Merge branch 'json-view/23-zmij' into json-view/08-view-builder

Signed-off-by: Niels Lohmann <mail@nlohmann.me>
This commit is contained in:
Niels Lohmann committed 2026-10-09 16:40:42 +02:00
commit 63dc0dce12
10 files changed
+232 -311

No files matched your search

+1 -1
View File
@@ -1402,7 +1402,7 @@ THE SOFTWARE IS PROVIDED “AS IS”, WITHOUT WARRANTY OF ANY KIND, EXPRESS OR I
- The class contains the UTF-8 Decoder from Bjoern Hoehrmann which is licensed under the [MIT License](https://opensource.org/licenses/MIT) (see above). Copyright &copy; 2008-2009 [Björn Hoehrmann](https://bjoern.hoehrmann.de/) <bjoern@hoehrmann.de>
- The class contains a slightly modified version of the Grisu2 algorithm from Florian Loitsch which is licensed under the [MIT License](https://opensource.org/licenses/MIT) (see above). Copyright &copy; 2009 [Florian Loitsch](https://florian.loitsch.com/)
- The class contains a port of the shortest double-to-decimal conversion of [Żmij](https://github.com/vitaut/zmij) by Victor Zverovich, which is licensed under the [MIT License](https://opensource.org/licenses/MIT) (see above). Copyright &copy; 2025 [Victor Zverovich](https://github.com/vitaut)
- The class contains a port of the shortest double-to-decimal conversion of [Żmij](https://github.com/vitaut/zmij) by Victor Zverovich, including the conversion of the digits to text by Xiang JunBo and the SIMD instruction sequence of Dougall Johnson, which is licensed under the [MIT License](https://opensource.org/licenses/MIT) (see above). Copyright &copy; 2025 [Victor Zverovich](https://github.com/vitaut)
- The class contains a copy of [Hedley](https://nemequ.github.io/hedley/) from Evan Nemerson which is licensed as [CC0-1.0](https://creativecommons.org/publicdomain/zero/1.0/).
- The class contains parts of [Google Abseil](https://github.com/abseil/abseil-cpp) which is licensed under the [Apache 2.0 License](https://opensource.org/licenses/Apache-2.0).
- The class contains an adapted version of the Eisel-Lemire algorithm, its table of powers of five, and its digit comparison for long numbers from [fast_float](https://github.com/fastfloat/fast_float) by Daniel Lemire and contributors, which is available under the [MIT License](https://opensource.org/licenses/MIT) (used here), the Apache 2.0 License, and the Boost Software License. Copyright &copy; 2021 The fast_float authors
@@ -23,10 +23,10 @@ type to use.
## Template parameters
`NumberFloatType`
: the type to store floating-point numbers. The parser converts `#!cpp float`, `#!cpp double`, and a
`#!cpp long double` that is IEEE 754 binary64 itself and other `#!cpp long double` formats with
`#!cpp std::from_chars` or `#!cpp std::strtold`, and serialization falls back to `#!cpp std::snprintf`, so the
type must be `#!cpp float`, `#!cpp double`, or `#!cpp long double`. The
: the type to store floating-point numbers. The type must be `#!cpp float`, `#!cpp double`, or
`#!cpp long double`. The parser converts `#!cpp float`, `#!cpp double`, and a `#!cpp long double` that is IEEE 754
binary64 itself. It converts other `#!cpp long double` formats with `#!cpp std::from_chars` where available, or
with `#!cpp std::strtold` otherwise. Serialization falls back to `#!cpp std::snprintf`. The
[binary formats](../../features/binary_formats/index.md) additionally require `#!cpp float` or `#!cpp double`,
because they have no encoding for `#!cpp long double`. See
[Template Parameter Requirements](../../features/types/template_parameters.md#numberfloattype).
@@ -353,16 +353,21 @@ using array_t = ArrayType<basic_json, AllocatorType<basic_json>>;
### Always required
- A member type `value_type` that is one byte wide and `char`-compatible. The library stores and processes UTF-8
encoded `char` data and passes `data()` to functions that take a `#!cpp const char*`, such as `#!cpp std::strtod`.
encoded `char` data and passes `data()` to functions that take a `#!cpp const char*`, such as `#!cpp std::strtold`
(only used to parse a `#!cpp long double` that is not IEEE 754 binary64, see
[`NumberFloatType`](#numberfloattype)).
`#!cpp std::wstring`, `#!cpp std::u16string`, and `#!cpp std::u32string` are **not** valid choices; see the FAQ on
[wide string handling](../../home/faq.md#wide-string-handling).
- Constructors: default, copy, move, from `#!cpp const char*` (which must not be `#!cpp explicit`), from
`#!cpp (const char*, size_type)`, and from `#!cpp (size_type, char)`; and copy or move assignment.
- Member functions `size()`, `clear()`, `resize(n, c)`, `data()`, `push_back(char)`, and `operator[]`
(const and non-const, returning references). `c_str()` and `back()` are **not** required.
- `data()` must return a pointer to a contiguous, **null-terminated** buffer -- the parser may hand it to
`#!cpp std::strtod`, which reads up to the null character. A type whose `data()` is not null-terminated does not
fail to compile; it can silently misparse floating-point numbers.
