Merge branch 'develop' into copy-scratch-allocator

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

.github/PULL_REQUEST_TEMPLATE.md

@@ -2,6 +2,7 @@
 
 - [ ] The changes are described in detail, both the what and why.
 - [ ] If applicable, an [existing issue](https://github.com/nlohmann/json/issues) is referenced.
+- [ ] If applicable, a fixed [OSS-Fuzz](https://issues.oss-fuzz.com) issue is referenced as `OSS-Fuzz: <id>` (see [fuzz testing](https://github.com/nlohmann/json/blob/develop/tests/fuzzing.md#handling-oss-fuzz-reports)).
 - [ ] The [Code coverage](https://coveralls.io/github/nlohmann/json) remained at 100%. A test case for every new line of code.
 - [ ] If applicable, the [documentation](https://json.nlohmann.me) is updated.
 - [ ] The source code is amalgamated by running `make amalgamate`.

.github/workflows/windows.yml

@@ -124,11 +124,11 @@ jobs:
     steps:
       - uses: actions/checkout@3d3c42e5aac5ba805825da76410c181273ba90b1 # v7.0.1
       - name: Run CMake (Release)
-        run: cmake -S . -B build -G "Visual Studio 17 2022" -A ARM64 -DJSON_BuildTests=On -DCMAKE_CXX_FLAGS="/W4 /WX"
+        run: cmake -S . -B build -G "Visual Studio 18 2026" -A ARM64 -DJSON_BuildTests=On -DCMAKE_CXX_FLAGS="/W4 /WX"
         if: matrix.build_type == 'Release'
         shell: pwsh
       - name: Run CMake (Debug)
-        run: cmake -S . -B build -G "Visual Studio 17 2022" -A ARM64 -DJSON_BuildTests=On -DJSON_FastTests=ON -DCMAKE_CXX_FLAGS="/W4 /WX"
+        run: cmake -S . -B build -G "Visual Studio 18 2026" -A ARM64 -DJSON_BuildTests=On -DJSON_FastTests=ON -DCMAKE_CXX_FLAGS="/W4 /WX"
         if: matrix.build_type == 'Debug'
         shell: pwsh
       - name: Build

BUILD.bazel

@@ -71,7 +71,6 @@ cc_library(
     ],
     includes = ["include"],
     visibility = ["//visibility:public"],
-    alwayslink = True,
 )
 
 cc_library(

cmake/ci.cmake

@@ -300,10 +300,10 @@ add_custom_target(ci_test_skiplibraryversioncheck
 # Disable thread-local storage.
 ###############################################################################
 
-# Without thread-local storage, the copy constructor cannot bound its descent
-# and copies every object and array without the call stack. That path is
-# otherwise only reached by values nested deeper than the bound, so this target
-# is what runs the whole test suite through it.
+# Without thread-local storage, copying and comparing cannot bound their
+# descent and handle every object and array without the call stack. Those paths
+# are otherwise only reached by values nested deeper than the bound, so this
+# target is what runs the whole test suite through them.
 add_custom_target(ci_test_no_thread_local
     COMMAND ${CMAKE_COMMAND}
     -DCMAKE_BUILD_TYPE=Debug -GNinja

cmake/scripts/gen_bazel_build_file.cmake

@@ -42,7 +42,6 @@ string(APPEND CONTENT [=[
     ],
     includes = ["include"],
     visibility = ["//visibility:public"],
-    alwayslink = True,
 )
 
 cc_library(

docs/mkdocs/docs/api/basic_json/get_to.md

@@ -21,6 +21,10 @@ This overload is chosen if:
 - `ValueType` is not `basic_json`,
 - `json_serializer<ValueType>` has a `from_json()` method of the form `void from_json(const basic_json&, ValueType&)`
 
+`v` must not be `const`. Passing a `const` object is a compile-time error. For types such as arithmetic types, enums,
+and C arrays, the error is a `static_assert` that names the problem. For other types, the overload is not viable, and
+the compiler reports that no matching `get_to` was found.
+
 ## Template parameters
 
 `ValueType`
@@ -67,3 +71,4 @@ Depends on the `json_serializer<ValueType>::from_json()` implementation.
 ## Version history
 
 - Since version 3.3.0.
+- Added a `static_assert` with a clear message for `const` arguments in version 3.13.0.

docs/mkdocs/docs/api/basic_json/sax_parse.md

@@ -69,7 +69,9 @@ The SAX event lister must follow the interface of [`json_sax`](../json_sax/index
      [`input_format_t`](input_format_t.md) for more information
 
 `strict` (in)
-:   whether the input has to be consumed completely (optional, `#!cpp true` by default)
+:   whether the input has to be consumed completely (optional, `#!cpp true` by default); when `#!cpp false` and the
+    input is a `#!cpp std::istream`, the character that terminates a number is consumed unless
+    [`JSON_PRECISE_STREAM_POSITION`](../macros/json_precise_stream_position.md) is defined to `1`; see [`operator>>`](../operator_gtgt.md#notes)
 
 `ignore_comments` (in)
 :   whether comments should be ignored and treated like whitespace (`#!cpp true`) or yield a parse error
@@ -136,6 +138,8 @@ A UTF-8 byte order mark is silently ignored.
 - Added `ignore_trailing_commas` in version 3.13.0.
 - Extended container support (1) to include types with lvalue-only ADL `begin`/`end` (matching `std::begin`/`std::end` semantics) in version 3.13.0.
 - Extended overload (2) to accept heterogeneous iterator+sentinel pairs (C++20 ranges support) in version 3.13.0.
+- `JSON_PRECISE_STREAM_POSITION` added in version 3.13.0 to optionally leave a `#!cpp std::istream` positioned right
+  after the parsed value when `strict` is `#!cpp false`.
 
 !!! warning "Deprecation"
 

docs/mkdocs/docs/api/macros/index.md

@@ -16,6 +16,8 @@ header. See also the [macro overview page](../../features/macros.md).
 
 ## Parsing
 
+- [**JSON_PRECISE_STREAM_POSITION**](json_precise_stream_position.md) - opt in to leaving an input stream positioned
+  right after a parsed number
 - [**JSON_STRICT_NUL_HANDLING**](json_strict_nul_handling.md) - opt in to rejecting a NUL byte in the input instead of
   treating it as end of input
 

docs/mkdocs/docs/api/macros/json_no_thread_local.md

@@ -7,16 +7,16 @@
 When defined, the library does not use `#!cpp thread_local` storage. This is relevant for the few environments whose
 toolchain does not support it.
 
-The copy constructor copies the first levels of a value by copying the containers, which copy their elements, and
-completes whatever is nested deeper than that without the call stack, so that copying a value cannot exhaust the stack
-however deeply it is nested. It counts the levels it has descended into in a `#!cpp thread_local` variable, as a counter
-shared between threads would be raced.
+Copying a value and comparing two values both descend into the first levels by letting the containers copy or compare
+themselves, and finish whatever is nested deeper than that without the call stack, so that neither can exhaust the stack
+however deeply the values are nested. Each counts the levels it has descended into in a `#!cpp thread_local` variable, as
+a counter shared between threads would be raced.
 
-Without that counter, no descent can be bounded safely, so objects and arrays are copied without the call stack right
-away. Copying keeps working exactly as it does otherwise - the same values come out, and deeply nested values are copied
-just as safely - but copying is slower, because the containers no longer copy themselves. Copying the benchmark
-documents takes 9% (`canada.json`) to 34% (`twitter.json`) longer; values built mostly from objects are affected the
-most.
+Without those counters, no descent can be bounded safely, so objects and arrays are copied and compared without the call
+stack right away. Both keep working exactly as they do otherwise - the same values come out, the same comparisons hold,
+and deeply nested values are handled just as safely - but both are slower, because the containers no longer copy or
+compare themselves. Copying the benchmark documents takes 9% (`canada.json`) to 34% (`twitter.json`) longer, and
+comparing two equal ones 10% (`citm_catalog.json`) to 90% (`canada.json`) longer.
 
 ## Default definition
 
@@ -28,6 +28,7 @@ By default, `#!cpp JSON_NO_THREAD_LOCAL` is not defined.
 
 The library defines it by itself for Clang targeting MinGW, which does not survive the `#!cpp thread_local` storage:
 copying a value segfaults there, with both old and current Clang versions, while GCC targeting MinGW is unaffected.
+Copying and comparing fall back to working without the call stack there, as they do whenever the macro is defined.
 
 ## Examples
 

docs/mkdocs/docs/api/macros/json_precise_stream_position.md

@@ -0,0 +1,131 @@
+# JSON_PRECISE_STREAM_POSITION
+
+```cpp
+#define JSON_PRECISE_STREAM_POSITION /* value */
+```
+
+When defined to `1`, [`operator>>`](../operator_gtgt.md) and [`sax_parse`](../basic_json/sax_parse.md) with
+`strict = false` leave a `#!cpp std::istream` positioned right after the parsed value for every value type. By default,
+the character that terminates a number is consumed as well.
+
+The macro only affects reading from a `#!cpp std::istream` when the rest of the stream is not required to be consumed.
+[`parse`](../basic_json/parse.md), [`accept`](../basic_json/accept.md), and all other inputs (strings, iterators,
+containers, `#!cpp FILE*`) are never affected.
+
+## Default definition
+
+The default value is `0` (disabled — existing behavior is preserved).
+
+```cpp
+#define JSON_PRECISE_STREAM_POSITION 0
+```
+
+## Notes
+
+!!! note "Background"
+
+    A number is the only JSON value whose end can be detected solely by reading the character that follows it. By
+    default, that character is consumed and not put back, so the stream is left one byte too far after a number, and
+    only after a number:
+
+    ```cpp
+    std::istringstream input("1true");
+    json j;
+    input >> j;  // j == 1, but the stream now starts at "rue"
+    ```
+
+    With this macro, the character is only looked at and left in the stream, so the stream starts at `true`. This
+    does not require the stream buffer to support putting a character back.
+
+    This was not changed unconditionally, because code can depend on the consumed character, even unknowingly (see
+    [#5340](https://github.com/nlohmann/json/issues/5340)). Both of the following work by default only because the
+    character after each number is swallowed, and behave differently with this macro:
+
+    ```cpp
+    std::istringstream input("1,2,3");
+    json j1, j2, j3;
+    input >> j1 >> j2 >> j3;  // default: 1, 2, 3
+                              // with the macro: throws parse_error.101 at the ','
+    ```
+
+    ```cpp
+    std::istringstream input("42\nfoo");
+    json j;
+    std::string line;
+    input >> j;
+    std::getline(input, line);  // default: "foo"
+                                // with the macro: "" (like after reading an int with >>)
+    ```
+
+    In both cases, the behavior with the macro is what you already get today when the value is not a number: `"a","b"`
+    fails at the `,`, and `std::getline` after `{}` returns an empty string. This macro offers an opt-in path to
+    the consistent behavior ahead of version 4.0.0, where it is planned to become the default.
+
+!!! warning "Opt-in only"
+
+    This macro must be defined **before** including `<nlohmann/json.hpp>`. Defining it after the include has no
+    effect.
+
+!!! note "ABI compatibility"
+
+    The value of this macro is encoded in the [namespace](../../features/namespace.md) (tag `_psp`), resulting in
+    distinct symbol names. Translation units compiled with and without it can therefore be linked into the same program
+    without One Definition Rule (ODR) violations, but they cannot exchange instances of library types.
+
+!!! tip "Workaround without the macro"
+
+    Separate the values in the stream with whitespace. The character consumed after a number is then the separator,
+    and whitespace before the next value is skipped anyway.
+
+## Examples
+
+??? example "Default behavior (macro not defined)"
+
+    Without the macro, the character after a number is consumed:
+
+    ```cpp
+    #include <iostream>
+    #include <sstream>
+    #include <nlohmann/json.hpp>
+
+    using json = nlohmann::json;
+
+    int main()
+    {
+        std::istringstream input("1true");
+        json j1, j2;
+        input >> j1;  // j1 == 1
+        input >> j2;  // throws parse_error.101: the stream now starts at "rue"
+    }
+    ```
+
+??? example "Opt-in precise stream position (macro defined to 1)"
+
+    With the macro, the stream is positioned right after the number:
+
+    ```cpp
+    #define JSON_PRECISE_STREAM_POSITION 1
+    #include <iostream>
+    #include <sstream>
+    #include <nlohmann/json.hpp>
+
+    using json = nlohmann::json;
+
+    int main()
+    {
+        std::istringstream input("1true");
+        json j1, j2;
+        input >> j1;  // j1 == 1
+        input >> j2;  // j2 == true
+    }
+    ```
+
+## See also
+
+- [**operator>>**](../operator_gtgt.md) - deserialize from stream
+- [**sax_parse**](../basic_json/sax_parse.md) - generate SAX events
+
+## Version history
+
+- Added in version 3.13.0.
+- Planned to become the default (with the macro removed) in version 4.0.0.

docs/mkdocs/docs/api/operator_gtgt.md

@@ -67,7 +67,9 @@ input >> j2;  // parses the next value
     Only numbers are affected. Values ending in a self-delimiting character do not read past themselves, so
     `truefalse`, `[1][2]`, `{"a":1}{"b":2}`, and `"a""b"` can be read back to back without a separator.
 
-    This is tracked in [#5340](https://github.com/nlohmann/json/issues/5340).
+    Define [`JSON_PRECISE_STREAM_POSITION`](macros/json_precise_stream_position.md) to `1` to leave the terminating character in the stream
+    instead, so that the stream is positioned right after the value for every value type and no separator is
+    needed. This is tracked in [#5340](https://github.com/nlohmann/json/issues/5340).
 
