mirrors/json
1
0
Fork 0
mirror of https://github.com/nlohmann/json.git synced 2025-05-11 13:43:57 +00:00
Code Issues Projects Releases Wiki Activity

Optimize binary get_number implementation by reading multiple bytes at once (#4391)

Browse Source 
* multibyte binary reader

* wide_string_input_adapter fallback to get_character

Update input_adapters.hpp

* Update json.hpp

* Add from msgpack test

* Test for broken msgpack with stream, address some warnings

* Reading binary number from wchar as an error, address warnings

* Not casting float to int, it violates strict aliasing rule
This commit is contained in:
Tianyi Chen 2024-11-29 00:19:58 -08:00 committed by GitHub
parent e41905fcb0
commit 935c6eeb5a
No known key found for this signature in database
GPG Key ID: B5690EEEBB952194
5 changed files with 258 additions and 42 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

77
include/nlohmann/detail/input/binary_reader.hpp
View File

@ -20,6 +20,9 @@
#include <string> // char_traits, string
#include <utility> // make_pair, move
#include <vector> // vector
#ifdef __cpp_lib_byteswap
#include <bit> //byteswap
#endif
#include <nlohmann/detail/exceptions.hpp>
#include <nlohmann/detail/input/input_adapters.hpp>
@ -2754,6 +2757,29 @@ class binary_reader
return current = ia.get_character();
}
/*!
@brief get_to read into a primitive type
This function provides the interface to the used input adapter. It does
not throw in case the input reached EOF, but returns false instead
@return bool, whether the read was successful
*/
template<class T>
bool get_to(T& dest, const input_format_t format, const char* context)
{
auto new_chars_read = ia.get_elements(&dest);
chars_read += new_chars_read;
if (JSON_HEDLEY_UNLIKELY(new_chars_read < sizeof(T)))
{
// in case of failure, advance position by 1 to report failing location
++chars_read;
sax->parse_error(chars_read, "<end of file>", parse_error::create(110, chars_read, exception_message(format, "unexpected end of input", context), nullptr));
return false;
}
return true;
}
/*!
@return character read from the input after ignoring all 'N' entries
*/
@ -2768,6 +2794,28 @@ class binary_reader
return current;
}
template<class NumberType>
static void byte_swap(NumberType& number)
{
constexpr std::size_t sz = sizeof(number);
#ifdef __cpp_lib_byteswap
if constexpr (sz == 1)
{
return;
}
if constexpr(std::is_integral_v<NumberType>)
{
number = std::byteswap(number);
return;
}
#endif
auto* ptr = reinterpret_cast<std::uint8_t*>(&number);
for (std::size_t i = 0; i < sz / 2; ++i)
{
std::swap(ptr[i], ptr[sz - i - 1]);
}
}
/*
@brief read a number from the input
@ -2786,29 +2834,16 @@ class binary_reader
template<typename NumberType, bool InputIsLittleEndian = false>
bool get_number(const input_format_t format, NumberType& result)
{
// step 1: read input into array with system's byte order
std::array<std::uint8_t, sizeof(NumberType)> vec{};
for (std::size_t i = 0; i < sizeof(NumberType); ++i)
// read in the original format
if (JSON_HEDLEY_UNLIKELY(!get_to(result, format, "number")))
{
get();
if (JSON_HEDLEY_UNLIKELY(!unexpect_eof(format, "number")))
{
return false;
}
// reverse byte order prior to conversion if necessary
if (is_little_endian != (InputIsLittleEndian || format == input_format_t::bjdata))
{
vec[sizeof(NumberType) - i - 1] = static_cast<std::uint8_t>(current);
}
else
{
vec[i] = static_cast<std::uint8_t>(current); // LCOV_EXCL_LINE
}
return false;
}
if (is_little_endian != (InputIsLittleEndian || format == input_format_t::bjdata))
{
byte_swap(result);
}
// step 2: convert array into number of type T and return
std::memcpy(&result, vec.data(), sizeof(NumberType));
return true;
}

45
include/nlohmann/detail/input/input_adapters.hpp
View File

@ -68,6 +68,13 @@ class file_input_adapter
return std::fgetc(m_file);
}
// returns the number of characters successfully read
template<class T>
std::size_t get_elements(T* dest, std::size_t count = 1)
{
return fread(dest, 1, sizeof(T) * count, m_file);
}
private:
/// the file pointer to read from
std::FILE* m_file;
@ -127,6 +134,17 @@ class input_stream_adapter
return res;
}
template<class T>
std::size_t get_elements(T* dest, std::size_t count = 1)
{
auto res = static_cast<std::size_t>(sb->sgetn(reinterpret_cast<char*>(dest), static_cast<std::streamsize>(count * sizeof(T))));
if (JSON_HEDLEY_UNLIKELY(res < count * sizeof(T)))
{
is->clear(is->rdstate() | std::ios::eofbit);
}
return res;
}
private:
/// the associated input stream
std::istream* is = nullptr;
@ -158,6 +176,26 @@ class iterator_input_adapter
return char_traits<char_type>::eof();
}
// for general iterators, we cannot really do something better than falling back to processing the range one-by-one
template<class T>
std::size_t get_elements(T* dest, std::size_t count = 1)
{
auto* ptr = reinterpret_cast<char*>(dest);
for (std::size_t read_index = 0; read_index < count * sizeof(T); ++read_index)
{
if (JSON_HEDLEY_LIKELY(current != end))
{
ptr[read_index] = static_cast<char>(*current);
std::advance(current, 1);
}
else
{
return read_index;
}
}
return count * sizeof(T);
}
private:
IteratorType current;
IteratorType end;
@ -321,6 +359,13 @@ class wide_string_input_adapter
return utf8_bytes[utf8_bytes_index++];
}
// parsing binary with wchar doesn't make sense, but since the parsing mode can be runtime, we need something here
template<class T>
std::size_t get_elements(T* /*dest*/, std::size_t /*count*/ = 1)
{
JSON_THROW(parse_error::create(112, 1, "wide string type cannot be interpreted as binary data", nullptr));
}
private:
BaseInputAdapter base_adapter;