- `data()` must return a pointer to a contiguous, **null-terminated** buffer. `#!cpp float`, `#!cpp double`, and a
`#!cpp long double` that is IEEE 754 binary64 are converted by the library itself and do not depend on this. For any
other `NumberFloatType` (a `#!cpp long double` of another format), the parser falls back to `#!cpp std::strtold` when
`#!cpp std::from_chars` is not available or declines the token, and `std::strtold` reads up to the null character. A type whose `data()`
is not null-terminated does not fail to compile; with such a `NumberFloatType` it can silently misparse
floating-point numbers.
- `append(const char*, size_type)`, used by [`dump`](../../api/basic_json/dump.md), and `append(const StringType&)`,
used by the CBOR reader for indefinite-length strings. The library's internal string concatenation additionally has
to append a `#!cpp char` and a `#!cpp const char*`; for each it selects between `append(arg)`, `#!cpp operator+=`,
+1 -1
View File
@@ -18,7 +18,7 @@ The class contains the UTF-8 Decoder from Bjoern Hoehrmann which is licensed und
The class contains a slightly modified version of the Grisu2 algorithm from Florian Loitsch which is licensed under the [MIT License](https://opensource.org/licenses/MIT) (see above). Copyright &copy; 2009 [Florian Loitsch](https://florian.loitsch.com/)
The class contains a port of the shortest double-to-decimal conversion of [Żmij](https://github.com/vitaut/zmij) by Victor Zverovich, which is licensed under the [MIT License](https://opensource.org/licenses/MIT) (see above). Copyright &copy; 2025 [Victor Zverovich](https://github.com/vitaut)
The class contains a port of the shortest double-to-decimal conversion of [Żmij](https://github.com/vitaut/zmij) by Victor Zverovich, including the conversion of the digits to text by Xiang JunBo and the SIMD instruction sequence of Dougall Johnson, which is licensed under the [MIT License](https://opensource.org/licenses/MIT) (see above). Copyright &copy; 2025 [Victor Zverovich](https://github.com/vitaut)
The class contains a copy of [Hedley](https://nemequ.github.io/hedley/) from Evan Nemerson which is licensed as [CC0-1.0](https://creativecommons.org/publicdomain/zero/1.0/).
+20 -5
View File
@@ -9,8 +9,9 @@
#pragma once
#include <cstdint> // uint64_t
#if !defined(__SIZEOF_INT128__) && defined(_MSC_VER) && (defined(_M_X64) || defined(_M_ARM64))
#include <intrin0.h> // __umulh, _umul128
#include <cstring> // memcpy
#if defined(_MSC_VER) && (defined(_M_X64) || defined(_M_ARM64)) && (!defined(__SIZEOF_INT128__) || (!defined(__GNUC__) && !defined(__clang__)))
#include <intrin0.h> // __umulh, _umul128, _BitScanForward64, _BitScanReverse64
#endif
#include <nlohmann/detail/macro_scope.hpp> // JSON_HEDLEY_ALWAYS_INLINE, NLOHMANN_JSON_NAMESPACE_BEGIN
@@ -28,6 +29,10 @@ inline int count_leading_zeros(std::uint64_t x) noexcept
{
#if defined(__GNUC__) || defined(__clang__)
return __builtin_clzll(x);
#elif defined(_MSC_VER) && (defined(_M_X64) || defined(_M_ARM64))
unsigned long index = 0;
_BitScanReverse64(&index, x);
return 63 - static_cast<int>(index);
#else
int n = 0;
for (int shift = 32; shift != 0; shift >>= 1)
@@ -47,6 +52,10 @@ inline int count_trailing_zeros(std::uint64_t x) noexcept
{
#if defined(__GNUC__) || defined(__clang__)
return __builtin_ctzll(x);
#elif defined(_MSC_VER) && (defined(_M_X64) || defined(_M_ARM64))
unsigned long index = 0;
_BitScanForward64(&index, x);
return static_cast<int>(index);
#else
int n = 0;
for (int shift = 32; shift != 0; shift >>= 1)
@@ -94,15 +103,21 @@ inline uint128_parts full_multiplication(std::uint64_t a, std::uint64_t b) noexc
#endif
}
/// eight bytes as a little-endian word (compilers fold this into one load on
/// little-endian targets; always inlined, as GCC otherwise calls it in the
/// number loops)
/// eight bytes as a little-endian word (a single load on little-endian
/// targets; always inlined, as GCC otherwise calls it in the number loops)
JSON_HEDLEY_ALWAYS_INLINE std::uint64_t read_eight_bytes(const unsigned char* b) noexcept
{
#if defined(_MSC_VER) || defined(__x86_64__) || defined(__i386__) || (defined(__BYTE_ORDER__) && defined(__ORDER_LITTLE_ENDIAN__) && __BYTE_ORDER__ == __ORDER_LITTLE_ENDIAN__)
// the byte order already matches (all MSVC targets are little-endian)
std::uint64_t result = 0;
std::memcpy(&result, b, sizeof(result));
return result;
#else
return static_cast<std::uint64_t>(b[0]) | (static_cast<std::uint64_t>(b[1]) << 8u)
| (static_cast<std::uint64_t>(b[2]) << 16u) | (static_cast<std::uint64_t>(b[3]) << 24u)
| (static_cast<std::uint64_t>(b[4]) << 32u) | (static_cast<std::uint64_t>(b[5]) << 40u)
| (static_cast<std::uint64_t>(b[6]) << 48u) | (static_cast<std::uint64_t>(b[7]) << 56u);
#endif
}
/// eight bytes as a little-endian word
+49 -124
View File
@@ -4,6 +4,7 @@
// |_____|_____|_____|_|___| https://github.com/nlohmann/json
//
// SPDX-FileCopyrightText: 2009 Florian Loitsch <https://florian.loitsch.com/>
// SPDX-FileCopyrightText: 2025 Victor Zverovich <https://github.com/vitaut/zmij>
// SPDX-FileCopyrightText: 2013-2026 Niels Lohmann <https://nlohmann.me>
// SPDX-License-Identifier: MIT
@@ -939,88 +940,6 @@ void grisu2(char* buf, int& len, int& decimal_exponent, FloatType value)
grisu2(buf, len, decimal_exponent, w.minus, w.w, w.plus);
}
/*!