 Note that reading concatenated values does **not** work for [JSON Lines](../features/parsing/json_lines.md)
 (newline-delimited JSON) input -- see that page for why and for the recommended alternative.
@@ -107,9 +109,12 @@ being read.
 - [parse](basic_json/parse.md) - deserialize from a compatible input
 - [`JSON_STRICT_NUL_HANDLING`](macros/json_strict_nul_handling.md) - opt in to rejecting a NUL byte in the input
   instead of treating it as end of input
+- [`JSON_PRECISE_STREAM_POSITION`](macros/json_precise_stream_position.md) - opt in to leaving the stream positioned right after a number
 
 ## Version history
 
 - Added in version 1.0.0.
 - `JSON_STRICT_NUL_HANDLING` added in version 3.13.0 to optionally reject a NUL byte in the input instead of treating
   it as end of input; planned to become the default in version 4.0.0.
+- `JSON_PRECISE_STREAM_POSITION` added in version 3.13.0 to optionally leave the character that terminates a number in
+  the stream; planned to become the default in version 4.0.0.

docs/mkdocs/docs/community/assurance_case.md

@@ -0,0 +1,70 @@
+# Assurance case
+
+This page argues why the library meets its security requirements. It describes the threats the library faces, where the
+trust boundaries lie, and how the library's design and the [quality assurance](quality_assurance.md) counter these
+threats. To report a vulnerability, see the [security policy](security_policy.md).
+
+## Threat model
+
+The library parses, stores, and serializes JSON values in memory. It does not open network connections, does not open
+files (it only reads from streams or `std::FILE*` handles that the caller has already opened), does not read environment
+variables, and does not implement cryptography or handle credentials.
+
+The primary threat is therefore **untrusted input**: JSON text or binary data (BJData, BSON, CBOR, MessagePack, UBJSON)
+that an attacker controls, passed to [`parse`](../api/basic_json/parse.md), [`accept`](../api/basic_json/accept.md),
+[`sax_parse`](../api/basic_json/sax_parse.md), or one of the `from_*` functions such as
+[`from_cbor`](../api/basic_json/from_cbor.md). Such input may try to
+
+- make the library read or write out of bounds (malformed lengths, truncated input, invalid UTF-8),
+- trigger undefined behavior (integer overflow in sizes or numbers, invalid casts),
+- exhaust memory (huge announced sizes), or
+- exhaust the call stack (deeply nested arrays and objects).
+
+## Trust boundaries
+
+- **Untrusted:** all serialized input read by the parser, the SAX interface, and the binary readers. The library must
+  handle every possible input by either producing a value or throwing a [`parse_error`](../home/exceptions.md#parse-errors)
+  (or returning `false` when exceptions are disabled for the call).
+- **Trusted:** the C++ code that calls the library. Calling a function with violated preconditions, for instance
+  accessing an array with [`operator[]`](../api/basic_json/operator%5B%5D.md) out of range, is a programming error and
+  not a security boundary. Such preconditions are checked with [runtime assertions](../features/assertions.md) in debug
+  builds; functions such as [`at`](../api/basic_json/at.md) offer checked access with exceptions.
+
+## Secure design
+
+- **Strict parsing.** The parser accepts exactly the JSON grammar of [RFC 8259](https://datatracker.ietf.org/doc/html/rfc8259).
+  Extensions such as [comments](../features/comments.md) and [trailing commas](../features/trailing_commas.md) must be
+  enabled explicitly. Invalid UTF-8 is rejected.
+- **Errors are reported, not ignored.** Malformed input results in a [`parse_error`](../home/exceptions.md#parse-errors)
+  with the byte position of the error. Binary readers do not trust announced sizes: strings and binary values grow
+  only as bytes are actually read, arrays reserve at most a fixed number of elements up front, and sizes that no
+  container can hold are rejected.
+- **Memory is owned by values.** Each `basic_json` value owns its content, and there is no manual memory management in
+  user code. The destructor does not recurse, so destroying a deeply nested value does not exhaust the stack.
+- **Bounded recursion.** The JSON parser and the binary readers keep their state in explicit stacks instead of
+  recursing per nesting level. Operations that walk a value, such as [`dump`](../api/basic_json/dump.md), copying,
+  hashing, and [`merge_patch`](../api/basic_json/merge_patch.md), recurse only up to a fixed depth and continue with an
+  explicit stack below it. Some operations, such as comparison, [`diff`](../api/basic_json/diff.md),
+  [`flatten`](../api/basic_json/flatten.md), and the binary writers, still recurse once per nesting level; work on them
+  is in progress. Applications that process untrusted input can limit its nesting depth with a
+  [parser callback](../features/parsing/parser_callbacks.md).
+- **Invariants are checked.** The class invariant (for instance, that the pointer for the stored type is never null) is
+  checked with runtime assertions throughout the test suite.
+
+## Common weaknesses
+
+The following table maps the relevant classes of the [Common Weakness Enumeration](https://cwe.mitre.org) to the
+measures that counter them. The measures are described in detail in [Quality assurance](quality_assurance.md).
+
+| Weakness                                                                  | Countermeasures                                                                                  |
+|---------------------------------------------------------------------------|--------------------------------------------------------------------------------------------------|
+| Out-of-bounds read/write ([CWE-125](https://cwe.mitre.org/data/definitions/125.html), [CWE-787](https://cwe.mitre.org/data/definitions/787.html)) | bounds checks on all reads from the input; AddressSanitizer and Valgrind on the test suite; OSS-Fuzz      |
+| Integer overflow ([CWE-190](https://cwe.mitre.org/data/definitions/190.html)) | UndefinedBehaviorSanitizer with integer overflow detection; Clang-Tidy; Cppcheck                 |
+| Use after free, double free ([CWE-416](https://cwe.mitre.org/data/definitions/416.html), [CWE-415](https://cwe.mitre.org/data/definitions/415.html)) | ownership of all memory by values; AddressSanitizer and Valgrind; Clang Static Analyzer          |
+| Memory leaks ([CWE-401](https://cwe.mitre.org/data/definitions/401.html)) | Valgrind (Memcheck) on the test suite                                                            |
+| Uncontrolled recursion ([CWE-674](https://cwe.mitre.org/data/definitions/674.html)) | iterative parser, binary readers, and destructor; bounded recursion in value operations; tests with deeply nested inputs |
+| Uncontrolled resource consumption ([CWE-400](https://cwe.mitre.org/data/definitions/400.html)) | allocations based on announced sizes are capped; OSS-Fuzz with memory limits                     |
+| Undefined behavior in general ([CWE-758](https://cwe.mitre.org/data/definitions/758.html)) | UndefinedBehaviorSanitizer; runtime assertions; Clang-Tidy, Cppcheck, Clang Static Analyzer, Infer |
+
+In addition, every line of the library is covered by the unit tests, and all parsers are fuzz-tested around the clock
+by [OSS-Fuzz](https://github.com/google/oss-fuzz/tree/master/projects/json).

docs/mkdocs/docs/community/index.md

@@ -5,4 +5,6 @@
 - [Contribution Guidelines](contribution_guidelines.md) - guidelines how to contribute to this project
 - [Governance](governance.md) - the governance model of this project
 - [Quality Assurance](quality_assurance.md) - how the quality of this project is assured
+- [Roadmap](roadmap.md) - what the project will and will not do
 - [Security Policy](security_policy.md) - the security policy of the project
+- [Assurance Case](assurance_case.md) - why the library meets its security requirements

docs/mkdocs/docs/community/quality_assurance.md

@@ -164,6 +164,9 @@ Note: Some modern features (like C++20 ranges or filesystem support) may be disa
 - [x] The parser is tested against extensive correctness suites for JSON compliance.
 - [x] In addition, the library is continuously fuzz-tested at [OSS-Fuzz](https://google.github.io/oss-fuzz/) where the
   library is checked against billions of inputs.
+- [x] Every crash reported by OSS-Fuzz is fixed together with a unit test that reproduces it, and the fix references
+  the OSS-Fuzz issue. The round-trip checks of the fuzzer drivers are also part of the unit tests. See the
+  [fuzz testing documentation](https://github.com/nlohmann/json/blob/develop/tests/fuzzing.md#handling-oss-fuzz-reports).
 
 ## Static analysis
 

docs/mkdocs/docs/community/roadmap.md

@@ -0,0 +1,43 @@
+# Roadmap
+
+This page describes what the project intends to do, and what it does not intend to do, over the next year. Concrete
+work items are tracked in the [GitHub milestones](https://github.com/nlohmann/json/milestones) and the
+[issue tracker](https://github.com/nlohmann/json/issues).
+
+## What the project will do
+
+- **Keep the C++11 baseline.** The library will continue to compile with every
+  [supported C++11 compiler](https://github.com/nlohmann/json/blob/develop/README.md#supported-compilers). Features of
+  later standards are only used when they are guarded by the `JSON_HAS_CPP_*` macros.
+- **Stay conformant to JSON.** The parser and serializer follow [RFC 8259](https://datatracker.ietf.org/doc/html/rfc8259).
+  Extensions such as [comments](../features/comments.md) or [trailing commas](../features/trailing_commas.md) remain
+  opt-in.
+- **Keep the 3.x public API stable.** Releases follow [semantic versioning](https://semver.org). Changes that would
+  break existing code are only added behind a feature macro, so users can opt in and test their code before a next
+  major release.
+- **Support a broad range of compilers and platforms.** The [CI](quality_assurance.md) keeps testing old and new
+  versions of GCC, Clang, MSVC, and other compilers on Linux, macOS, and Windows.
+- **Keep the quality assurance up.** Every change keeps the test coverage at 100%, passes the static and dynamic
+  analysis, and is fuzz-tested by OSS-Fuzz, see [Quality assurance](quality_assurance.md).
+- **Harden the library against hostile input.** Handling deeply nested values without exhausting the call stack is
+  ongoing work.
+- **Fix bugs and security issues** reported through the issue tracker and the [security policy](security_policy.md).
+
+## What the project will not do
+
+- **Break the public API of version 3.x.** See the
+  [contribution guidelines](https://github.com/nlohmann/json/blob/develop/.github/CONTRIBUTING.md#break-the-public-api)
+  for what counts as a breaking change.
+- **Require a newer C++ standard than C++11.**
+- **Break JSON conformance** or enable non-standard extensions by default.
+- **Add dependencies** or require a build step. The library remains header-only, and the single header
+  `json.hpp` remains a complete distribution.
+- **Trade simplicity for speed or memory efficiency.** Performance improvements are welcome, but the library is not
+  meant to compete with the fastest JSON libraries, see [Design goals](../home/design_goals.md).
+
+## Version 4.0
+
+There is no decision yet on whether or when a version 4.0 with breaking changes will be released. Proposals that need
+a major version, for instance stricter type conversions, are collected in issue
+[#3453](https://github.com/nlohmann/json/issues/3453). Until then, such changes are only added as opt-in behavior
+behind feature macros.

docs/mkdocs/docs/features/binary_formats/bjdata.md

@@ -208,6 +208,16 @@ The library maps BJData types to JSON value types as follows:
 
     The mapping is **complete** in the sense that any BJData value can be converted to a JSON value.
 
+!!! info "Round trips"
+
+    A value returned by [`from_bjdata`](../../api/basic_json/from_bjdata.md) can be serialized with
+    [`to_bjdata`](../../api/basic_json/to_bjdata.md) using any combination of options and parsed back into an equal
+    value, and serializing that value again with the same options produces the same bytes. The exception is binary
+    values: they are only written as an optimized binary array (`[$B`) if Draft 3 is enabled and both `use_size` and
+    `use_type` are set. Otherwise, they are written as arrays of integers and parsed back as such (see the notes on
+    binary values above), and serializing such an array again may choose different, but equally valid, type markers.
+    The bytes can then differ, but parsing them again yields the same value.
+
 ??? example
 
     ```cpp

docs/mkdocs/docs/features/macros.md

@@ -93,11 +93,21 @@ See [full documentation of `JSON_NO_IO`](../api/macros/json_no_io.md).
 
 ## `JSON_NO_THREAD_LOCAL`
 
-When defined, the library does not use `#!cpp thread_local` storage. Copying a value then always avoids the call stack
-rather than descending into a bounded number of levels first, which is slower but yields the same values.
+When defined, the library does not use `#!cpp thread_local` storage. Copying a value and comparing two values then
+always avoid the call stack rather than descending into a bounded number of levels first, which is slower but yields the
+same values and the same comparisons.
 
 See [full documentation of `JSON_NO_THREAD_LOCAL`](../api/macros/json_no_thread_local.md).
 
+## `JSON_PRECISE_STREAM_POSITION`
+
+When defined to `1`, [`operator>>`](../api/operator_gtgt.md) and non-strict
+[`sax_parse`](../api/basic_json/sax_parse.md) leave an input stream positioned right after the parsed value, instead of
+also consuming the character that terminates a number. The default value is `0`, which preserves the existing behavior;
+this is planned to become the default in version 4.0.0.
+
+See [full documentation of `JSON_PRECISE_STREAM_POSITION`](../api/macros/json_precise_stream_position.md).
+
 ## `JSON_SKIP_LIBRARY_VERSION_CHECK`
 
 When defined, the library will not create a compiler warning when a different version of the library was already

docs/mkdocs/docs/features/namespace.md

@@ -18,6 +18,7 @@ The complete default namespace name is derived as follows:
     - [`JSON_DIAGNOSTIC_POSITIONS`](../api/macros/json_diagnostic_positions.md) defined non-zero appends `_dp`.
     - [`JSON_BRACE_INIT_COPY_SEMANTICS`](../api/macros/json_brace_init_copy_semantics.md) defined non-zero appends
       `_bics`.
+    - [`JSON_PRECISE_STREAM_POSITION`](../api/macros/json_precise_stream_position.md) defined non-zero appends `_psp`.
 - The inline namespace ends with the suffix `_v` followed by the 3 components of the version number separated by
   underscores. To omit the version component, see [Disabling the version component](#disabling-the-version-component)
   below.

docs/mkdocs/docs/features/parsing/index.md

@@ -41,7 +41,8 @@ document followed by trailing bytes" is accepted rather than rejected. If you ar
 reject any input that is not exactly one JSON document, prefer `parse`.
 