122
single_include/nlohmann/json.hpp
View File

@ -6220,6 +6220,9 @@ NLOHMANN_JSON_NAMESPACE_END
#include <string> // char_traits, string
#include <utility> // make_pair, move
#include <vector> // vector
#ifdef __cpp_lib_byteswap
#include <bit> //byteswap
#endif
// #include <nlohmann/detail/exceptions.hpp>
@ -6298,6 +6301,13 @@ class file_input_adapter
return std::fgetc(m_file);
}
// returns the number of characters successfully read
template<class T>
std::size_t get_elements(T* dest, std::size_t count = 1)
{
return fread(dest, 1, sizeof(T) * count, m_file);
}
private:
/// the file pointer to read from
std::FILE* m_file;
@ -6357,6 +6367,17 @@ class input_stream_adapter
return res;
}
template<class T>
std::size_t get_elements(T* dest, std::size_t count = 1)
{
auto res = static_cast<std::size_t>(sb->sgetn(reinterpret_cast<char*>(dest), static_cast<std::streamsize>(count * sizeof(T))));
if (JSON_HEDLEY_UNLIKELY(res < count * sizeof(T)))
{
is->clear(is->rdstate() | std::ios::eofbit);
}
return res;
}
private:
/// the associated input stream
std::istream* is = nullptr;
@ -6388,6 +6409,26 @@ class iterator_input_adapter
return char_traits<char_type>::eof();
}
// for general iterators, we cannot really do something better than falling back to processing the range one-by-one
template<class T>
std::size_t get_elements(T* dest, std::size_t count = 1)
{
auto* ptr = reinterpret_cast<char*>(dest);
for (std::size_t read_index = 0; read_index < count * sizeof(T); ++read_index)
{
if (JSON_HEDLEY_LIKELY(current != end))
{
ptr[read_index] = static_cast<char>(*current);
std::advance(current, 1);
}
else
{
return read_index;
}
}
return count * sizeof(T);
}
private:
IteratorType current;
IteratorType end;
@ -6551,6 +6592,13 @@ class wide_string_input_adapter
return utf8_bytes[utf8_bytes_index++];
}
// parsing binary with wchar doesn't make sense, but since the parsing mode can be runtime, we need something here
template<class T>
std::size_t get_elements(T* /*dest*/, std::size_t /*count*/ = 1)
{
JSON_THROW(parse_error::create(112, 1, "wide string type cannot be interpreted as binary data", nullptr));
}
private:
BaseInputAdapter base_adapter;
@ -12007,6 +12055,29 @@ class binary_reader
return current = ia.get_character();
}
/*!
@brief get_to read into a primitive type
This function provides the interface to the used input adapter. It does
not throw in case the input reached EOF, but returns false instead
@return bool, whether the read was successful
*/
template<class T>
bool get_to(T& dest, const input_format_t format, const char* context)
{
auto new_chars_read = ia.get_elements(&dest);
chars_read += new_chars_read;
if (JSON_HEDLEY_UNLIKELY(new_chars_read < sizeof(T)))
{
// in case of failure, advance position by 1 to report failing location
++chars_read;
sax->parse_error(chars_read, "<end of file>", parse_error::create(110, chars_read, exception_message(format, "unexpected end of input", context), nullptr));
return false;
}
return true;
}
/*!
@return character read from the input after ignoring all 'N' entries
*/
@ -12021,6 +12092,28 @@ class binary_reader
return current;
}
template<class NumberType>
static void byte_swap(NumberType& number)
{
constexpr std::size_t sz = sizeof(number);
#ifdef __cpp_lib_byteswap
if constexpr (sz == 1)
{
return;
}
if constexpr(std::is_integral_v<NumberType>)
{
number = std::byteswap(number);
return;
}
#endif
auto* ptr = reinterpret_cast<std::uint8_t*>(&number);
for (std::size_t i = 0; i < sz / 2; ++i)
{
std::swap(ptr[i], ptr[sz - i - 1]);
}
}
/*
@brief read a number from the input
@ -12039,29 +12132,16 @@ class binary_reader
template<typename NumberType, bool InputIsLittleEndian = false>
bool get_number(const input_format_t format, NumberType& result)
{
// step 1: read input into array with system's byte order
std::array<std::uint8_t, sizeof(NumberType)> vec{};
for (std::size_t i = 0; i < sizeof(NumberType); ++i)
// read in the original format
if (JSON_HEDLEY_UNLIKELY(!get_to(result, format, "number")))
{
get();
if (JSON_HEDLEY_UNLIKELY(!unexpect_eof(format, "number")))
{
return false;
}
// reverse byte order prior to conversion if necessary
if (is_little_endian != (InputIsLittleEndian || format == input_format_t::bjdata))
{
vec[sizeof(NumberType) - i - 1] = static_cast<std::uint8_t>(current);
}
else
{
vec[i] = static_cast<std::uint8_t>(current); // LCOV_EXCL_LINE
}
return false;
}
if (is_little_endian != (InputIsLittleEndian || format == input_format_t::bjdata))
{
byte_swap(result);
}
// step 2: convert array into number of type T and return
std::memcpy(&result, vec.data(), sizeof(NumberType));
return true;
}