@brief the shortest digits of a positive finite float (other than double): Grisu2
*/
template<typename FloatType>
JSON_HEDLEY_NON_NULL(1)
void shortest_digits(char* buf, int& len, int& decimal_exponent, FloatType value)
{
grisu2(buf, len, decimal_exponent, value);
}
/*!
@brief the shortest digits of a positive finite double: the conversion of
Zmij (see zmij.hpp), which always finds the shortest digits that read back as
the same value (Grisu2 does not for about one double in a thousand), and the
closest of them if there are several
v = buf * 10^decimal_exponent, as for grisu2()
*/
JSON_HEDLEY_NON_NULL(1)
inline void shortest_digits(char* buf, int& len, int& decimal_exponent, double value)
{
static_assert(std::numeric_limits<double>::is_iec559 && std::numeric_limits<double>::digits == 53,
"internal error: the conversion of Zmij needs IEEE 754 binary64 doubles");
JSON_ASSERT(std::isfinite(value));
JSON_ASSERT(value > 0);
std::uint64_t bits = 0;
std::memcpy(&bits, &value, sizeof(bits));
zmij::decimal d = zmij::to_decimal(bits);
// without trailing zeros (up to 16): 8, 4, 2, 1 at a time
while (d.significand % 100000000 == 0)
{
d.significand /= 100000000;
d.exponent += 8;
}
if (d.significand % 10000 == 0)
{
d.significand /= 10000;
d.exponent += 4;
}
if (d.significand % 100 == 0)
{
d.significand /= 100;
d.exponent += 2;
}
if (d.significand % 10 == 0)
{
d.significand /= 10;
d.exponent += 1;
}
// at most 17 digits, written from the back two at a time
static constexpr const char* pairs =
"00010203040506070809101112131415161718192021222324252627282930313233343536373839"
"40414243444546474849505152535455565758596061626364656667686970717273747576777879"
"8081828384858687888990919293949596979899";
std::array<char, 20> digits{};
std::size_t n = digits.size();
while (d.significand >= 100)
{
const std::uint64_t two_digits = d.significand % 100; // a variable: GCC calls a cast of the remainder useless where std::uint64_t is std::size_t
const auto i = static_cast<std::size_t>(two_digits) * 2;
d.significand /= 100;
n -= 2;
digits[n] = pairs[i];
digits[n + 1] = pairs[i + 1];
}
if (d.significand >= 10)
{
const auto i = static_cast<std::size_t>(d.significand) * 2;
n -= 2;
digits[n] = pairs[i];
digits[n + 1] = pairs[i + 1];
}
else
{
digits[--n] = static_cast<char>('0' + d.significand);
}
len = static_cast<int>(digits.size() - n);
std::memcpy(buf, digits.data() + n, static_cast<std::size_t>(len));
decimal_exponent = d.exponent;
}
/*!
@brief appends a decimal representation of e to buf
@return a pointer to the element following the exponent.
@@ -1423,53 +1342,30 @@ inline char* write_shortest(char* first, const zmij::shortest_decimal d) noexcep
return end + (three ? 5 : 4);
}
/// the powers of ten up to 10^16
inline const std::array<std::uint64_t, 17>& powers_of_ten_16() noexcept
{
static const std::array<std::uint64_t, 17> powers =
{
{
1u, 10u, 100u, 1000u, 10000u, 100000u, 1000000u, 10000000u, 100000000u, 1000000000u, 10000000000u,
100000000000u, 1000000000000u, 10000000000000u, 100000000000000u, 1000000000000000u, 10000000000000000u
}
};
return powers;
}
/*!
@brief digits * 10^exp, as write_decimal() writes it, for the digits of a
double that need no conversion (count digits, at most 15, the first not 0;
trailing zeros allowed): extended to 16 digits and written by write_shortest()
@brief whether FloatType is an IEEE 754 binary64 type (a double, or a long double
that has the same format, as with MSVC and on Apple's Arm CPUs)
@return a pointer past the text; up to 41 bytes at @a first are written
(some beyond the returned end)
These are the types the conversion of Zmij (see zmij.hpp) is used for; all
others (binary32, or a format the library does not know) use Grisu2.