 When using `operator>>` to read several concatenated values this way, a value that is a number must be followed by
-whitespace, because `operator>>` consumes the character that terminates a number — see the
+whitespace, because `operator>>` consumes the character that terminates a number, unless
+[`JSON_PRECISE_STREAM_POSITION`](../../api/macros/json_precise_stream_position.md) is defined to `1` — see the
 [`operator>>` notes](../../api/operator_gtgt.md#notes) for details and examples.
 
 ## SAX vs. DOM parsing

docs/mkdocs/docs/features/types/template_parameters.md

@@ -562,7 +562,11 @@ binary32 or binary64 field and have no encoding for `#!cpp long double`.
 ## `AllocatorType`
 
 `AllocatorType` is instantiated with **one** argument, for each of `object_t`, `array_t`, `string_t`, `binary_t`,
-`basic_json`, and `#!cpp std::pair<const StringType, basic_json>`.
+`basic_json`, `#!cpp std::pair<const StringType, basic_json>`, and `#!cpp std::pair<StringType, basic_json>`.
+
+`AllocatorType` is not the only allocator a `basic_json` uses. It allocates the JSON values themselves, but most
+temporary storage is allocated with `#!cpp std::allocator`. This includes the parser's stacks and the stacks that
+process deeply nested values without recursion.
 
 ### Always required
 

docs/mkdocs/docs/home/architecture.md

@@ -1,34 +1,125 @@
 # Architecture
 
-!!! info
-
-    This page is still under construction. Its goal is to provide a high-level overview of the library's architecture.
-    This should help new contributors to get an idea of the used concepts and where to make changes.
+This page gives a high-level overview of the library's architecture. It should help new contributors to get an idea of
+the used concepts and where to make changes.
 
 ## Overview
 
-The main structure is class [nlohmann::basic_json](../api/basic_json/index.md).
+The library is built around a single class template, [`nlohmann::basic_json`](../api/basic_json/index.md). A
+`basic_json` value is a node in a tree of JSON values. All other components either create such a tree from an input
+(parsing), write a tree to an output (serialization), or give access to it (iterators, JSON Pointer, conversions).
 
-- public API
-- container interface
-- iterators
+```mermaid
+flowchart LR
+    input[/"input<br>(string, stream,<br>iterator range, file)"/]
+    ia["input adapter"]
+    lexer["lexer"]
+    parser["parser"]
+    breader["binary_reader"]
+    sax["SAX interface"]
+    value[("basic_json<br>value tree")]
+    serializer["serializer"]
+    bwriter["binary_writer"]
+    oa["output adapter"]
+    output[/"output<br>(string, stream,<br>vector)"/]
 
-## Template specializations
+    input --> ia
+    ia --> lexer --> parser --> sax
+    ia --> breader --> sax
+    sax --> value
+    value --> serializer --> oa
+    value --> bwriter --> oa
+    oa --> output
+```
 
-- describe template parameters of `basic_json`
-- [`json`](../api/json.md)
-- [`ordered_json`](../api/ordered_json.md) via [`ordered_map`](../api/ordered_map.md)
+- **JSON text** is read by an [input adapter](#input-adapters), tokenized by the lexer, and turned into SAX events by
+  the parser.
+- **Binary formats** (BJData, BSON, CBOR, MessagePack, UBJSON) are read by an input adapter and turned into the same SAX
+  events by the `binary_reader`.
+- A [SAX consumer](#sax-interface) receives the events. The one used by [`parse`](../api/basic_json/parse.md) builds a
+  `basic_json` value tree.
+- The `serializer` (JSON text) or the `binary_writer` (binary formats) writes a value tree to an
+  [output adapter](#output-adapters).
+
+## Source layout
+
+The public headers are in [`include/nlohmann`](https://github.com/nlohmann/json/tree/develop/include/nlohmann):
+
+- [`json.hpp`](https://github.com/nlohmann/json/blob/develop/include/nlohmann/json.hpp) defines class [`basic_json`](../api/basic_json/index.md).
+- [`json_fwd.hpp`](https://github.com/nlohmann/json/blob/develop/include/nlohmann/json_fwd.hpp) contains forward declarations.
+- [`adl_serializer.hpp`](https://github.com/nlohmann/json/blob/develop/include/nlohmann/adl_serializer.hpp), [`byte_container_with_subtype.hpp`](https://github.com/nlohmann/json/blob/develop/include/nlohmann/byte_container_with_subtype.hpp), and [`ordered_map.hpp`](https://github.com/nlohmann/json/blob/develop/include/nlohmann/ordered_map.hpp) define
+  [`adl_serializer`](../api/adl_serializer/index.md),
+  [`byte_container_with_subtype`](../api/byte_container_with_subtype/index.md), and
+  [`ordered_map`](../api/ordered_map.md).
+
+Everything else lives in [`detail/`](https://github.com/nlohmann/json/tree/develop/include/nlohmann/detail) and namespace `nlohmann::detail`, which is not part of the public API. Paths
+below are relative to `include/nlohmann`.
+
+| Component | Location |
+|-----------|----------|
+| Value type enumeration | [`detail/value_t.hpp`](https://github.com/nlohmann/json/blob/develop/include/nlohmann/detail/value_t.hpp) |
+| Input adapters | [`detail/input/input_adapters.hpp`](https://github.com/nlohmann/json/blob/develop/include/nlohmann/detail/input/input_adapters.hpp) |
+| Lexer | [`detail/input/lexer.hpp`](https://github.com/nlohmann/json/blob/develop/include/nlohmann/detail/input/lexer.hpp), [`detail/input/number_parse.hpp`](https://github.com/nlohmann/json/blob/develop/include/nlohmann/detail/input/number_parse.hpp), [`detail/input/string_scan.hpp`](https://github.com/nlohmann/json/blob/develop/include/nlohmann/detail/input/string_scan.hpp) |
+| Parser | [`detail/input/parser.hpp`](https://github.com/nlohmann/json/blob/develop/include/nlohmann/detail/input/parser.hpp) |
+| SAX interface and DOM builders | [`detail/input/json_sax.hpp`](https://github.com/nlohmann/json/blob/develop/include/nlohmann/detail/input/json_sax.hpp) |
+| Binary format readers | [`detail/input/binary_reader.hpp`](https://github.com/nlohmann/json/blob/develop/include/nlohmann/detail/input/binary_reader.hpp) |
+| JSON serializer | [`detail/output/serializer.hpp`](https://github.com/nlohmann/json/blob/develop/include/nlohmann/detail/output/serializer.hpp), [`detail/conversions/to_chars.hpp`](https://github.com/nlohmann/json/blob/develop/include/nlohmann/detail/conversions/to_chars.hpp) |
+| Binary format writers | [`detail/output/binary_writer.hpp`](https://github.com/nlohmann/json/blob/develop/include/nlohmann/detail/output/binary_writer.hpp) |
+| Output adapters | [`detail/output/output_adapters.hpp`](https://github.com/nlohmann/json/blob/develop/include/nlohmann/detail/output/output_adapters.hpp) |
+| Iterators | [`detail/iterators/`](https://github.com/nlohmann/json/tree/develop/include/nlohmann/detail/iterators) |
+| Conversions from/to arbitrary types | [`detail/conversions/from_json.hpp`](https://github.com/nlohmann/json/blob/develop/include/nlohmann/detail/conversions/from_json.hpp), [`detail/conversions/to_json.hpp`](https://github.com/nlohmann/json/blob/develop/include/nlohmann/detail/conversions/to_json.hpp) |
+| JSON Pointer | [`detail/json_pointer.hpp`](https://github.com/nlohmann/json/blob/develop/include/nlohmann/detail/json_pointer.hpp) |
+| Exceptions | [`detail/exceptions.hpp`](https://github.com/nlohmann/json/blob/develop/include/nlohmann/detail/exceptions.hpp) |
+| Type traits and C++ feature backports | [`detail/meta/`](https://github.com/nlohmann/json/tree/develop/include/nlohmann/detail/meta) |
+| Macros | [`detail/macro_scope.hpp`](https://github.com/nlohmann/json/blob/develop/include/nlohmann/detail/macro_scope.hpp), [`detail/macro_unscope.hpp`](https://github.com/nlohmann/json/blob/develop/include/nlohmann/detail/macro_unscope.hpp), [`detail/abi_macros.hpp`](https://github.com/nlohmann/json/blob/develop/include/nlohmann/detail/abi_macros.hpp) |
+
+The single-header version [`single_include/nlohmann/json.hpp`](https://github.com/nlohmann/json/blob/develop/single_include/nlohmann/json.hpp)
+is generated from these files with `make amalgamate` and must not be edited by hand.
+
+## Template parameters
+
+[`basic_json`](../api/basic_json/index.md) is parameterized by the types it uses to store values and to convert from and to other types:
+
+| Template parameter   | Default                     | Used for                                                          |
+|----------------------|-----------------------------|-------------------------------------------------------------------|
+| `ObjectType`         | `std::map`                  | objects, see [`object_t`](../api/basic_json/object_t.md)          |
+| `ArrayType`          | `std::vector`               | arrays, see [`array_t`](../api/basic_json/array_t.md)             |
+| `StringType`         | `std::string`               | strings and object keys, see [`string_t`](../api/basic_json/string_t.md) |
+| `BooleanType`        | `bool`                      | Booleans, see [`boolean_t`](../api/basic_json/boolean_t.md)       |
+| `NumberIntegerType`  | `std::int64_t`              | signed integers, see [`number_integer_t`](../api/basic_json/number_integer_t.md) |
+| `NumberUnsignedType` | `std::uint64_t`             | unsigned integers, see [`number_unsigned_t`](../api/basic_json/number_unsigned_t.md) |
+| `NumberFloatType`    | `double`                    | floating-point numbers, see [`number_float_t`](../api/basic_json/number_float_t.md) |
+| `AllocatorType`      | `std::allocator`            | allocating objects, arrays, strings, and binary values            |
+| `JSONSerializer`     | `adl_serializer`            | conversions from/to other types, see [`adl_serializer`](../api/adl_serializer/index.md) |
+| `BinaryType`         | `std::vector<std::uint8_t>` | binary values, see [`binary_t`](../api/basic_json/binary_t.md)  |
+| `CustomBaseClass`    | `void`                      | an optional base class, see [`json_base_class_t`](../api/basic_json/json_base_class_t.md) |
+
+The library provides two specializations:
+
+- [`json`](../api/json.md) uses all default template arguments.
+- [`ordered_json`](../api/ordered_json.md) uses [`ordered_map`](../api/ordered_map.md) as `ObjectType` to keep the
+  insertion order of object keys.
+
+The requirements on the template arguments are listed in
+[Template Parameter Requirements](../features/types/template_parameters.md).
 
 ## Value storage
 
-Values are stored as a tagged union of [value_t](../api/basic_json/value_t.md) and json_value.
+Each [`basic_json`](../api/basic_json/index.md) value stores its content as a tagged union: an enumeration [`value_t`](../api/basic_json/value_t.md)
+names the type of the value, and a union `json_value` holds the value itself. Both are members of the nested struct
+`data`, which is the only data member `m_data` of `basic_json`:
 
 ```cpp
-/// the type of the current element
-value_t m_type = value_t::null;
+struct data
+{
+    /// the type of the current element
+    value_t m_type = value_t::null;
 
-/// the value of the current element
-json_value m_value = {};
+    /// the value of the current element
+    json_value m_value = {};
+};
+
+data m_data = {};
 ```
 
 with
@@ -68,42 +159,83 @@ union json_value {
 };
 ```
 
-## Parsing inputs (deserialization)
+Objects, arrays, strings, and binary values are allocated on the heap with `AllocatorType`, and the union only stores a
+pointer to them. This keeps a `basic_json` value small: one pointer-sized union and one byte for the type. The class
+maintains the invariant that the pointer matching `m_type` is never null; `assert_invariant()` checks it with
+[runtime assertions](../features/assertions.md).
 
-Input is read via **input adapters** that abstract a source with a common interface:
+## Input adapters
+
+Input is read via **input adapters** that abstract a source. Every input adapter provides this interface:
 
 ```cpp
-/// read a single character
-std::char_traits<char>::int_type get_character() noexcept;
+/// the type of the characters in the input
+using char_type = ...;
 
-/// read multiple characters to a destination buffer and
-/// returns the number of characters successfully read
+/// read a single character; returns std::char_traits<char_type>::eof() at the end of the input
+typename std::char_traits<char_type>::int_type get_character();
+
+/// read up to count * sizeof(T) bytes into dest and return the number of bytes read
+/// (used by the binary readers)
 template<class T>
 std::size_t get_elements(T* dest, std::size_t count = 1);
 ```
 
-List examples of input adapters.
+The lexer detects two optional extensions at compile time. Only `iterator_input_adapter` provides them, and only for
+random-access input of single-byte characters:
 
-## SAX Interface
+- `supports_seek`, `get_consumed_count()`, and `copy_consumed_range()` let the lexer reconstruct already consumed input
+  for error messages instead of copying every character it reads.
+- `supports_bulk_scan`, `bulk_data()`, `bulk_remaining()`, and `bulk_skip()` let the lexer scan strings directly in
+  contiguous memory, several bytes at a time.
 
-TODO
+The function `input_adapter` picks the right adapter for the argument passed to `parse`, `accept`, `sax_parse`, or the
+`from_*` functions:
 
-## Writing outputs (serialization)
+- `iterator_input_adapter` reads from an iterator range, which also covers strings, containers, and pointers.
+- `wide_string_input_adapter` reads from ranges of `wchar_t`, `char16_t`, or `char32_t` and converts them to UTF-8.
+  It cannot be used for binary formats; its `get_elements()` throws.
+- `input_stream_adapter` reads from a `std::istream`.
+- `file_input_adapter` reads from a `std::FILE*`.
+
+## SAX interface
+
+The parser does not build values itself. It reports what it reads as events to a [SAX](../features/parsing/sax_interface.md)
+consumer, which implements the interface [`json_sax`](../api/json_sax/index.md): `null`, `boolean`, `number_integer`,
+`number_unsigned`, `number_float`, `string`, `binary`, `start_object`, `key`, `end_object`, `start_array`, `end_array`,
+and `parse_error`.
+
+The library comes with two consumers in `detail/input/json_sax.hpp`:
+
+- `json_sax_dom_parser` builds a [`basic_json`](../api/basic_json/index.md) value tree. [`parse`](../api/basic_json/parse.md) uses it.
+- `json_sax_dom_callback_parser` does the same, but calls a [parser callback](../features/parsing/parser_callbacks.md)
+  for each event, which can skip values. `parse` uses it when a callback is given.
+
+The `binary_reader` emits the same events for binary formats, so [`sax_parse`](../api/basic_json/sax_parse.md) works
+with a user-defined consumer for JSON and for all binary formats alike.
+
+## Output adapters
 
 Output is written via **output adapters**:
 