40
tests/benchmarks/src/benchmarks.cpp
View File

@ -140,6 +140,46 @@ BENCHMARK_CAPTURE(ToCbor, signed_ints, TEST_DATA_DIRECTORY "/regression/si
BENCHMARK_CAPTURE(ToCbor, unsigned_ints, TEST_DATA_DIRECTORY "/regression/unsigned_ints.json");
BENCHMARK_CAPTURE(ToCbor, small_signed_ints, TEST_DATA_DIRECTORY "/regression/small_signed_ints.json");
//////////////////////////////////////////////////////////////////////////////
// Parse Msgpack
//////////////////////////////////////////////////////////////////////////////
static void FromMsgpack(benchmark::State& state, const char* filename)
{
std::ifstream f(filename);
std::string str((std::istreambuf_iterator<char>(f)), std::istreambuf_iterator<char>());
auto bytes = json::to_msgpack(json::parse(str));
std::ofstream o("test.msgpack");
o.write((char*)bytes.data(), bytes.size());
o.flush();
o.close();
for (auto _ : state)
{
state.PauseTiming();
auto* j = new json();
auto file = fopen("test.msgpack", "rb");
state.ResumeTiming();
*j = json::from_msgpack(file);
state.PauseTiming();
fclose(file);
delete j;
state.ResumeTiming();
}
state.SetBytesProcessed(state.iterations() * bytes.size());
}
BENCHMARK_CAPTURE(FromMsgpack, jeopardy, TEST_DATA_DIRECTORY "/jeopardy/jeopardy.json");
BENCHMARK_CAPTURE(FromMsgpack, canada, TEST_DATA_DIRECTORY "/nativejson-benchmark/canada.json");
BENCHMARK_CAPTURE(FromMsgpack, citm_catalog, TEST_DATA_DIRECTORY "/nativejson-benchmark/citm_catalog.json");
BENCHMARK_CAPTURE(FromMsgpack, twitter, TEST_DATA_DIRECTORY "/nativejson-benchmark/twitter.json");
BENCHMARK_CAPTURE(FromMsgpack, floats, TEST_DATA_DIRECTORY "/regression/floats.json");
BENCHMARK_CAPTURE(FromMsgpack, signed_ints, TEST_DATA_DIRECTORY "/regression/signed_ints.json");
BENCHMARK_CAPTURE(FromMsgpack, unsigned_ints, TEST_DATA_DIRECTORY "/regression/unsigned_ints.json");
BENCHMARK_CAPTURE(FromMsgpack, small_signed_ints, TEST_DATA_DIRECTORY "/regression/small_signed_ints.json");
//////////////////////////////////////////////////////////////////////////////
// serialize binary CBOR
//////////////////////////////////////////////////////////////////////////////

16
tests/src/unit-msgpack.cpp
View File

@ -1508,6 +1508,22 @@ TEST_CASE("MessagePack")
CHECK(json::from_msgpack(std::vector<uint8_t>({0xc4}), true, false).is_discarded());
}
SECTION("unexpected end inside int with stream")
{
json _;
const std::string data = {static_cast<char>(0xd2u), static_cast<char>(0x12u), static_cast<char>(0x34u), static_cast<char>(0x56u)};
CHECK_THROWS_WITH_AS(_ = json::from_msgpack(std::istringstream(data, std::ios::binary)),
"[json.exception.parse_error.110] parse error at byte 5: syntax error while parsing MessagePack number: unexpected end of input", json::parse_error&);
}
SECTION("misuse wchar for binary")
{
json _;
// creates 0xd2 after UTF-8 decoding, triggers get_elements in wide_string_input_adapter for code coverage
const std::u32string data = {static_cast<char32_t>(0x0280)};
CHECK_THROWS_WITH_AS(_ = json::from_msgpack(data),
"[json.exception.parse_error.112] parse error at byte 1: wide string type cannot be interpreted as binary data", json::parse_error&);
}
SECTION("unsupported bytes")
{
SECTION("concrete examples")