*/
JSON_HEDLEY_NON_NULL(1)
JSON_HEDLEY_RETURNS_NON_NULL
inline char* write_short_decimal(char* first, std::uint64_t digits, int count, int exp) noexcept
template<typename FloatType>
constexpr bool has_binary64_format() noexcept
{
JSON_ASSERT(digits >= powers_of_ten_16()[static_cast<std::size_t>(count - 1)] && count <= 15);
const int scale = 16 - count;
return write_shortest(first, zmij::shortest_decimal{digits * powers_of_ten_16()[static_cast<std::size_t>(scale)], exp - scale - 1, 0, false});
return std::numeric_limits<FloatType>::is_iec559
&& std::numeric_limits<FloatType>::digits == 53
&& std::numeric_limits<FloatType>::max_exponent == 1024
&& sizeof(FloatType) == sizeof(std::uint64_t);
}
/// as write_short_decimal(), counting the digits (not 0, less than 10^15)
JSON_HEDLEY_NON_NULL(1)
JSON_HEDLEY_RETURNS_NON_NULL
inline char* write_short_decimal(char* first, std::uint64_t digits, int exp) noexcept
{
JSON_ASSERT(digits != 0 && digits < 1000000000000000u);
// floor(log10(2^bits)) + 1 digits, or one less
const int log2_bound = ((64 - count_leading_zeros(digits)) * 1233) >> 12;
const int count = log2_bound + (digits >= powers_of_ten_16()[static_cast<std::size_t>(log2_bound)] ? 1 : 0);
return write_short_decimal(first, digits, count, exp);
}
template<typename FloatType>
struct is_binary64 : std::integral_constant<bool, has_binary64_format<FloatType>()> {};
/// a positive finite float (other than double): Grisu2 and format_buffer()
/// a positive finite float (other than binary64): Grisu2 and format_buffer()
template<typename FloatType>
JSON_HEDLEY_NON_NULL(1, 2)
JSON_HEDLEY_RETURNS_NON_NULL
char* write_positive(char* first, const char* last, FloatType value)
char* write_positive_grisu2(char* first, const char* last, FloatType value)
{
JSON_ASSERT(last - first >= std::numeric_limits<FloatType>::max_digits10);
static_cast<void>(last); // (only used in the assertion)
@@ -1480,7 +1376,7 @@ char* write_positive(char* first, const char* last, FloatType value)
// len is the length of the buffer, i.e., the number of decimal digits.
int len = 0;
int decimal_exponent = 0;
shortest_digits(first, len, decimal_exponent, value);
grisu2(first, len, decimal_exponent, value);
JSON_ASSERT(len <= std::numeric_limits<FloatType>::max_digits10);
@@ -1496,15 +1392,16 @@ char* write_positive(char* first, const char* last, FloatType value)
return format_buffer(first, len, decimal_exponent, kMinExp, kMaxExp);
}
/// a positive finite double: the shortest digits (Zmij), laid out by
/// a positive finite binary64 number: the shortest digits (Zmij), laid out by
/// write_shortest() (through a local buffer if [first, last) is shorter than
/// the 41 bytes it may write)
template<typename FloatType>
JSON_HEDLEY_NON_NULL(1, 2)
JSON_HEDLEY_RETURNS_NON_NULL
inline char* write_positive(char* first, const char* last, double value)
char* write_positive_zmij(char* first, const char* last, FloatType value)
{
static_assert(std::numeric_limits<double>::is_iec559 && std::numeric_limits<double>::digits == 53,
"internal error: the conversion of Zmij needs IEEE 754 binary64 doubles");
static_assert(is_binary64<FloatType>::value,
"internal error: the conversion of Zmij needs IEEE 754 binary64 numbers");
std::uint64_t bits = 0;
std::memcpy(&bits, &value, sizeof(bits));
const zmij::shortest_decimal d = zmij::to_shortest(bits);
@@ -1519,6 +1416,34 @@ inline char* write_positive(char* first, const char* last, double value)
return first + len;
}
/// a positive finite binary64 number: Zmij (as a long double has the format of
/// a double here, its bits are those of the double of the same value)
template<typename FloatType>
JSON_HEDLEY_NON_NULL(1, 2)
JSON_HEDLEY_RETURNS_NON_NULL
char* write_positive(char* first, const char* last, FloatType value, std::true_type /*is_binary64*/)
{
return write_positive_zmij(first, last, value);
}
/// a positive finite float of any other format: Grisu2
template<typename FloatType>
JSON_HEDLEY_NON_NULL(1, 2)
JSON_HEDLEY_RETURNS_NON_NULL
char* write_positive(char* first, const char* last, FloatType value, std::false_type /*is_binary64*/)
{
return write_positive_grisu2(first, last, value);
}
/// a positive finite float: Zmij for binary64 numbers, Grisu2 otherwise
template<typename FloatType>
JSON_HEDLEY_NON_NULL(1, 2)
JSON_HEDLEY_RETURNS_NON_NULL
char* write_positive(char* first, const char* last, FloatType value)
{
return write_positive(first, last, value, is_binary64<FloatType> {});
}
} // namespace dtoa_impl
/*!