 ```cpp
-template<typename T>
 void write_character(CharType c);
 
-template<typename CharType>
 void write_characters(const CharType* s, std::size_t length);
 ```
 
-List examples of output adapters.
+The `serializer` (used by [`dump`](../api/basic_json/dump.md) and [`operator<<`](../api/operator_ltlt.md)) and the
+`binary_writer` (used by the `to_*` functions) write to one of these adapters:
+
+- `output_vector_adapter` appends to a `std::vector`.
+- `output_stream_adapter` writes to a `std::ostream`.
+- `output_string_adapter` appends to a string.
 
 ## Value conversion
 
+Values are converted from and to other types with the `JSONSerializer` template parameter. The default,
+[`adl_serializer`](../api/adl_serializer/index.md), calls the free functions
+
 ```cpp
 template<class T>
 void to_json(basic_json& j, const T& t);
@@ -112,13 +244,23 @@ template<class T>
 void from_json(const basic_json& j, T& t);
 ```
 
+found by argument-dependent lookup. The library defines them for standard types in `detail/conversions`; users add them
+for their own types, see [Arbitrary Type Conversions](../features/arbitrary_types.md). The
+[serialization macros](../features/macros.md) generate these functions.
+
 ## Additional features
 
-- JSON Pointers
-- Binary formats
-- Custom base class
-- Conversion macros
+- [JSON Pointer](../features/json_pointer.md) (class `json_pointer`) addresses values inside a tree. It is also the
+  basis of [JSON Patch](../features/json_patch.md).
+- [Binary formats](../features/binary_formats/index.md) are read by `binary_reader` and written by `binary_writer`.
+- A [custom base class](../api/basic_json/json_base_class_t.md) can add members to every [`basic_json`](../api/basic_json/index.md) value.
+- [Serialization macros](../features/macros.md) generate `to_json` and `from_json` functions for user-defined types.
 
 ## Details namespace
 
-- C++ feature backports
+Namespace `nlohmann::detail` contains all implementation details. It is not part of the public API and may change in any
+release. Besides the components above, it contains:
+
+- type traits to detect the capabilities of user-defined types (`detail/meta/type_traits.hpp`),
+- backports of C++14/17 features to C++11 (`detail/meta/cpp_future.hpp`), and
+- helpers such as `string_concat` and `string_escape`.

docs/mkdocs/mkdocs.yml

@@ -293,6 +293,7 @@ nav:
           - 'JSON_NOEXCEPTION': api/macros/json_noexception.md
           - 'JSON_NO_IO': api/macros/json_no_io.md
           - 'JSON_NO_THREAD_LOCAL': api/macros/json_no_thread_local.md
+          - 'JSON_PRECISE_STREAM_POSITION': api/macros/json_precise_stream_position.md
           - 'JSON_SKIP_LIBRARY_VERSION_CHECK': api/macros/json_skip_library_version_check.md
           - 'JSON_SKIP_UNSUPPORTED_COMPILER_CHECK': api/macros/json_skip_unsupported_compiler_check.md
           - 'JSON_STRICT_NUL_HANDLING': api/macros/json_strict_nul_handling.md
@@ -317,7 +318,9 @@ nav:
       - community/contribution_guidelines.md
       - community/quality_assurance.md
       - community/governance.md
+      - community/roadmap.md
       - community/security_policy.md
+      - community/assurance_case.md
 
 # Extras
 extra:

include/nlohmann/detail/abi_macros.hpp

@@ -38,6 +38,10 @@
     #define JSON_BRACE_INIT_COPY_SEMANTICS 0
 #endif
 
+#ifndef JSON_PRECISE_STREAM_POSITION
+    #define JSON_PRECISE_STREAM_POSITION 0
+#endif
+
 #if JSON_DIAGNOSTICS
     #define NLOHMANN_JSON_ABI_TAG_DIAGNOSTICS _diag
 #else
@@ -62,21 +66,28 @@
     #define NLOHMANN_JSON_ABI_TAG_BRACE_INIT_COPY_SEMANTICS
 #endif
 
+#if JSON_PRECISE_STREAM_POSITION
+    #define NLOHMANN_JSON_ABI_TAG_PRECISE_STREAM_POSITION _psp
+#else
+    #define NLOHMANN_JSON_ABI_TAG_PRECISE_STREAM_POSITION
+#endif
+
 #ifndef NLOHMANN_JSON_NAMESPACE_NO_VERSION
     #define NLOHMANN_JSON_NAMESPACE_NO_VERSION 0
 #endif
 
 // Construct the namespace ABI tags component
-#define NLOHMANN_JSON_ABI_TAGS_CONCAT_EX(a, b, c, d) json_abi ## a ## b ## c ## d
-#define NLOHMANN_JSON_ABI_TAGS_CONCAT(a, b, c, d) \
-    NLOHMANN_JSON_ABI_TAGS_CONCAT_EX(a, b, c, d)
+#define NLOHMANN_JSON_ABI_TAGS_CONCAT_EX(a, b, c, d, e) json_abi ## a ## b ## c ## d ## e
+#define NLOHMANN_JSON_ABI_TAGS_CONCAT(a, b, c, d, e) \
+    NLOHMANN_JSON_ABI_TAGS_CONCAT_EX(a, b, c, d, e)
 
 #define NLOHMANN_JSON_ABI_TAGS                                       \
     NLOHMANN_JSON_ABI_TAGS_CONCAT(                                   \
             NLOHMANN_JSON_ABI_TAG_DIAGNOSTICS,                       \
             NLOHMANN_JSON_ABI_TAG_LEGACY_DISCARDED_VALUE_COMPARISON, \
             NLOHMANN_JSON_ABI_TAG_DIAGNOSTIC_POSITIONS,              \
-            NLOHMANN_JSON_ABI_TAG_BRACE_INIT_COPY_SEMANTICS)
+            NLOHMANN_JSON_ABI_TAG_BRACE_INIT_COPY_SEMANTICS,         \
+            NLOHMANN_JSON_ABI_TAG_PRECISE_STREAM_POSITION)
 
 // Construct the namespace version component
 #define NLOHMANN_JSON_NAMESPACE_VERSION_CONCAT_EX(major, minor, patch) \

include/nlohmann/detail/input/input_adapters.hpp

@@ -101,6 +101,11 @@ class input_stream_adapter
         // maintain ifstream flags, except eof
         if (is != nullptr)
         {
+#if JSON_PRECISE_STREAM_POSITION
+            // consume the character last returned by get_character() unless it
+            // was given back with release_lookahead()
+            commit_lookahead();
+#endif
             is->clear(is->rdstate() & std::ios::eofbit);
         }
     }
@@ -114,6 +119,58 @@ class input_stream_adapter
     input_stream_adapter& operator=(input_stream_adapter&) = delete;
     input_stream_adapter& operator=(input_stream_adapter&&) = delete;
 
+#if JSON_PRECISE_STREAM_POSITION
+    input_stream_adapter(input_stream_adapter&& rhs) noexcept
+        : is(rhs.is), sb(rhs.sb), lookahead(rhs.lookahead)
+    {
+        rhs.is = nullptr;
+        rhs.sb = nullptr;
+        rhs.lookahead = false;
+    }
+
+    // Whether the character last returned by get_character() can be given back
+    // to the input with release_lookahead().
+    static constexpr bool supports_lookahead = true;
+
+    // std::istream/std::streambuf use std::char_traits<char>::to_int_type, to
+    // ensure that std::char_traits<char>::eof() and the character 0xFF do not
+    // end up as the same value, e.g., 0xFFFFFFFF.
+    //
+    // The character is peeked rather than consumed: it is only stepped over
+    // once the next character is requested, or when the adapter is destroyed.
+    // Until then, release_lookahead() can leave it in the input.
+    std::char_traits<char>::int_type get_character()
+    {
+        if (lookahead)
+        {
+            // step over the character returned by the previous call
+            sb->sbumpc();
+        }
+
+        auto res = sb->sgetc();
+        // set eof manually, as we don't use the istream interface.
+        if (JSON_HEDLEY_UNLIKELY(res == std::char_traits<char>::eof()))
+        {
+            // there is nothing to step over next time
+            lookahead = false;
+            is->clear(is->rdstate() | std::ios::eofbit);
+        }
+        else
+        {
+            lookahead = true;
+        }
+        return res;
+    }
+
+    // Leave the character last returned by get_character() in the input, so
+    // that the next read from the stream - by this adapter or by the caller
+    // once parsing is done - sees it again. Unlike putting a consumed
+    // character back, this cannot fail.
+    void release_lookahead() noexcept
+    {
+        lookahead = false;
+    }
+#else
     input_stream_adapter(input_stream_adapter&& rhs) noexcept
         : is(rhs.is), sb(rhs.sb)
     {
@@ -124,6 +181,9 @@ class input_stream_adapter
     // std::istream/std::streambuf use std::char_traits<char>::to_int_type, to
     // ensure that std::char_traits<char>::eof() and the character 0xFF do not
     // end up as the same value, e.g., 0xFFFFFFFF.
+    //
+    // The character is consumed, so the character that terminates a number
+    // stays consumed after parsing; see JSON_PRECISE_STREAM_POSITION.
     std::char_traits<char>::int_type get_character()
     {
         auto res = sb->sbumpc();
@@ -134,10 +194,14 @@ class input_stream_adapter
         }
         return res;
     }
+#endif
 
     template<class T>
     std::size_t get_elements(T* dest, std::size_t count = 1)
     {
+#if JSON_PRECISE_STREAM_POSITION
+        commit_lookahead();
+#endif
         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)))
         {
@@ -147,9 +211,27 @@ class input_stream_adapter
     }
 
   private:
+#if JSON_PRECISE_STREAM_POSITION
+    // Step over the character last returned by get_character(). The character
+    // has already been peeked successfully, so for every streambuf with a get
+    // area this is a pointer increment that cannot fail.
+    void commit_lookahead()
+    {
+        if (lookahead)
+        {
+            lookahead = false;
+            sb->sbumpc();
+        }
+    }
+#endif
+
     /// the associated input stream
     std::istream* is = nullptr;
     std::streambuf* sb = nullptr;
+#if JSON_PRECISE_STREAM_POSITION
+    /// whether get_character() peeked a character that is not consumed yet
+    bool lookahead = false;
+#endif
 };
 #endif  // JSON_NO_IO
 

include/nlohmann/detail/input/lexer.hpp

@@ -127,6 +127,25 @@ constexpr bool input_adapter_supports_seek(std::false_type /*detected*/)
     return false;
 }
 
+// Detect whether an input adapter reads with one character of lookahead that
+// can be left in the input (see input_stream_adapter::supports_lookahead,
+// which is only defined with JSON_PRECISE_STREAM_POSITION), detected like
+// supports_seek above.
+template<typename InputAdapterType>
+using detect_supports_lookahead = decltype(InputAdapterType::supports_lookahead);
+
+template<typename InputAdapterType>
+constexpr bool input_adapter_supports_lookahead(std::true_type /*detected*/)
+{
+    return InputAdapterType::supports_lookahead;
+}
+
+template<typename InputAdapterType>
+constexpr bool input_adapter_supports_lookahead(std::false_type /*detected*/)
+{
+    return false;
+}
+
 // Detect whether an input adapter exposes a contiguous byte block that the
 // lexer can scan directly (see iterator_input_adapter::supports_bulk_scan).
 // Adapters without the flag - file, stream, wide-string, user-defined - fall
@@ -167,6 +186,12 @@ class lexer : public lexer_base<BasicJsonType>
     static constexpr bool lazy_token_string =
         input_adapter_supports_seek<InputAdapterType>(is_detected<detect_supports_seek, InputAdapterType> {});
 
+    /// whether a simulated unget can be passed on to the input adapter, which
+    /// then leaves the character in the input; see
+    /// input_adapter_supports_lookahead
+    static constexpr bool can_release_lookahead =
+        input_adapter_supports_lookahead<InputAdapterType>(is_detected<detect_supports_lookahead, InputAdapterType> {});
+
     /// whether string scanning may bulk-consume runs of ordinary characters
     /// directly from a contiguous input buffer (SWAR fast path). This requires
     /// the token to be reconstructible lazily (lazy_token_string), so bypassing
@@ -1898,6 +1923,21 @@ scan_number_done:
         uncapture_char(std::integral_constant<bool, lazy_token_string> {});
     }
 
+    /// adapter without lookahead: nothing to do (see release_lookahead)
+    void release_lookahead_impl(std::false_type /*can_release*/) const noexcept {}
+
+    /// adapter with lookahead: leave the character in the input instead
+    void release_lookahead_impl(std::true_type /*can_release*/)
+    {
+        if (next_unget)
+        {
+            // the character is read from the input again rather than replayed
+            // from current, so the adapter must not step over it
+            next_unget = false;
+            ia.release_lookahead();
+        }
+    }
+
     /// seekable adapter: nothing was captured, so nothing to undo
     void uncapture_char(std::true_type /*lazy*/) const noexcept {}
 
@@ -1961,6 +2001,31 @@ scan_number_done:
         return position;
     }
 
+    /*!
+    @brief pass a pending simulated unget on to the input
+
+    unget() only rewinds the lexer's own bookkeeping, so the character that
+    terminated the last token (e.g. the character after a number) would still
+    be stepped over when the input adapter is done. Callers that hand the
+    input back to the user afterwards - operator>> and non-strict sax_parse -
+    call this once when scanning is done, so that the input is positioned
+    right after the value.
+
+    Adapters without lookahead (see input_adapter_supports_lookahead) are not
+    handed back to the user, so this is a no-op for them. Without
+    JSON_PRECISE_STREAM_POSITION, no adapter has lookahead, so this is always a
+    no-op and the terminating character stays consumed.
+
+    Scanning may continue after this call: @a next_unget is cleared, and the
+    character is read from the input again instead of being replayed from
+    @a current. A pending unget of EOF needs no special case, because reaching
+    EOF leaves no lookahead to release.
+    */
+    void release_lookahead()
+    {
+        release_lookahead_impl(std::integral_constant<bool, can_release_lookahead> {});
+    }
+
 #if JSON_DIAGNOSTIC_POSITIONS
     /// return the offset of the first character of the last read token; unlike
     /// the token's parsed value, this accounts for escape sequences

include/nlohmann/detail/input/parser.hpp

@@ -100,13 +100,22 @@ class parser
             json_sax_dom_callback_parser<BasicJsonType, InputAdapterType> sdp(result, callback, allow_exceptions, &m_lexer);
             sax_parse_internal(&sdp);
 
-            // in strict mode, input must be completely read
-            if (strict && (get_token() != token_type::end_of_input))
+            if (strict)
             {
-                sdp.parse_error(m_lexer.get_position(),
-                                m_lexer.get_token_string(),
-                                parse_error::create(101, m_lexer.get_position(),
-                                                    exception_message(token_type::end_of_input, "value"), nullptr));
+                // in strict mode, input must be completely read
+                if (get_token() != token_type::end_of_input)
+                {
+                    sdp.parse_error(m_lexer.get_position(),
+                                    m_lexer.get_token_string(),
+                                    parse_error::create(101, m_lexer.get_position(),
+                                                        exception_message(token_type::end_of_input, "value"), nullptr));
+                }
+            }
+            else
+            {
+                // the caller keeps using the input: position it right after
+                // the value by leaving the character that terminated it
+                m_lexer.release_lookahead();
             }
 
             // in case of an error, return a discarded value
@@ -128,12 +137,20 @@ class parser
             json_sax_dom_parser<BasicJsonType, InputAdapterType> sdp(result, allow_exceptions, &m_lexer);
             sax_parse_internal(&sdp);
 
