Merge remote-tracking branch 'origin/develop' into claude/fix-issue-3989-db7e45

Signed-off-by: Niels Lohmann <mail@nlohmann.me>

# Conflicts:
#	include/nlohmann/detail/input/binary_reader.hpp
#	include/nlohmann/json.hpp
#	single_include/nlohmann/json.hpp
This commit is contained in:
Niels Lohmann committed 2026-09-30 22:53:03 +02:00
commit 7e8e8e219b
55 files changed
+3174 -2080

No files matched your search

@@ -159,6 +159,8 @@ basic_json(basic_json&& other) noexcept;
- `CompatibleType` is not `basic_json` (to avoid hijacking copy/move constructors),
- `CompatibleType` is not a different `basic_json` type (i.e. with different template arguments)
- `CompatibleType` is not a `basic_json` nested type (e.g., `json_pointer`, `iterator`, etc.)
- if [`JSON_DISABLE_TUPLE_REFERENCE_CONVERSION`](../macros/json_disable_tuple_reference_conversion.md) is defined
to `1`: `CompatibleType` is not a one-element `std::tuple` holding a reference to `basic_json`
- `json_serializer<U>` (with `U = uncvref_t<CompatibleType>`) has a `to_json(basic_json_t&, CompatibleType&&)`
method
@@ -51,7 +51,7 @@ range will yield over/underflow when used in a constructor. During deserializati
will automatically be stored as [`number_unsigned_t`](number_unsigned_t.md) or [`number_float_t`](number_float_t.md).
[RFC 8259](https://tools.ietf.org/html/rfc8259) further states:
> Note that when such software is used, numbers that are integers and are in the range $[-2^{53}+1, 2^{53}-1]$ are
> Note that when such software is used, numbers that are integers and are in the range [-2<sup>53</sup>+1, 2<sup>53</sup>-1] are
> interoperable in the sense that implementations will agree exactly on their numeric values.
As this range is a subrange of the exactly supported range [INT64_MIN, INT64_MAX], this class's integer type is
@@ -52,7 +52,7 @@ when used in a constructor. During deserialization, too large or small integer n
as [`number_integer_t`](number_integer_t.md) or [`number_float_t`](number_float_t.md).
[RFC 8259](https://tools.ietf.org/html/rfc8259) further states:
> Note that when such software is used, numbers that are integers and are in the range $[-2^{53}+1, 2^{53}-1]$ are
> Note that when such software is used, numbers that are integers and are in the range [-2<sup>53</sup>+1, 2<sup>53</sup>-1] are
> interoperable in the sense that implementations will agree exactly on their numeric values.
As this range is a subrange (when considered in conjunction with the `number_integer_t` type) of the exactly supported
@@ -54,7 +54,7 @@ classDiagram
## Notes
For an input with $n$ bytes, 1 is the index of the first character and $n+1$ is the index of the terminating null byte
For an input with <i>n</i> bytes, 1 is the index of the first character and <i>n</i>+1 is the index of the terminating null byte
or the end of file. This also holds true when reading a byte vector for binary formats.
## Examples
+1
View File
@@ -53,6 +53,7 @@ header. See also the [macro overview page](../../features/macros.md).
- [**JSON_BRACE_INIT_COPY_SEMANTICS**](json_brace_init_copy_semantics.md) - opt in to copy/move semantics for single-element brace initialization
- [**JSON_DISABLE_ENUM_SERIALIZATION**](json_disable_enum_serialization.md) - switch off default serialization/deserialization functions for enums
- [**JSON_DISABLE_TUPLE_REFERENCE_CONVERSION**](json_disable_tuple_reference_conversion.md) - switch off conversion from a one-element tuple of a JSON reference
- [**JSON_USE_IMPLICIT_CONVERSIONS**](json_use_implicit_conversions.md) - control implicit conversions
## Comparison behavior
@@ -0,0 +1,114 @@
# JSON_DISABLE_TUPLE_REFERENCE_CONVERSION
```cpp
#define JSON_DISABLE_TUPLE_REFERENCE_CONVERSION /* value */
```
When defined to `1`, a `basic_json` value can no longer be constructed from a one-element `std::tuple` whose element is
a reference to that `basic_json` type, such as `std::tuple<json&>`, `std::tuple<const json&>`, or `std::tuple<json&&>`.