@@ -36,13 +36,6 @@ computed from the compressed tables of Zmij beyond it.
namespace zmij
{
/// significand * 10^exponent
struct decimal
{
std::uint64_t significand;
int exponent;
};
/// the compressed powers of ten of Zmij
inline const std::array<std::uint64_t, 28>& pow10_minor() noexcept
{
@@ -221,18 +214,6 @@ JSON_HEDLEY_ALWAYS_INLINE shortest_decimal to_shortest(std::uint64_t bits) noexc
return shortest_decimal{integral, dec_exp, static_cast<unsigned char>(digit), !round_up && !round_down};
}
/// The shortest decimal in the rounding interval of a positive finite double
/// given by its bits, as one number. The significand can end in zeros.
inline decimal to_decimal(std::uint64_t bits) noexcept
{
const shortest_decimal d = to_shortest(bits);
if (d.has_digit)
{
return decimal{(d.integral * 10) + d.digit, d.exponent};
}
return decimal{d.integral, d.exponent + 1};
}
} // namespace zmij
} // namespace detail
NLOHMANN_JSON_NAMESPACE_END
+69 -148
View File
@@ -8829,8 +8829,9 @@ NLOHMANN_JSON_NAMESPACE_END
#include <cstdint> // uint64_t
#if !defined(__SIZEOF_INT128__) && defined(_MSC_VER) && (defined(_M_X64) || defined(_M_ARM64))
#include <intrin0.h> // __umulh, _umul128
#include <cstring> // memcpy
#if defined(_MSC_VER) && (defined(_M_X64) || defined(_M_ARM64)) && (!defined(__SIZEOF_INT128__) || (!defined(__GNUC__) && !defined(__clang__)))
#include <intrin0.h> // __umulh, _umul128, _BitScanForward64, _BitScanReverse64
#endif
// #include <nlohmann/detail/macro_scope.hpp>
@@ -8849,6 +8850,10 @@ inline int count_leading_zeros(std::uint64_t x) noexcept
{
#if defined(__GNUC__) || defined(__clang__)
return __builtin_clzll(x);
#elif defined(_MSC_VER) && (defined(_M_X64) || defined(_M_ARM64))
unsigned long index = 0;
_BitScanReverse64(&index, x);
return 63 - static_cast<int>(index);
#else
int n = 0;
for (int shift = 32; shift != 0; shift >>= 1)
@@ -8868,6 +8873,10 @@ inline int count_trailing_zeros(std::uint64_t x) noexcept
{
#if defined(__GNUC__) || defined(__clang__)
return __builtin_ctzll(x);
#elif defined(_MSC_VER) && (defined(_M_X64) || defined(_M_ARM64))
unsigned long index = 0;
_BitScanForward64(&index, x);
return static_cast<int>(index);
#else
int n = 0;
for (int shift = 32; shift != 0; shift >>= 1)
@@ -8915,15 +8924,21 @@ inline uint128_parts full_multiplication(std::uint64_t a, std::uint64_t b) noexc
#endif
}
/// eight bytes as a little-endian word (compilers fold this into one load on
/// little-endian targets; always inlined, as GCC otherwise calls it in the
/// number loops)
/// eight bytes as a little-endian word (a single load on little-endian
/// targets; always inlined, as GCC otherwise calls it in the number loops)
JSON_HEDLEY_ALWAYS_INLINE std::uint64_t read_eight_bytes(const unsigned char* b) noexcept
{
#if defined(_MSC_VER) || defined(__x86_64__) || defined(__i386__) || (defined(__BYTE_ORDER__) && defined(__ORDER_LITTLE_ENDIAN__) && __BYTE_ORDER__ == __ORDER_LITTLE_ENDIAN__)
// the byte order already matches (all MSVC targets are little-endian)
std::uint64_t result = 0;
std::memcpy(&result, b, sizeof(result));
return result;
#else
return static_cast<std::uint64_t>(b[0]) | (static_cast<std::uint64_t>(b[1]) << 8u)
| (static_cast<std::uint64_t>(b[2]) << 16u) | (static_cast<std::uint64_t>(b[3]) << 24u)
| (static_cast<std::uint64_t>(b[4]) << 32u) | (static_cast<std::uint64_t>(b[5]) << 40u)
| (static_cast<std::uint64_t>(b[6]) << 48u) | (static_cast<std::uint64_t>(b[7]) << 56u);
#endif
}
/// eight bytes as a little-endian word
@@ -25117,6 +25132,7 @@ NLOHMANN_JSON_NAMESPACE_END
// |_____|_____|_____|_|___| https://github.com/nlohmann/json
//
// SPDX-FileCopyrightText: 2009 Florian Loitsch <https://florian.loitsch.com/>
// SPDX-FileCopyrightText: 2025 Victor Zverovich <https://github.com/vitaut/zmij>
// SPDX-FileCopyrightText: 2013-2026 Niels Lohmann <https://nlohmann.me>
// SPDX-License-Identifier: MIT
@@ -25192,13 +25208,6 @@ computed from the compressed tables of Zmij beyond it.
namespace zmij
{
/// significand * 10^exponent
struct decimal
{
std::uint64_t significand;
int exponent;
};
/// the compressed powers of ten of Zmij
inline const std::array<std::uint64_t, 28>& pow10_minor() noexcept
{
@@ -25377,18 +25386,6 @@ JSON_HEDLEY_ALWAYS_INLINE shortest_decimal to_shortest(std::uint64_t bits) noexc
return shortest_decimal{integral, dec_exp, static_cast<unsigned char>(digit), !round_up && !round_down};
}
/// The shortest decimal in the rounding interval of a positive finite double
/// given by its bits, as one number. The significand can end in zeros.
inline decimal to_decimal(std::uint64_t bits) noexcept
{
const shortest_decimal d = to_shortest(bits);
if (d.has_digit)
{
return decimal{(d.integral * 10) + d.digit, d.exponent};
}
return decimal{d.integral, d.exponent + 1};
}
} // namespace zmij
} // namespace detail
NLOHMANN_JSON_NAMESPACE_END
@@ -26296,88 +26293,6 @@ void grisu2(char* buf, int& len, int& decimal_exponent, FloatType value)
grisu2(buf, len, decimal_exponent, w.minus, w.w, w.plus);
}
/*!