-            // in strict mode, input must be completely read
-            if (strict && (get_token() != token_type::end_of_input))
+            if (strict)
             {
-                sdp.parse_error(m_lexer.get_position(),
-                                m_lexer.get_token_string(),
-                                parse_error::create(101, m_lexer.get_position(), exception_message(token_type::end_of_input, "value"), nullptr));
+                // in strict mode, input must be completely read
+                if (get_token() != token_type::end_of_input)
+                {
+                    sdp.parse_error(m_lexer.get_position(),
+                                    m_lexer.get_token_string(),
+                                    parse_error::create(101, m_lexer.get_position(), exception_message(token_type::end_of_input, "value"), nullptr));
+                }
+            }
+            else
+            {
+                // see above
+                m_lexer.release_lookahead();
             }
 
             // in case of an error, return a discarded value
@@ -166,12 +183,24 @@ class parser
         (void)detail::is_sax_static_asserts<SAX, BasicJsonType> {};
         const bool result = sax_parse_internal(sax);
 
-        // strict mode: next byte must be EOF
-        if (result && strict && (get_token() != token_type::end_of_input))
+        if (result)
         {
-            return sax->parse_error(m_lexer.get_position(),
-                                    m_lexer.get_token_string(),
-                                    parse_error::create(101, m_lexer.get_position(), exception_message(token_type::end_of_input, "value"), nullptr));
+            if (strict)
+            {
+                // strict mode: next byte must be EOF
+                if (get_token() != token_type::end_of_input)
+                {
+                    return sax->parse_error(m_lexer.get_position(),
+                                            m_lexer.get_token_string(),
+                                            parse_error::create(101, m_lexer.get_position(), exception_message(token_type::end_of_input, "value"), nullptr));
+                }
+            }
+            else
+            {
+                // the caller keeps using the input: position it right after
+                // the value by leaving the character that terminated it
+                m_lexer.release_lookahead();
+            }
         }
 
         return result;

include/nlohmann/detail/macro_unscope.hpp

@@ -45,6 +45,7 @@
     #undef JSON_HAS_STATIC_RTTI
     #undef JSON_USE_LEGACY_DISCARDED_VALUE_COMPARISON
     #undef JSON_BRACE_INIT_COPY_SEMANTICS
+    #undef JSON_PRECISE_STREAM_POSITION
 #endif
 
 #include <nlohmann/thirdparty/hedley/hedley_undef.hpp>

include/nlohmann/json.hpp

@@ -924,6 +924,31 @@ class basic_json // NOLINT(cppcoreguidelines-special-member-functions,hicpp-spec
     }
 #endif
 
+    /*!
+    @brief whether a descent must stop here and finish without the call stack
+
+    @a may_descend says whether the operator descends at all; it is a constant
+    at every call site, and is passed rather than tested by the caller so that
+    the test does not become a constant condition there, which MSVC reports as
+    C4127.
+
+    The comparison operators use this rather than @ref nesting_depth_guard::okay,
+    because they are written as a macro and a macro cannot use the preprocessor
+    the way the guard's constructor does; @ref copy_structured, which can, asks
+    the guard instead and never calls this.
+    */
+    static bool nesting_depth_exhausted(bool may_descend = true) noexcept
+    {
+#ifdef JSON_NO_THREAD_LOCAL
+        // without a count of its own per thread, a descent cannot be bounded
+        // without racing another one, so none is made
+        static_cast<void>(may_descend);
+        return true;
+#else
+        return !may_descend || nesting_depth() >= nesting_depth_limit();
+#endif
+    }
+
     /*!
     @brief counts one level of a bounded descent for as long as it runs, and
            reports whether the descent was still within the limit when it began
@@ -978,7 +1003,8 @@ class basic_json // NOLINT(cppcoreguidelines-special-member-functions,hicpp-spec
     using copy_worklist_t = std::vector<std::pair<const basic_json*, basic_json*>>;
 
     /// scratch space to build the key skeleton of an object copy in one go
-    using copy_scratch_t = std::vector<std::pair<typename object_t::key_type, basic_json>>;
+    using copy_scratch_value_t = std::pair<typename object_t::key_type, basic_json>;
+    using copy_scratch_t = std::vector<copy_scratch_value_t, AllocatorType<copy_scratch_value_t>>;
 
     /// @brief copy everything of @a src into @a dst but its type and value
     static void copy_metadata(const basic_json& src, basic_json& dst)
@@ -1243,6 +1269,274 @@ class basic_json // NOLINT(cppcoreguidelines-special-member-functions,hicpp-spec
     }
 
 
+    /// the result of comparing two values, including values that cannot be
+    /// ordered at all, such as a discarded value or a NaN
+    enum class compare_result { less, equal, greater, unordered };
+
+#if JSON_HAS_THREE_WAY_COMPARISON
+    /// @brief the ordering that @a result stands for
+    static std::partial_ordering to_partial_ordering(compare_result result) noexcept // *NOPAD*
+    {
+        switch (result)
+        {
+            case compare_result::less:
+                return std::partial_ordering::less;
+            case compare_result::greater:
+                return std::partial_ordering::greater;
+            case compare_result::equal:
+                return std::partial_ordering::equivalent;
+            case compare_result::unordered:
+            default:
+                return std::partial_ordering::unordered;
+        }
+    }
+#endif
+
+    /*!
+    @brief compare two values that are not both an array or both an object
+
+    Such a pair is compared by the operators themselves, which cannot descend
+    into it and therefore cannot recurse.
+
+    That holds for a pair whose types differ as much as for a pair of leaves: an
+    array and an object are told apart by their types alone, because an operator
+    only ever descends into two values of the same type. So `==` reports them as
+    unequal without looking inside either, and an ordering falls back to the
+    order of the types - an object sorts before an array - exactly as it does
+    for a value that is not nested deeply enough to get here.
+    */
+    template<bool Ordered>
+    static compare_result compare_leaves(const_reference lhs, const_reference rhs) noexcept
+    {
+        if (lhs == rhs)
+        {
+            return compare_result::equal;
+        }
+
+        return order_leaves(lhs, rhs, std::integral_constant<bool, Ordered> {});
+    }
+
+    /*!
+    @brief compare two object keys
+
+    An object compares its entries as pairs of a key and a value, so its keys
+    are compared exactly as std::pair compares them: with < where the objects
+    are being ordered, and with == where they are only checked for equality.
+    Note that this is not the object's own comparator, which for a vector-backed
+    object type such as nlohmann::ordered_map tells equality rather than order.
+    */
+    static compare_result compare_keys(const typename object_t::key_type& lhs,
+                                       const typename object_t::key_type& rhs,
+                                       std::true_type /*ordered*/)
+    {
+        if (lhs < rhs)
+        {
+            return compare_result::less;
+        }
+
+        if (rhs < lhs)
+        {
+            return compare_result::greater;
+        }
+
+        return compare_result::equal;
+    }
+
+    /// @brief check two object keys for equality
+    static compare_result compare_keys(const typename object_t::key_type& lhs,
+                                       const typename object_t::key_type& rhs,
+                                       std::false_type /*ordered*/)
+    {
+        return lhs == rhs ? compare_result::equal : compare_result::unordered;
+    }
+
+    /// @brief tell apart two values that are not equal
+    /// @note only instantiated where the values are being ordered, as a key or
+    ///       string type is not required to be ordered to be compared for equality
+    static compare_result order_leaves(const_reference lhs, const_reference rhs, std::true_type /*ordered*/) noexcept
+    {
+        if (lhs < rhs)
+        {
+            return compare_result::less;
+        }
+
+        if (rhs < lhs)
+        {
+            return compare_result::greater;
+        }
+
+        return compare_result::unordered;
+    }
+
+    /// @brief report two values as not equal without ordering them
+    static compare_result order_leaves(const_reference /*lhs*/, const_reference /*rhs*/, std::false_type /*ordered*/) noexcept
+    {
+        return compare_result::unordered;
+    }
+
+    /*!
+    @brief compare @a lhs and @a rhs without descending into them
+
+    Reached once a comparison has descended @ref nesting_depth_limit levels, so
+    that comparing values cannot exhaust the call stack however deeply they are
+    nested. The two values are walked in lockstep on an explicit stack and
+    compared lexicographically, element by element in the order the containers
+    enumerate them - which is how the container types this library ships compare
+    themselves: a std::map enumerates its entries in key order, and
+    nlohmann::ordered_map in insertion order. An object type that enumerates its
+    entries in an unspecified order, such as std::unordered_map, compares them
+    pairwise instead; the difference could only ever show below the bound.
+
+    Note that the stack this walks with is allocated, while the comparison
+    operators are noexcept and the container comparison this replaces allocated
+    nothing. Failing that allocation therefore ends the process rather than
+    throwing. It only arises for values nested past the bound, and only when
+    memory has run out - where the same comparison used to exhaust the call
+    stack instead - but it is a way to fail that the operators did not have.
+    */
+    template<bool Ordered>
+    static compare_result compare_iteratively(const_reference lhs, const_reference rhs,
+            const bool unordered_compares_equal) noexcept
+    {
+        /// a pair of containers being compared in lockstep
+        struct frame
+        {
+            const basic_json* lhs_value{nullptr};
+            const basic_json* rhs_value{nullptr};
+            typename array_t::const_iterator lhs_array_it{};
+            typename array_t::const_iterator rhs_array_it{};
+            typename object_t::const_iterator lhs_object_it{};
+            typename object_t::const_iterator rhs_object_it{};
+        };
+
+        std::vector<frame> stack;
+        const basic_json* left = &lhs;
+        const basic_json* right = &rhs;
+
+        for (;;)
+        {
+            const auto type = left->m_data.m_type;
+
+            if (type == right->m_data.m_type && (type == value_t::array || type == value_t::object))
+            {
+                // descend: the elements decide, and are compared further down
+                stack.emplace_back();
+                frame& pushed = stack.back();
+                pushed.lhs_value = left;
+                pushed.rhs_value = right;
+
+                if (type == value_t::array)
+                {
+                    pushed.lhs_array_it = left->m_data.m_value.array->cbegin();
+                    pushed.rhs_array_it = right->m_data.m_value.array->cbegin();
+                }
+                else
+                {
+                    pushed.lhs_object_it = left->m_data.m_value.object->cbegin();
+                    pushed.rhs_object_it = right->m_data.m_value.object->cbegin();
+                }
+            }
+            else
+            {
+                const compare_result result = compare_leaves<Ordered>(*left, *right);
+
+                // Values that cannot be ordered - a NaN, say - end an ordered
+                // comparison for std::lexicographical_compare_three_way, but
+                // std::lexicographical_compare treats them as equivalent and
+                // carries on with the next element. Both are reproduced here,
+                // so that a value nested too deeply to descend into compares
+                // exactly as one that is not.
+                if (result != compare_result::equal &&
+                        !(unordered_compares_equal && result == compare_result::unordered))
+                {
+                    return result;
+                }
+            }
+
+            // walk back up past the containers that are exhausted, then take the
+            // next pair of elements from the innermost one that is not
+            for (;;)
+            {
+                if (stack.empty())
+                {
+                    return compare_result::equal;
+                }
+
+                frame& current = stack.back();
+                const bool is_object = current.lhs_value->m_data.m_type == value_t::object;
+
+                const bool lhs_done = is_object
+                                      ? current.lhs_object_it == current.lhs_value->m_data.m_value.object->cend()
+                                      : current.lhs_array_it == current.lhs_value->m_data.m_value.array->cend();
+                const bool rhs_done = is_object
+                                      ? current.rhs_object_it == current.rhs_value->m_data.m_value.object->cend()
+                                      : current.rhs_array_it == current.rhs_value->m_data.m_value.array->cend();
+
+                if (lhs_done || rhs_done)
+                {
+                    // whichever ran out first holds the smaller container; if
+                    // both did, they are equal and the container above decides
+                    if (lhs_done != rhs_done)
+                    {
+                        return lhs_done ? compare_result::less : compare_result::greater;
+                    }
+
+                    stack.pop_back();
+                    continue;
+                }
+
+                if (is_object)
+                {
+                    // an entry is a key and a value, and the key decides first
+                    const compare_result key_result =
+                        compare_keys(current.lhs_object_it->first, current.rhs_object_it->first,
+                                     std::integral_constant<bool, Ordered> {});
+
+                    if (key_result != compare_result::equal)
+                    {
+                        return key_result;
+                    }
+
+                    left = &(current.lhs_object_it->second);
+                    right = &(current.rhs_object_it->second);
+                    ++current.lhs_object_it;
+                    ++current.rhs_object_it;
+                }
+                else
+                {
+                    left = &(*current.lhs_array_it);
+                    right = &(*current.rhs_array_it);
+                    ++current.lhs_array_it;
