remain_ += used_;
    used_ = 0;
}

void
SerialIter::skip(int length)
{
    if (remain_ < length)
        Throw<std::runtime_error>("invalid SerialIter skip");
    p_ += length;
    used_ += length;
    remain_ -= length;
}

unsigned char
SerialIter::get8()
{
    if (remain_ < 1)
        Throw<std::runtime_error>("invalid SerialIter get8");
    unsigned char const t = *p_;
    ++p_;
    ++used_;
    --remain_;
    return t;
}

std::uint16_t
SerialIter::get16()
{
    if (remain_ < 2)
        Throw<std::runtime_error>("invalid SerialIter get16");
    auto t = p_;
    p_ += 2;
    used_ += 2;
    remain_ -= 2;
    return (std::uint64_t(t[0]) << 8) + std::uint64_t(t[1]);
}

std::uint32_t
SerialIter::get32()
{
    if (remain_ < 4)
        Throw<std::runtime_error>("invalid SerialIter get32");
    auto t = p_;
    p_ += 4;
    used_ += 4;
    remain_ -= 4;
    return (std::uint64_t(t[0]) << 24) + (std::uint64_t(t[1]) << 16) + (std::uint64_t(t[2]) << 8) +
        std::uint64_t(t[3]);
}

std::uint64_t
SerialIter::get64()
{
    if (remain_ < 8)
        Throw<std::runtime_error>("invalid SerialIter get64");
    auto t = p_;
    p_ += 8;
    used_ += 8;
    remain_ -= 8;
    return (std::uint64_t(t[0]) << 56) + (std::uint64_t(t[1]) << 48) + (std::uint64_t(t[2]) << 40) +
        (std::uint64_t(t[3]) << 32) + (std::uint64_t(t[4]) << 24) + (std::uint64_t(t[5]) << 16) +
        (std::uint64_t(t[6]) << 8) + std::uint64_t(t[7]);
}

std::int32_t
SerialIter::geti32()
{
    if (remain_ < 4)
        Throw<std::runtime_error>("invalid SerialIter geti32");
    auto t = p_;
    p_ += 4;
    used_ += 4;
    remain_ -= 4;
    return boost::endian::load_big_s32(t);
}

std::int64_t
SerialIter::geti64()
{
    if (remain_ < 8)
        Throw<std::runtime_error>("invalid SerialIter geti64");
    auto t = p_;
    p_ += 8;
    used_ += 8;
    remain_ -= 8;
    return boost::endian::load_big_s64(t);
}

void
SerialIter::getFieldID(int& type, int& name)
{
    type = get8();
    name = type & 15;
    type >>= 4;

    if (type == 0)
    {
        // uncommon type
        type = get8();
        if (type < 16)
            Throw<std::runtime_error>("gFID: uncommon type out of range " + std::to_string(type));
    }

    if (name == 0)
    {
        // uncommon name
        name = get8();
        if (name < 16)
            Throw<std::runtime_error>("gFID: uncommon name out of range " + std::to_string(name));
    }
}

// getRaw for blob or buffer
template <class T>
T
SerialIter::getRawHelper(int size)
{
    static_assert(std::is_same_v<T, Blob> || std::is_same_v<T, Buffer>);
    if (remain_ < size)
        Throw<std::runtime_error>("invalid SerialIter getRaw");
    T result(size);
    if (size != 0)
    {
        // It's normally safe to call memcpy with size set to 0 (see the
        // C99 standard 7.21.1/2). However, here this could mean that
        // result.data would be null, which would trigger undefined behavior.
        std::memcpy(result.data(), p_, size);
        p_ += size;
        used_ += size;
        remain_ -= size;
    }
    return result;
}

// VFALCO DEPRECATED Returns a copy
Blob
SerialIter::getRaw(int size)
{
    return getRawHelper<Blob>(size);
}

int
SerialIter::getVLDataLength()
{
    std::byte const firstByte{get8()};
    int datLen = 0;
    int const lenLen = Serializer::decodeLengthLength(firstByte);
    if (lenLen == 1)
    {
        datLen = Serializer::decodeVLLength(firstByte);
    }
    else if (lenLen == 2)
    {
        std::byte const secondByte{get8()};
        datLen = Serializer::decodeVLLength(firstByte, secondByte);
    }
    else
    {
        XRPL_ASSERT(lenLen == 3, "xrpl::SerialIter::getVLDataLength : lenLen is 3");
        std::byte const secondByte{get8()};
        std::byte const thirdByte{get8()};
        datLen = Serializer::decodeVLLength(firstByte, secondByte, thirdByte);
    }
    return datLen;
}

Slice
SerialIter::getSlice(std::size_t bytes)
{
    if (bytes > remain_)
        Throw<std::runtime_error>("invalid SerialIter getSlice");
    Slice const s(p_, bytes);
    p_ += bytes;
    used_ += bytes;
    remain_ -= bytes;
    return s;
}

// VFALCO DEPRECATED Returns a copy
Blob
SerialIter::getVL()
{
    return getRaw(getVLDataLength());
}

Buffer
SerialIter::getVLBuffer()
{
    return getRawHelper<Buffer>(getVLDataLength());
}

}  // namespace xrpl