@brief the shortest digits of a positive finite float (other than double): Grisu2
*/
template<typename FloatType>
JSON_HEDLEY_NON_NULL(1)
void shortest_digits(char* buf, int& len, int& decimal_exponent, FloatType value)
{
grisu2(buf, len, decimal_exponent, value);
}
/*!
@brief the shortest digits of a positive finite double: the conversion of
Zmij (see zmij.hpp), which always finds the shortest digits that read back as
the same value (Grisu2 does not for about one double in a thousand), and the
closest of them if there are several
v = buf * 10^decimal_exponent, as for grisu2()
*/
JSON_HEDLEY_NON_NULL(1)
inline void shortest_digits(char* buf, int& len, int& decimal_exponent, double value)
{
static_assert(std::numeric_limits<double>::is_iec559 && std::numeric_limits<double>::digits == 53,
"internal error: the conversion of Zmij needs IEEE 754 binary64 doubles");
JSON_ASSERT(std::isfinite(value));
JSON_ASSERT(value > 0);
std::uint64_t bits = 0;
std::memcpy(&bits, &value, sizeof(bits));
zmij::decimal d = zmij::to_decimal(bits);
// without trailing zeros (up to 16): 8, 4, 2, 1 at a time
while (d.significand % 100000000 == 0)
{
d.significand /= 100000000;
d.exponent += 8;
}
if (d.significand % 10000 == 0)
{
d.significand /= 10000;
d.exponent += 4;
}
if (d.significand % 100 == 0)
{
d.significand /= 100;
d.exponent += 2;
}
if (d.significand % 10 == 0)
{
d.significand /= 10;
d.exponent += 1;
}
// at most 17 digits, written from the back two at a time
static constexpr const char* pairs =
"00010203040506070809101112131415161718192021222324252627282930313233343536373839"
"40414243444546474849505152535455565758596061626364656667686970717273747576777879"
"8081828384858687888990919293949596979899";
std::array<char, 20> digits{};
std::size_t n = digits.size();
while (d.significand >= 100)
{
const std::uint64_t two_digits = d.significand % 100; // a variable: GCC calls a cast of the remainder useless where std::uint64_t is std::size_t
const auto i = static_cast<std::size_t>(two_digits) * 2;
d.significand /= 100;
n -= 2;
digits[n] = pairs[i];
digits[n + 1] = pairs[i + 1];
}
if (d.significand >= 10)
{
const auto i = static_cast<std::size_t>(d.significand) * 2;
n -= 2;
digits[n] = pairs[i];
digits[n + 1] = pairs[i + 1];
}
else
{
digits[--n] = static_cast<char>('0' + d.significand);
}
len = static_cast<int>(digits.size() - n);
std::memcpy(buf, digits.data() + n, static_cast<std::size_t>(len));
decimal_exponent = d.exponent;
}
/*!
@brief appends a decimal representation of e to buf
@return a pointer to the element following the exponent.
@@ -26780,53 +26695,30 @@ inline char* write_shortest(char* first, const zmij::shortest_decimal d) noexcep
return end + (three ? 5 : 4);
}
/// the powers of ten up to 10^16
inline const std::array<std::uint64_t, 17>& powers_of_ten_16() noexcept
{
static const std::array<std::uint64_t, 17> powers =
{
{
1u, 10u, 100u, 1000u, 10000u, 100000u, 1000000u, 10000000u, 100000000u, 1000000000u, 10000000000u,
100000000000u, 1000000000000u, 10000000000000u, 100000000000000u, 1000000000000000u, 10000000000000000u
}
};
return powers;
}
/*!
@brief digits * 10^exp, as write_decimal() writes it, for the digits of a
double that need no conversion (count digits, at most 15, the first not 0;
trailing zeros allowed): extended to 16 digits and written by write_shortest()
@brief whether FloatType is an IEEE 754 binary64 type (a double, or a long double
that has the same format, as with MSVC and on Apple's Arm CPUs)
@return a pointer past the text; up to 41 bytes at @a first are written
(some beyond the returned end)
These are the types the conversion of Zmij (see zmij.hpp) is used for; all
others (binary32, or a format the library does not know) use Grisu2.