+                    ++current.rhs_array_it;
+                }
+
+                break;
+            }
+        }
+    }
+
+
+    /// @brief restore the parent pointers after erasing from an object
+    /// ordered_json keeps its members in a vector, and erasing a member
+    /// re-constructs every member after it in place, which resets their
+    /// parent pointers
+    void set_parents_after_object_erase()
+    {
+#if JSON_DIAGNOSTICS
+#ifdef JSON_HEDLEY_MSVC_VERSION
+#pragma warning(push )
+#pragma warning(disable : 4127) // ignore warning to replace if with if constexpr
+#endif
+        if (detail::is_ordered_map<object_t>::value)
+        {
+            set_parents();
+        }
+#ifdef JSON_HEDLEY_MSVC_VERSION
+#pragma warning( pop )
+#endif
+#endif
+    }
+
   public:
     //////////////////////////
     // JSON parser callback //
@@ -2260,6 +2554,7 @@ class basic_json // NOLINT(cppcoreguidelines-special-member-functions,hicpp-spec
     ValueType & get_to(ValueType& v) const noexcept(noexcept(
                 JSONSerializer<ValueType>::from_json(std::declval<const basic_json_t&>(), v)))
     {
+        static_assert(!std::is_const<ValueType>::value, "get_to() cannot deserialize into a const value");
         JSONSerializer<ValueType>::from_json(*this, v);
         return v;
     }
@@ -2285,6 +2580,7 @@ class basic_json // NOLINT(cppcoreguidelines-special-member-functions,hicpp-spec
     noexcept(noexcept(JSONSerializer<Array>::from_json(
                           std::declval<const basic_json_t&>(), v)))
     {
+        static_assert(!std::is_const<T>::value, "get_to() cannot deserialize into a const value");
         JSONSerializer<Array>::from_json(*this, v);
         return v;
     }
@@ -2931,6 +3227,7 @@ class basic_json // NOLINT(cppcoreguidelines-special-member-functions,hicpp-spec
             case value_t::object:
             {
                 result.m_it.object_iterator = erase_from_object(pos.m_it.object_iterator);
+                set_parents_after_object_erase();
                 break;
             }
 
@@ -3003,6 +3300,7 @@ class basic_json // NOLINT(cppcoreguidelines-special-member-functions,hicpp-spec
             {
                 result.m_it.object_iterator = m_data.m_value.object->erase(first.m_it.object_iterator,
                                               last.m_it.object_iterator);
+                set_parents_after_object_erase();
                 break;
             }
 
@@ -3033,7 +3331,9 @@ class basic_json // NOLINT(cppcoreguidelines-special-member-functions,hicpp-spec
             JSON_THROW(type_error::create(307, detail::concat("cannot use erase() with ", type_name()), this));
         }
 
-        return m_data.m_value.object->erase(std::forward<KeyType>(key));
+        const auto erased = m_data.m_value.object->erase(std::forward<KeyType>(key));
+        set_parents_after_object_erase();
+        return erased;
     }
 
     template < typename KeyType, detail::enable_if_t <
@@ -3050,6 +3350,7 @@ class basic_json // NOLINT(cppcoreguidelines-special-member-functions,hicpp-spec
         if (it != m_data.m_value.object->end())
         {
             m_data.m_value.object->erase(it);
+            set_parents_after_object_erase();
             return 1;
         }
         return 0;
@@ -3961,16 +4262,12 @@ class basic_json // NOLINT(cppcoreguidelines-special-member-functions,hicpp-spec
                 if (it2 != m_data.m_value.object->end() && it2->second.is_object())
                 {
                     it2->second.update_members(it.value().cbegin(), it.value().cend(), true, depth + 1);
-#if JSON_DIAGNOSTICS
-                    it2->second.set_parents();
-#endif
                     continue;
                 }
             }
-            m_data.m_value.object->operator[](it.key()) = it.value();
-#if JSON_DIAGNOSTICS
-            m_data.m_value.object->operator[](it.key()).m_parent = this;
-#endif
+            // set_parent() also repairs the other members, which ordered_json
+            // relocates when adding a key makes its vector grow
+            set_parent(m_data.m_value.object->operator[](it.key()) = it.value());
         }
     }
 
@@ -3999,9 +4296,6 @@ class basic_json // NOLINT(cppcoreguidelines-special-member-functions,hicpp-spec
                 }
 
                 // a nested object is merged: continue with its parent
-#if JSON_DIAGNOSTICS
-                target->set_parents();
-#endif
                 target = stack.back().target;
                 first = stack.back().position;
                 last = stack.back().last;
@@ -4023,10 +4317,9 @@ class basic_json // NOLINT(cppcoreguidelines-special-member-functions,hicpp-spec
                     continue;
                 }
             }
-            target->m_data.m_value.object->operator[](first.key()) = first.value();
-#if JSON_DIAGNOSTICS
-            target->m_data.m_value.object->operator[](first.key()).m_parent = target;
-#endif
+            // set_parent() also repairs the other members, which ordered_json
+            // relocates when adding a key makes its vector grow
+            target->set_parent(target->m_data.m_value.object->operator[](first.key()) = first.value());
             ++first;
         }
     }
@@ -4165,7 +4458,7 @@ class basic_json // NOLINT(cppcoreguidelines-special-member-functions,hicpp-spec
     // because any negative signed value is smaller than any unsigned value.
     // Otherwise, the non-negative signed value is cast to unsigned before the
     // comparison to avoid wraparound.
-#define JSON_IMPLEMENT_OPERATOR(op, null_result, unordered_result, default_result)                       \
+#define JSON_IMPLEMENT_OPERATOR(op, null_result, unordered_result, default_result, deep_result, may_descend) \
     const auto lhs_type = lhs.type();                                                                    \
     const auto rhs_type = rhs.type();                                                                    \
     \
@@ -4174,11 +4467,25 @@ class basic_json // NOLINT(cppcoreguidelines-special-member-functions,hicpp-spec
         switch (lhs_type)                                                                                \
         {                                                                                                \
             case value_t::array:                                                                         \
+            {                                                                                            \
+                if (JSON_HEDLEY_UNLIKELY(nesting_depth_exhausted(may_descend)))                        \
+                {                                                                                        \
+                    return (deep_result);                                                                \
+                }                                                                                        \
+                const nesting_depth_guard guard;                                                         \
                 return (*lhs.m_data.m_value.array) op (*rhs.m_data.m_value.array);                                     \
-                \
+            }                                                                                            \
+            \
             case value_t::object:                                                                        \
+            {                                                                                            \
+                if (JSON_HEDLEY_UNLIKELY(nesting_depth_exhausted(may_descend)))                        \
+                {                                                                                        \
+                    return (deep_result);                                                                \
+                }                                                                                        \
+                const nesting_depth_guard guard;                                                         \
                 return (*lhs.m_data.m_value.object) op (*rhs.m_data.m_value.object);                                   \
-                \
+            }                                                                                            \
+            \
             case value_t::null:                                                                          \
                 return (null_result);                                                                    \
                 \
@@ -4278,7 +4585,8 @@ class basic_json // NOLINT(cppcoreguidelines-special-member-functions,hicpp-spec
         JSON_HEDLEY_PRAGMA(GCC diagnostic ignored "-Wfloat-equal")
 #endif
         const_reference lhs = *this;
-        JSON_IMPLEMENT_OPERATOR( ==, true, false, false)
+        JSON_IMPLEMENT_OPERATOR( ==, true, false, false,
+                                 compare_iteratively<false>(lhs, rhs, false) == compare_result::equal, true)
 #ifdef __GNUC__
         JSON_HEDLEY_DIAGNOSTIC_POP
 #endif
@@ -4303,7 +4611,8 @@ class basic_json // NOLINT(cppcoreguidelines-special-member-functions,hicpp-spec
         JSON_IMPLEMENT_OPERATOR(<=>, // *NOPAD*
                                 std::partial_ordering::equivalent,
                                 std::partial_ordering::unordered,
-                                lhs_type <=> rhs_type) // *NOPAD*
+                                lhs_type <=> rhs_type, // *NOPAD*
+                                to_partial_ordering(compare_iteratively<true>(lhs, rhs, false)), true)
     }
 
     /// @brief comparison: 3-way
@@ -4370,7 +4679,8 @@ class basic_json // NOLINT(cppcoreguidelines-special-member-functions,hicpp-spec
         JSON_HEDLEY_DIAGNOSTIC_PUSH
         JSON_HEDLEY_PRAGMA(GCC diagnostic ignored "-Wfloat-equal")
 #endif
-        JSON_IMPLEMENT_OPERATOR( ==, true, false, false)
+        JSON_IMPLEMENT_OPERATOR( ==, true, false, false,
+                                 compare_iteratively<false>(lhs, rhs, false) == compare_result::equal, true)
 #ifdef __GNUC__
         JSON_HEDLEY_DIAGNOSTIC_POP
 #endif
@@ -4426,7 +4736,8 @@ class basic_json // NOLINT(cppcoreguidelines-special-member-functions,hicpp-spec
         // default_result is used if we cannot compare values. In that case,
         // we compare types. Note we have to call the operator explicitly,
         // because MSVC has problems otherwise.
-        JSON_IMPLEMENT_OPERATOR( <, false, false, operator<(lhs_type, rhs_type))
+        JSON_IMPLEMENT_OPERATOR( <, false, false, operator<(lhs_type, rhs_type),
+                                 compare_iteratively<true>(lhs, rhs, true) == compare_result::less, false)
     }
 
     /// @brief comparison: less than

single_include/nlohmann/json.hpp

@@ -95,6 +95,10 @@
     #define JSON_BRACE_INIT_COPY_SEMANTICS 0
 #endif
 
+#ifndef JSON_PRECISE_STREAM_POSITION
+    #define JSON_PRECISE_STREAM_POSITION 0
+#endif
+
 #if JSON_DIAGNOSTICS
     #define NLOHMANN_JSON_ABI_TAG_DIAGNOSTICS _diag
 #else
@@ -119,21 +123,28 @@
     #define NLOHMANN_JSON_ABI_TAG_BRACE_INIT_COPY_SEMANTICS
 #endif
 
+#if JSON_PRECISE_STREAM_POSITION
+    #define NLOHMANN_JSON_ABI_TAG_PRECISE_STREAM_POSITION _psp
+#else
+    #define NLOHMANN_JSON_ABI_TAG_PRECISE_STREAM_POSITION
+#endif
+
 #ifndef NLOHMANN_JSON_NAMESPACE_NO_VERSION
     #define NLOHMANN_JSON_NAMESPACE_NO_VERSION 0
 #endif
 
 // Construct the namespace ABI tags component
-#define NLOHMANN_JSON_ABI_TAGS_CONCAT_EX(a, b, c, d) json_abi ## a ## b ## c ## d
-#define NLOHMANN_JSON_ABI_TAGS_CONCAT(a, b, c, d) \
-    NLOHMANN_JSON_ABI_TAGS_CONCAT_EX(a, b, c, d)
+#define NLOHMANN_JSON_ABI_TAGS_CONCAT_EX(a, b, c, d, e) json_abi ## a ## b ## c ## d ## e
+#define NLOHMANN_JSON_ABI_TAGS_CONCAT(a, b, c, d, e) \
+    NLOHMANN_JSON_ABI_TAGS_CONCAT_EX(a, b, c, d, e)
 
 #define NLOHMANN_JSON_ABI_TAGS                                       \
     NLOHMANN_JSON_ABI_TAGS_CONCAT(                                   \
             NLOHMANN_JSON_ABI_TAG_DIAGNOSTICS,                       \
             NLOHMANN_JSON_ABI_TAG_LEGACY_DISCARDED_VALUE_COMPARISON, \
             NLOHMANN_JSON_ABI_TAG_DIAGNOSTIC_POSITIONS,              \
-            NLOHMANN_JSON_ABI_TAG_BRACE_INIT_COPY_SEMANTICS)
+            NLOHMANN_JSON_ABI_TAG_BRACE_INIT_COPY_SEMANTICS,         \
+            NLOHMANN_JSON_ABI_TAG_PRECISE_STREAM_POSITION)
 
 // Construct the namespace version component
 #define NLOHMANN_JSON_NAMESPACE_VERSION_CONCAT_EX(major, minor, patch) \
@@ -7419,6 +7430,11 @@ class input_stream_adapter
         // maintain ifstream flags, except eof
         if (is != nullptr)
         {
+#if JSON_PRECISE_STREAM_POSITION
+            // consume the character last returned by get_character() unless it
+            // was given back with release_lookahead()
+            commit_lookahead();
+#endif
             is->clear(is->rdstate() & std::ios::eofbit);
         }
     }
@@ -7432,6 +7448,58 @@ class input_stream_adapter
     input_stream_adapter& operator=(input_stream_adapter&) = delete;
     input_stream_adapter& operator=(input_stream_adapter&&) = delete;
 
+#if JSON_PRECISE_STREAM_POSITION
+    input_stream_adapter(input_stream_adapter&& rhs) noexcept
+        : is(rhs.is), sb(rhs.sb), lookahead(rhs.lookahead)
+    {
+        rhs.is = nullptr;
+        rhs.sb = nullptr;
+        rhs.lookahead = false;
+    }
+
+    // Whether the character last returned by get_character() can be given back
+    // to the input with release_lookahead().
+    static constexpr bool supports_lookahead = true;
+
+    // std::istream/std::streambuf use std::char_traits<char>::to_int_type, to
+    // ensure that std::char_traits<char>::eof() and the character 0xFF do not
+    // end up as the same value, e.g., 0xFFFFFFFF.
+    //
+    // The character is peeked rather than consumed: it is only stepped over
+    // once the next character is requested, or when the adapter is destroyed.
+    // Until then, release_lookahead() can leave it in the input.
+    std::char_traits<char>::int_type get_character()
+    {
+        if (lookahead)
+        {
+            // step over the character returned by the previous call
+            sb->sbumpc();
+        }
+
+        auto res = sb->sgetc();
+        // set eof manually, as we don't use the istream interface.
+        if (JSON_HEDLEY_UNLIKELY(res == std::char_traits<char>::eof()))
+        {
+            // there is nothing to step over next time
+            lookahead = false;
+            is->clear(is->rdstate() | std::ios::eofbit);
+        }
+        else
+        {
+            lookahead = true;
+        }
+        return res;
+    }
+
+    // Leave the character last returned by get_character() in the input, so
+    // that the next read from the stream - by this adapter or by the caller
+    // once parsing is done - sees it again. Unlike putting a consumed
+    // character back, this cannot fail.
+    void release_lookahead() noexcept
+    {
+        lookahead = false;
+    }
+#else
     input_stream_adapter(input_stream_adapter&& rhs) noexcept
         : is(rhs.is), sb(rhs.sb)
     {
@@ -7442,6 +7510,9 @@ class input_stream_adapter