These are the tuples created by `std::forward_as_tuple(j)`.
## Default definition
The default value is `0` (disabled — existing behavior is preserved).
```cpp
#define JSON_DISABLE_TUPLE_REFERENCE_CONVERSION 0
```
## Notes
!!! note "Background"
By default, `basic_json` can be constructed from any `std::tuple` whose elements can be converted to JSON; the result
is an array. This includes `std::tuple<json&>`, which becomes a one-element array.
`std::tuple` only converts another tuple element by element if its element type cannot be constructed from the whole
source tuple. Because `json` *can* be constructed from `std::tuple<json&>`, `std::tuple` instead converts the whole
tuple into a single `json` value. This has two surprising effects:
```cpp
json j = true;
// rejected by some standard libraries (e.g., libc++); with others, the
// reference binds to a temporary that is destroyed right away
std::tuple<const json&> t1(std::forward_as_tuple(j));
// compiles, but std::get<0>(t2) is [true], not true
std::tuple<json> t2(std::forward_as_tuple(j));
```
Enabling this macro removes the conversion, so both tuples are converted element by element: `std::get<0>(t1)`
refers to `j`, and `std::get<0>(t2)` is a copy of `j` (see [#2226](https://github.com/nlohmann/json/issues/2226)).
!!! warning "Opt-in only"
This macro must be defined **before** including `<nlohmann/json.hpp>`. Defining it after the include has no effect.
!!! note "Affected conversions"
Only one-element tuples holding a reference to the **same** `basic_json` type are affected. Constructing a JSON value
from them no longer compiles:
```cpp
json j = true;
json a = std::forward_as_tuple(j); // error with the macro enabled
json b = json::array({j}); // use this instead: [true]
```
Tuples holding a JSON value (`std::make_tuple(j)`), tuples with more than one element, and tuples holding references
to other types (including other `basic_json` specializations) are converted to arrays as before.
!!! hint "CMake option"
This behavior can also be controlled with the CMake option
[`JSON_DisableTupleReferenceConversion`](../../integration/cmake.md#json_disabletuplereferenceconversion)
(`OFF` by default) which defines `JSON_DISABLE_TUPLE_REFERENCE_CONVERSION` accordingly.
## Examples
??? example "Default behavior (macro not defined)"
```cpp
#include <nlohmann/json.hpp>
using json = nlohmann::json;
int main()
{
json j = true;
std::tuple<json> t(std::forward_as_tuple(j));
// std::get<0>(t) is [true] -- the whole tuple was converted
}
```
??? example "Conversion disabled (macro defined to 1)"
```cpp
#define JSON_DISABLE_TUPLE_REFERENCE_CONVERSION 1
#include <nlohmann/json.hpp>
using json = nlohmann::json;
int main()
{
json j = true;
std::tuple<json> t(std::forward_as_tuple(j));
// std::get<0>(t) is true -- a copy of j
std::tuple<const json&> r(std::forward_as_tuple(j));
// std::get<0>(r) refers to j
}
```
## See also
- [**basic_json(CompatibleType&&)**](../basic_json/basic_json.md) - the affected constructor
- [:simple-cmake: JSON_DisableTupleReferenceConversion](../../integration/cmake.md#json_disabletuplereferenceconversion) -
CMake option to control the macro
## Version history
- Added in version 3.13.0.
+6
View File
@@ -18,6 +18,10 @@ Deserializes an input stream to a JSON value.
the stream `i`
## Exception safety
Strong guarantee: if an exception is thrown, there are no changes in `j`.
## Exceptions
- Throws [`parse_error.101`](../home/exceptions.md#jsonexceptionparse_error101) in case of an unexpected token, or if
@@ -125,3 +129,5 @@ being read.
the stream; planned to become the default in version 4.0.0.
- Fixed a null pointer dereference for an `std::istream` without a stream buffer (now throws `parse_error.101`), and a
crash (`std::terminate`) when `i` has `eofbit` in its exception mask, in version 3.13.0.
- Changed to the strong exception safety guarantee in version 3.13.0: `j` is no longer left with a partially parsed
value if parsing throws.