*/
JSON_HEDLEY_NON_NULL(1)
JSON_HEDLEY_RETURNS_NON_NULL
inline char* write_short_decimal(char* first, std::uint64_t digits, int count, int exp) noexcept
template<typename FloatType>
constexpr bool has_binary64_format() noexcept
{
JSON_ASSERT(digits >= powers_of_ten_16()[static_cast<std::size_t>(count - 1)] && count <= 15);
const int scale = 16 - count;
return write_shortest(first, zmij::shortest_decimal{digits * powers_of_ten_16()[static_cast<std::size_t>(scale)], exp - scale - 1, 0, false});
return std::numeric_limits<FloatType>::is_iec559
&& std::numeric_limits<FloatType>::digits == 53
&& std::numeric_limits<FloatType>::max_exponent == 1024
&& sizeof(FloatType) == sizeof(std::uint64_t);
}
/// as write_short_decimal(), counting the digits (not 0, less than 10^15)
JSON_HEDLEY_NON_NULL(1)
JSON_HEDLEY_RETURNS_NON_NULL
inline char* write_short_decimal(char* first, std::uint64_t digits, int exp) noexcept
{
JSON_ASSERT(digits != 0 && digits < 1000000000000000u);
// floor(log10(2^bits)) + 1 digits, or one less
const int log2_bound = ((64 - count_leading_zeros(digits)) * 1233) >> 12;
const int count = log2_bound + (digits >= powers_of_ten_16()[static_cast<std::size_t>(log2_bound)] ? 1 : 0);
return write_short_decimal(first, digits, count, exp);
}
template<typename FloatType>
struct is_binary64 : std::integral_constant<bool, has_binary64_format<FloatType>()> {};
/// a positive finite float (other than double): Grisu2 and format_buffer()
/// a positive finite float (other than binary64): Grisu2 and format_buffer()
template<typename FloatType>
JSON_HEDLEY_NON_NULL(1, 2)
JSON_HEDLEY_RETURNS_NON_NULL
char* write_positive(char* first, const char* last, FloatType value)
char* write_positive_grisu2(char* first, const char* last, FloatType value)
{
JSON_ASSERT(last - first >= std::numeric_limits<FloatType>::max_digits10);
static_cast<void>(last); // (only used in the assertion)
@@ -26837,7 +26729,7 @@ char* write_positive(char* first, const char* last, FloatType value)
// len is the length of the buffer, i.e., the number of decimal digits.
int len = 0;
int decimal_exponent = 0;
shortest_digits(first, len, decimal_exponent, value);
grisu2(first, len, decimal_exponent, value);
JSON_ASSERT(len <= std::numeric_limits<FloatType>::max_digits10);
@@ -26853,15 +26745,16 @@ char* write_positive(char* first, const char* last, FloatType value)
return format_buffer(first, len, decimal_exponent, kMinExp, kMaxExp);
}
/// a positive finite double: the shortest digits (Zmij), laid out by
/// a positive finite binary64 number: the shortest digits (Zmij), laid out by
/// write_shortest() (through a local buffer if [first, last) is shorter than
/// the 41 bytes it may write)
template<typename FloatType>
JSON_HEDLEY_NON_NULL(1, 2)
JSON_HEDLEY_RETURNS_NON_NULL
inline char* write_positive(char* first, const char* last, double value)
char* write_positive_zmij(char* first, const char* last, FloatType value)
{
static_assert(std::numeric_limits<double>::is_iec559 && std::numeric_limits<double>::digits == 53,
"internal error: the conversion of Zmij needs IEEE 754 binary64 doubles");
static_assert(is_binary64<FloatType>::value,
"internal error: the conversion of Zmij needs IEEE 754 binary64 numbers");
std::uint64_t bits = 0;
std::memcpy(&bits, &value, sizeof(bits));
const zmij::shortest_decimal d = zmij::to_shortest(bits);
@@ -26876,6 +26769,34 @@ inline char* write_positive(char* first, const char* last, double value)
return first + len;
}
/// a positive finite binary64 number: Zmij (as a long double has the format of
/// a double here, its bits are those of the double of the same value)
template<typename FloatType>
JSON_HEDLEY_NON_NULL(1, 2)
JSON_HEDLEY_RETURNS_NON_NULL
char* write_positive(char* first, const char* last, FloatType value, std::true_type /*is_binary64*/)
{
return write_positive_zmij(first, last, value);
}
/// a positive finite float of any other format: Grisu2
template<typename FloatType>
JSON_HEDLEY_NON_NULL(1, 2)
JSON_HEDLEY_RETURNS_NON_NULL
char* write_positive(char* first, const char* last, FloatType value, std::false_type /*is_binary64*/)
{
return write_positive_grisu2(first, last, value);
}
/// a positive finite float: Zmij for binary64 numbers, Grisu2 otherwise
template<typename FloatType>
JSON_HEDLEY_NON_NULL(1, 2)
JSON_HEDLEY_RETURNS_NON_NULL
char* write_positive(char* first, const char* last, FloatType value)
{
return write_positive(first, last, value, is_binary64<FloatType> {});
}
} // namespace dtoa_impl
/*!