     // std::istream/std::streambuf use std::char_traits<char>::to_int_type, to
     // ensure that std::char_traits<char>::eof() and the character 0xFF do not
     // end up as the same value, e.g., 0xFFFFFFFF.
+    //
+    // The character is consumed, so the character that terminates a number
+    // stays consumed after parsing; see JSON_PRECISE_STREAM_POSITION.
     std::char_traits<char>::int_type get_character()
     {
         auto res = sb->sbumpc();
@@ -7452,10 +7523,14 @@ class input_stream_adapter
         }
         return res;
     }
+#endif
 
     template<class T>
     std::size_t get_elements(T* dest, std::size_t count = 1)
     {
+#if JSON_PRECISE_STREAM_POSITION
+        commit_lookahead();
+#endif
         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)))
         {
@@ -7465,9 +7540,27 @@ class input_stream_adapter
     }
 
   private:
+#if JSON_PRECISE_STREAM_POSITION
+    // Step over the character last returned by get_character(). The character
+    // has already been peeked successfully, so for every streambuf with a get
+    // area this is a pointer increment that cannot fail.
+    void commit_lookahead()
+    {
+        if (lookahead)
+        {
+            lookahead = false;
+            sb->sbumpc();
+        }
+    }
+#endif
+
     /// the associated input stream
     std::istream* is = nullptr;
     std::streambuf* sb = nullptr;
+#if JSON_PRECISE_STREAM_POSITION
+    /// whether get_character() peeked a character that is not consumed yet
+    bool lookahead = false;
+#endif
 };
 #endif  // JSON_NO_IO
 
@@ -8879,6 +8972,25 @@ constexpr bool input_adapter_supports_seek(std::false_type /*detected*/)
     return false;
 }
 
+// Detect whether an input adapter reads with one character of lookahead that
+// can be left in the input (see input_stream_adapter::supports_lookahead,
+// which is only defined with JSON_PRECISE_STREAM_POSITION), detected like
+// supports_seek above.
+template<typename InputAdapterType>
+using detect_supports_lookahead = decltype(InputAdapterType::supports_lookahead);
+
+template<typename InputAdapterType>
+constexpr bool input_adapter_supports_lookahead(std::true_type /*detected*/)
+{
+    return InputAdapterType::supports_lookahead;
+}
+
+template<typename InputAdapterType>
+constexpr bool input_adapter_supports_lookahead(std::false_type /*detected*/)
+{
+    return false;
+}
+
 // Detect whether an input adapter exposes a contiguous byte block that the
 // lexer can scan directly (see iterator_input_adapter::supports_bulk_scan).
 // Adapters without the flag - file, stream, wide-string, user-defined - fall
@@ -8919,6 +9031,12 @@ class lexer : public lexer_base<BasicJsonType>
     static constexpr bool lazy_token_string =
         input_adapter_supports_seek<InputAdapterType>(is_detected<detect_supports_seek, InputAdapterType> {});
 
+    /// whether a simulated unget can be passed on to the input adapter, which
+    /// then leaves the character in the input; see
+    /// input_adapter_supports_lookahead
+    static constexpr bool can_release_lookahead =
+        input_adapter_supports_lookahead<InputAdapterType>(is_detected<detect_supports_lookahead, InputAdapterType> {});
+
     /// whether string scanning may bulk-consume runs of ordinary characters
     /// directly from a contiguous input buffer (SWAR fast path). This requires
     /// the token to be reconstructible lazily (lazy_token_string), so bypassing
@@ -10650,6 +10768,21 @@ scan_number_done:
         uncapture_char(std::integral_constant<bool, lazy_token_string> {});
     }
 
+    /// adapter without lookahead: nothing to do (see release_lookahead)
+    void release_lookahead_impl(std::false_type /*can_release*/) const noexcept {}
+
+    /// adapter with lookahead: leave the character in the input instead
+    void release_lookahead_impl(std::true_type /*can_release*/)
+    {
+        if (next_unget)
+        {
+            // the character is read from the input again rather than replayed
+            // from current, so the adapter must not step over it
+            next_unget = false;
+            ia.release_lookahead();
+        }
+    }
+
     /// seekable adapter: nothing was captured, so nothing to undo
     void uncapture_char(std::true_type /*lazy*/) const noexcept {}
 
@@ -10713,6 +10846,31 @@ scan_number_done:
         return position;
     }
 
+    /*!
+    @brief pass a pending simulated unget on to the input
+
+    unget() only rewinds the lexer's own bookkeeping, so the character that
+    terminated the last token (e.g. the character after a number) would still
+    be stepped over when the input adapter is done. Callers that hand the
+    input back to the user afterwards - operator>> and non-strict sax_parse -
+    call this once when scanning is done, so that the input is positioned
+    right after the value.
+
+    Adapters without lookahead (see input_adapter_supports_lookahead) are not
+    handed back to the user, so this is a no-op for them. Without
+    JSON_PRECISE_STREAM_POSITION, no adapter has lookahead, so this is always a
+    no-op and the terminating character stays consumed.
+
+    Scanning may continue after this call: @a next_unget is cleared, and the
+    character is read from the input again instead of being replayed from
+    @a current. A pending unget of EOF needs no special case, because reaching
+    EOF leaves no lookahead to release.
+    */
+    void release_lookahead()
+    {
+        release_lookahead_impl(std::integral_constant<bool, can_release_lookahead> {});
+    }
+
 #if JSON_DIAGNOSTIC_POSITIONS
     /// return the offset of the first character of the last read token; unlike
     /// the token's parsed value, this accounts for escape sequences
@@ -15960,13 +16118,22 @@ class parser
             json_sax_dom_callback_parser<BasicJsonType, InputAdapterType> sdp(result, callback, allow_exceptions, &m_lexer);
             sax_parse_internal(&sdp);
 
-            // in strict mode, input must be completely read
-            if (strict && (get_token() != token_type::end_of_input))
+            if (strict)
             {
-                sdp.parse_error(m_lexer.get_position(),
-                                m_lexer.get_token_string(),
-                                parse_error::create(101, m_lexer.get_position(),
-                                                    exception_message(token_type::end_of_input, "value"), nullptr));
+                // in strict mode, input must be completely read
+                if (get_token() != token_type::end_of_input)
+                {
+                    sdp.parse_error(m_lexer.get_position(),
+                                    m_lexer.get_token_string(),
+                                    parse_error::create(101, m_lexer.get_position(),
+                                                        exception_message(token_type::end_of_input, "value"), nullptr));
+                }
+            }
+            else
+            {
+                // the caller keeps using the input: position it right after
+                // the value by leaving the character that terminated it
+                m_lexer.release_lookahead();
             }
 
             // in case of an error, return a discarded value
@@ -15988,12 +16155,20 @@ class parser
             json_sax_dom_parser<BasicJsonType, InputAdapterType> sdp(result, allow_exceptions, &m_lexer);
             sax_parse_internal(&sdp);
 
-            // in strict mode, input must be completely read
-            if (strict && (get_token() != token_type::end_of_input))
+            if (strict)
             {
-                sdp.parse_error(m_lexer.get_position(),
-                                m_lexer.get_token_string(),
-                                parse_error::create(101, m_lexer.get_position(), exception_message(token_type::end_of_input, "value"), nullptr));
+                // in strict mode, input must be completely read
+                if (get_token() != token_type::end_of_input)
+                {
+                    sdp.parse_error(m_lexer.get_position(),
+                                    m_lexer.get_token_string(),
+                                    parse_error::create(101, m_lexer.get_position(), exception_message(token_type::end_of_input, "value"), nullptr));
+                }
+            }
+            else
+            {
+                // see above
+                m_lexer.release_lookahead();
             }
 
             // in case of an error, return a discarded value
@@ -16026,12 +16201,24 @@ class parser
         (void)detail::is_sax_static_asserts<SAX, BasicJsonType> {};
         const bool result = sax_parse_internal(sax);
 
-        // strict mode: next byte must be EOF
-        if (result && strict && (get_token() != token_type::end_of_input))
+        if (result)
         {
-            return sax->parse_error(m_lexer.get_position(),
-                                    m_lexer.get_token_string(),
-                                    parse_error::create(101, m_lexer.get_position(), exception_message(token_type::end_of_input, "value"), nullptr));
+            if (strict)
+            {
+                // strict mode: next byte must be EOF
+                if (get_token() != token_type::end_of_input)
+                {
+                    return sax->parse_error(m_lexer.get_position(),
+                                            m_lexer.get_token_string(),
+                                            parse_error::create(101, m_lexer.get_position(), exception_message(token_type::end_of_input, "value"), nullptr));
+                }
+            }
+            else
+            {
+                // the caller keeps using the input: position it right after
+                // the value by leaving the character that terminated it
+                m_lexer.release_lookahead();
+            }
         }
 
         return result;
@@ -25382,6 +25569,31 @@ class basic_json // NOLINT(cppcoreguidelines-special-member-functions,hicpp-spec
     }
 #endif
 
+    /*!
+    @brief whether a descent must stop here and finish without the call stack
+
+    @a may_descend says whether the operator descends at all; it is a constant
+    at every call site, and is passed rather than tested by the caller so that
+    the test does not become a constant condition there, which MSVC reports as
+    C4127.
+
+    The comparison operators use this rather than @ref nesting_depth_guard::okay,
+    because they are written as a macro and a macro cannot use the preprocessor
+    the way the guard's constructor does; @ref copy_structured, which can, asks
+    the guard instead and never calls this.
+    */
+    static bool nesting_depth_exhausted(bool may_descend = true) noexcept
+    {
+#ifdef JSON_NO_THREAD_LOCAL
+        // without a count of its own per thread, a descent cannot be bounded
+        // without racing another one, so none is made
+        static_cast<void>(may_descend);
+        return true;
+#else
+        return !may_descend || nesting_depth() >= nesting_depth_limit();
+#endif
+    }
+
     /*!
     @brief counts one level of a bounded descent for as long as it runs, and
            reports whether the descent was still within the limit when it began
@@ -25436,7 +25648,8 @@ class basic_json // NOLINT(cppcoreguidelines-special-member-functions,hicpp-spec
     using copy_worklist_t = std::vector<std::pair<const basic_json*, basic_json*>>;
 
     /// scratch space to build the key skeleton of an object copy in one go
-    using copy_scratch_t = std::vector<std::pair<typename object_t::key_type, basic_json>>;
+    using copy_scratch_value_t = std::pair<typename object_t::key_type, basic_json>;
+    using copy_scratch_t = std::vector<copy_scratch_value_t, AllocatorType<copy_scratch_value_t>>;
 
     /// @brief copy everything of @a src into @a dst but its type and value
     static void copy_metadata(const basic_json& src, basic_json& dst)
@@ -25701,6 +25914,274 @@ class basic_json // NOLINT(cppcoreguidelines-special-member-functions,hicpp-spec
     }
 
 
+    /// the result of comparing two values, including values that cannot be
+    /// ordered at all, such as a discarded value or a NaN
+    enum class compare_result { less, equal, greater, unordered };
+
+#if JSON_HAS_THREE_WAY_COMPARISON
+    /// @brief the ordering that @a result stands for
+    static std::partial_ordering to_partial_ordering(compare_result result) noexcept // *NOPAD*
+    {
+        switch (result)
+        {
+            case compare_result::less:
+                return std::partial_ordering::less;
+            case compare_result::greater:
+                return std::partial_ordering::greater;
+            case compare_result::equal:
+                return std::partial_ordering::equivalent;
+            case compare_result::unordered:
+            default:
+                return std::partial_ordering::unordered;
+        }
+    }
+#endif
+
+    /*!
+    @brief compare two values that are not both an array or both an object
+
+    Such a pair is compared by the operators themselves, which cannot descend
+    into it and therefore cannot recurse.
+
+    That holds for a pair whose types differ as much as for a pair of leaves: an
+    array and an object are told apart by their types alone, because an operator
+    only ever descends into two values of the same type. So `==` reports them as
+    unequal without looking inside either, and an ordering falls back to the
+    order of the types - an object sorts before an array - exactly as it does
+    for a value that is not nested deeply enough to get here.
+    */
+    template<bool Ordered>
+    static compare_result compare_leaves(const_reference lhs, const_reference rhs) noexcept
+    {
+        if (lhs == rhs)
+        {
+            return compare_result::equal;
+        }
+
+        return order_leaves(lhs, rhs, std::integral_constant<bool, Ordered> {});
+    }
+
+    /*!
+    @brief compare two object keys
+
+    An object compares its entries as pairs of a key and a value, so its keys
+    are compared exactly as std::pair compares them: with < where the objects
+    are being ordered, and with == where they are only checked for equality.
+    Note that this is not the object's own comparator, which for a vector-backed
+    object type such as nlohmann::ordered_map tells equality rather than order.
+    */
+    static compare_result compare_keys(const typename object_t::key_type& lhs,
+                                       const typename object_t::key_type& rhs,
+                                       std::true_type /*ordered*/)
+    {
+        if (lhs < rhs)
+        {
+            return compare_result::less;
+        }
+
+        if (rhs < lhs)
+        {
+            return compare_result::greater;
+        }
+
+        return compare_result::equal;
+    }
+
+    /// @brief check two object keys for equality
+    static compare_result compare_keys(const typename object_t::key_type& lhs,
+                                       const typename object_t::key_type& rhs,
+                                       std::false_type /*ordered*/)
+    {