+7
View File
@@ -1792,4 +1792,11 @@ TEST_CASE("string scanning kernels")
CHECK(nlohmann::detail::count_trailing_zeros(bit) == k);
CHECK(nlohmann::detail::count_trailing_zeros(bit | (bit << 1u) | 0x8000000000000000u) == k);
}
// eight bytes as a little-endian word, at any alignment
const unsigned char bytes[16] = {0x01, 0x02, 0x03, 0x04, 0x05, 0x06, 0x07, 0x08, 0x09, 0x0A, 0x0B, 0x0C, 0x0D, 0x0E, 0x0F, 0xFF};
CHECK(nlohmann::detail::read_eight_bytes(bytes) == 0x0807060504030201u);
CHECK(nlohmann::detail::read_eight_bytes(bytes + 1) == 0x0908070605040302u);
CHECK(nlohmann::detail::read_eight_bytes(bytes + 8) == 0xFF0F0E0D0C0B0A09u);
CHECK(nlohmann::detail::read_eight_bytes(reinterpret_cast<const char*>(bytes) + 3) == 0x0B0A090807060504u);
}
+72 -5
View File
@@ -15,6 +15,7 @@
#include <nlohmann/json.hpp>
using nlohmann::detail::dtoa_impl::reinterpret_bits;
#include <algorithm>
#include <array>
#include <cmath>
#include <cstdint>
@@ -666,13 +667,24 @@ void check_shortest(double v)
const std::string text(buf.data(), end);
CAPTURE(text)
CHECK(parse_double(text) == v);
// the layout is that of format_buffer() for the same digits
// the layout is that of format_buffer() for the digits of Zmij
const auto sd = nlohmann::detail::zmij::to_shortest(reinterpret_bits<std::uint64_t>(v));
const std::uint64_t significand = sd.has_digit ? (sd.integral * 10) + sd.digit : sd.integral;
int exponent = sd.has_digit ? sd.exponent : sd.exponent + 1;
std::string significand_digits = std::to_string(significand);
while (significand_digits.size() > 1 && significand_digits.back() == '0')
{
significand_digits.pop_back();
++exponent;
}
std::array<char, 64> reference{};
int len = 0;
int exponent = 0;
nlohmann::detail::dtoa_impl::shortest_digits(reference.data(), len, exponent, v);
const char* const reference_end = nlohmann::detail::dtoa_impl::format_buffer(reference.data(), len, exponent, -4, 15);
std::copy(significand_digits.begin(), significand_digits.end(), reference.begin());
const char* const reference_end = nlohmann::detail::dtoa_impl::format_buffer(reference.data(), static_cast<int>(significand_digits.size()), exponent, -4, 15);
CHECK(text == std::string(reference.data(), static_cast<std::size_t>(reference_end - reference.data())));
// and write_positive() is what to_chars() calls
std::array<char, 64> positive{};
const char* const positive_end = nlohmann::detail::dtoa_impl::write_positive(positive.data(), positive.data() + positive.size(), v);
CHECK(text == std::string(positive.data(), static_cast<std::size_t>(positive_end - positive.data())));
const auto de = digits_and_exponent(text);
const std::string& digits = de.first;
if (digits.size() > 1)
@@ -785,3 +797,58 @@ TEST_CASE("shortest digits of doubles")
}
}
}
TEST_CASE("choice of the conversion")
{
using nlohmann::detail::dtoa_impl::is_binary64;
SECTION("by the format of the type")
{
// Zmij needs binary64 numbers; everything else uses Grisu2
static_assert(!is_binary64<float>::value, "float is not binary64");
static_assert(is_binary64<double>::value == (std::numeric_limits<double>::is_iec559 && std::numeric_limits<double>::digits == 53),
"double is binary64 where it is IEEE 754 with 53 digits");
static_assert(!is_binary64<int>::value, "integers are not binary64");
static_assert(is_binary64<long double>::value == (std::numeric_limits<long double>::is_iec559 && std::numeric_limits<long double>::digits == 53 && sizeof(long double) == 8),
"long double is binary64 where it has the format of a double");
CHECK(!is_binary64<float>::value);
CHECK(is_binary64<double>::value);
}
SECTION("float: Grisu2, double: Zmij")
{
// 5.3165205877497296e+16 is one of the doubles for which Grisu2 does not find the shortest digits
constexpr double value = 5.3165205877497296e+16;
std::array<char, 64> buf{};
const char* const last = buf.data() + buf.size();
char* end = nlohmann::detail::dtoa_impl::write_positive(buf.data(), last, value);
CHECK(std::string(buf.data(), end) == "5.31652058774973e+16");
end = nlohmann::detail::dtoa_impl::write_positive_grisu2(buf.data(), last, value);
CHECK(std::string(buf.data(), end) == "5.3165205877497296e+16");
constexpr float f = 1.1754944e-38f;
end = nlohmann::detail::dtoa_impl::write_positive(buf.data(), last, f);
const std::string dispatched(buf.data(), end);
end = nlohmann::detail::dtoa_impl::write_positive_grisu2(buf.data(), last, f);
CHECK(dispatched == std::string(buf.data(), end));
}
SECTION("long double with the format of a double: Zmij")
{
// (on platforms where long double is wider, Grisu2 does not apply either: the snprintf fallback does)
if (std::numeric_limits<long double>::digits == 53 && std::numeric_limits<long double>::is_iec559)
{
using long_double_json = nlohmann::json::with_float_t<long double>;
for (const double d :
{
5.3165205877497296e+16, 1.0, 0.1, 123456.789, 2.2250738585072014e-308, 1.7976931348623157e+308, -5.3165205877497296e+16
})
{
CAPTURE(d)
CHECK(long_double_json(static_cast<long double>(d)).dump() == nlohmann::json(d).dump());
}
CHECK(long_double_json(5.3165205877497296e+16L).dump() == "5.31652058774973e+16");
}
}
}