+        return lhs == rhs ? compare_result::equal : compare_result::unordered;
+    }
+
+    /// @brief tell apart two values that are not equal
+    /// @note only instantiated where the values are being ordered, as a key or
+    ///       string type is not required to be ordered to be compared for equality
+    static compare_result order_leaves(const_reference lhs, const_reference rhs, std::true_type /*ordered*/) noexcept
+    {
+        if (lhs < rhs)
+        {
+            return compare_result::less;
+        }
+
+        if (rhs < lhs)
+        {
+            return compare_result::greater;
+        }
+
+        return compare_result::unordered;
+    }
+
+    /// @brief report two values as not equal without ordering them
+    static compare_result order_leaves(const_reference /*lhs*/, const_reference /*rhs*/, std::false_type /*ordered*/) noexcept
+    {
+        return compare_result::unordered;
+    }
+
+    /*!
+    @brief compare @a lhs and @a rhs without descending into them
+
+    Reached once a comparison has descended @ref nesting_depth_limit levels, so
+    that comparing values cannot exhaust the call stack however deeply they are
+    nested. The two values are walked in lockstep on an explicit stack and
+    compared lexicographically, element by element in the order the containers
+    enumerate them - which is how the container types this library ships compare
+    themselves: a std::map enumerates its entries in key order, and
+    nlohmann::ordered_map in insertion order. An object type that enumerates its
+    entries in an unspecified order, such as std::unordered_map, compares them
+    pairwise instead; the difference could only ever show below the bound.
+
+    Note that the stack this walks with is allocated, while the comparison
+    operators are noexcept and the container comparison this replaces allocated
+    nothing. Failing that allocation therefore ends the process rather than
+    throwing. It only arises for values nested past the bound, and only when
+    memory has run out - where the same comparison used to exhaust the call
+    stack instead - but it is a way to fail that the operators did not have.
+    */
+    template<bool Ordered>
+    static compare_result compare_iteratively(const_reference lhs, const_reference rhs,
+            const bool unordered_compares_equal) noexcept
+    {
+        /// a pair of containers being compared in lockstep
+        struct frame
+        {
+            const basic_json* lhs_value{nullptr};
+            const basic_json* rhs_value{nullptr};
+            typename array_t::const_iterator lhs_array_it{};
+            typename array_t::const_iterator rhs_array_it{};
+            typename object_t::const_iterator lhs_object_it{};
+            typename object_t::const_iterator rhs_object_it{};
+        };
+
+        std::vector<frame> stack;
+        const basic_json* left = &lhs;
+        const basic_json* right = &rhs;
+
+        for (;;)
+        {
+            const auto type = left->m_data.m_type;
+
+            if (type == right->m_data.m_type && (type == value_t::array || type == value_t::object))
+            {
+                // descend: the elements decide, and are compared further down
+                stack.emplace_back();
+                frame& pushed = stack.back();
+                pushed.lhs_value = left;
+                pushed.rhs_value = right;
+
+                if (type == value_t::array)
+                {
+                    pushed.lhs_array_it = left->m_data.m_value.array->cbegin();
+                    pushed.rhs_array_it = right->m_data.m_value.array->cbegin();
+                }
+                else
+                {
+                    pushed.lhs_object_it = left->m_data.m_value.object->cbegin();
+                    pushed.rhs_object_it = right->m_data.m_value.object->cbegin();
+                }
+            }
+            else
+            {
+                const compare_result result = compare_leaves<Ordered>(*left, *right);
+
+                // Values that cannot be ordered - a NaN, say - end an ordered
+                // comparison for std::lexicographical_compare_three_way, but
+                // std::lexicographical_compare treats them as equivalent and
+                // carries on with the next element. Both are reproduced here,
+                // so that a value nested too deeply to descend into compares
+                // exactly as one that is not.
+                if (result != compare_result::equal &&
+                        !(unordered_compares_equal && result == compare_result::unordered))
+                {
+                    return result;
+                }
+            }
+
+            // walk back up past the containers that are exhausted, then take the
+            // next pair of elements from the innermost one that is not
+            for (;;)
+            {
+                if (stack.empty())
+                {
+                    return compare_result::equal;
+                }
+
+                frame& current = stack.back();
+                const bool is_object = current.lhs_value->m_data.m_type == value_t::object;
+
+                const bool lhs_done = is_object
+                                      ? current.lhs_object_it == current.lhs_value->m_data.m_value.object->cend()
+                                      : current.lhs_array_it == current.lhs_value->m_data.m_value.array->cend();
+                const bool rhs_done = is_object
+                                      ? current.rhs_object_it == current.rhs_value->m_data.m_value.object->cend()
+                                      : current.rhs_array_it == current.rhs_value->m_data.m_value.array->cend();
+
+                if (lhs_done || rhs_done)
+                {
+                    // whichever ran out first holds the smaller container; if
+                    // both did, they are equal and the container above decides
+                    if (lhs_done != rhs_done)
+                    {
+                        return lhs_done ? compare_result::less : compare_result::greater;
+                    }
+
+                    stack.pop_back();
+                    continue;
+                }
+
+                if (is_object)
+                {
+                    // an entry is a key and a value, and the key decides first
+                    const compare_result key_result =
+                        compare_keys(current.lhs_object_it->first, current.rhs_object_it->first,
+                                     std::integral_constant<bool, Ordered> {});
+
+                    if (key_result != compare_result::equal)
+                    {
+                        return key_result;
+                    }
+
+                    left = &(current.lhs_object_it->second);
+                    right = &(current.rhs_object_it->second);
+                    ++current.lhs_object_it;
+                    ++current.rhs_object_it;
+                }
+                else
+                {
+                    left = &(*current.lhs_array_it);
+                    right = &(*current.rhs_array_it);
+                    ++current.lhs_array_it;
+                    ++current.rhs_array_it;
+                }
+
+                break;
+            }
+        }
+    }
+
+
+    /// @brief restore the parent pointers after erasing from an object
+    /// ordered_json keeps its members in a vector, and erasing a member
+    /// re-constructs every member after it in place, which resets their
+    /// parent pointers
+    void set_parents_after_object_erase()
+    {
+#if JSON_DIAGNOSTICS
+#ifdef JSON_HEDLEY_MSVC_VERSION
+#pragma warning(push )
+#pragma warning(disable : 4127) // ignore warning to replace if with if constexpr
+#endif
+        if (detail::is_ordered_map<object_t>::value)
+        {
+            set_parents();
+        }
+#ifdef JSON_HEDLEY_MSVC_VERSION
+#pragma warning( pop )
+#endif
+#endif
+    }
+
   public:
     //////////////////////////
     // JSON parser callback //
@@ -26718,6 +27199,7 @@ class basic_json // NOLINT(cppcoreguidelines-special-member-functions,hicpp-spec
     ValueType & get_to(ValueType& v) const noexcept(noexcept(
             JSONSerializer<ValueType>::from_json(std::declval<const basic_json_t&>(), v)))
     {
+        static_assert(!std::is_const<ValueType>::value, "get_to() cannot deserialize into a const value");
         JSONSerializer<ValueType>::from_json(*this, v);
         return v;
     }
@@ -26743,6 +27225,7 @@ class basic_json // NOLINT(cppcoreguidelines-special-member-functions,hicpp-spec
     noexcept(noexcept(JSONSerializer<Array>::from_json(
                           std::declval<const basic_json_t&>(), v)))
     {
+        static_assert(!std::is_const<T>::value, "get_to() cannot deserialize into a const value");
         JSONSerializer<Array>::from_json(*this, v);
         return v;
     }
@@ -27389,6 +27872,7 @@ class basic_json // NOLINT(cppcoreguidelines-special-member-functions,hicpp-spec
             case value_t::object:
             {
                 result.m_it.object_iterator = erase_from_object(pos.m_it.object_iterator);
+                set_parents_after_object_erase();
                 break;
             }
 
@@ -27461,6 +27945,7 @@ class basic_json // NOLINT(cppcoreguidelines-special-member-functions,hicpp-spec
             {
                 result.m_it.object_iterator = m_data.m_value.object->erase(first.m_it.object_iterator,
                                               last.m_it.object_iterator);
+                set_parents_after_object_erase();
                 break;
             }
 
@@ -27491,7 +27976,9 @@ class basic_json // NOLINT(cppcoreguidelines-special-member-functions,hicpp-spec
             JSON_THROW(type_error::create(307, detail::concat("cannot use erase() with ", type_name()), this));
         }
 
-        return m_data.m_value.object->erase(std::forward<KeyType>(key));
+        const auto erased = m_data.m_value.object->erase(std::forward<KeyType>(key));
+        set_parents_after_object_erase();
+        return erased;
     }
 
     template < typename KeyType, detail::enable_if_t <
@@ -27508,6 +27995,7 @@ class basic_json // NOLINT(cppcoreguidelines-special-member-functions,hicpp-spec
         if (it != m_data.m_value.object->end())
         {
             m_data.m_value.object->erase(it);
+            set_parents_after_object_erase();
             return 1;
         }
         return 0;
@@ -28419,16 +28907,12 @@ class basic_json // NOLINT(cppcoreguidelines-special-member-functions,hicpp-spec
                 if (it2 != m_data.m_value.object->end() && it2->second.is_object())
                 {
                     it2->second.update_members(it.value().cbegin(), it.value().cend(), true, depth + 1);
-#if JSON_DIAGNOSTICS
-                    it2->second.set_parents();
-#endif
                     continue;
                 }
             }
-            m_data.m_value.object->operator[](it.key()) = it.value();
-#if JSON_DIAGNOSTICS
-            m_data.m_value.object->operator[](it.key()).m_parent = this;
-#endif
+            // set_parent() also repairs the other members, which ordered_json
+            // relocates when adding a key makes its vector grow
+            set_parent(m_data.m_value.object->operator[](it.key()) = it.value());
         }
     }
 
@@ -28457,9 +28941,6 @@ class basic_json // NOLINT(cppcoreguidelines-special-member-functions,hicpp-spec
                 }
 
                 // a nested object is merged: continue with its parent
-#if JSON_DIAGNOSTICS
-                target->set_parents();
-#endif
                 target = stack.back().target;
                 first = stack.back().position;
                 last = stack.back().last;
@@ -28481,10 +28962,9 @@ class basic_json // NOLINT(cppcoreguidelines-special-member-functions,hicpp-spec
                     continue;
                 }
             }
-            target->m_data.m_value.object->operator[](first.key()) = first.value();
-#if JSON_DIAGNOSTICS
-            target->m_data.m_value.object->operator[](first.key()).m_parent = target;
-#endif
+            // set_parent() also repairs the other members, which ordered_json
+            // relocates when adding a key makes its vector grow
+            target->set_parent(target->m_data.m_value.object->operator[](first.key()) = first.value());
             ++first;
         }
     }
@@ -28623,7 +29103,7 @@ class basic_json // NOLINT(cppcoreguidelines-special-member-functions,hicpp-spec
     // because any negative signed value is smaller than any unsigned value.
     // Otherwise, the non-negative signed value is cast to unsigned before the
     // comparison to avoid wraparound.
-#define JSON_IMPLEMENT_OPERATOR(op, null_result, unordered_result, default_result)                       \
+#define JSON_IMPLEMENT_OPERATOR(op, null_result, unordered_result, default_result, deep_result, may_descend) \
     const auto lhs_type = lhs.type();                                                                    \
     const auto rhs_type = rhs.type();                                                                    \
     \
@@ -28632,11 +29112,25 @@ class basic_json // NOLINT(cppcoreguidelines-special-member-functions,hicpp-spec
         switch (lhs_type)                                                                                \
         {                                                                                                \
             case value_t::array:                                                                         \
+            {                                                                                            \
+                if (JSON_HEDLEY_UNLIKELY(nesting_depth_exhausted(may_descend)))                        \
+                {                                                                                        \
+                    return (deep_result);                                                                \
+                }                                                                                        \
+                const nesting_depth_guard guard;                                                         \
                 return (*lhs.m_data.m_value.array) op (*rhs.m_data.m_value.array);                                     \
-                \
+            }                                                                                            \
+            \
             case value_t::object:                                                                        \
+            {                                                                                            \
+                if (JSON_HEDLEY_UNLIKELY(nesting_depth_exhausted(may_descend)))                        \
+                {                                                                                        \
+                    return (deep_result);                                                                \
+                }                                                                                        \
+                const nesting_depth_guard guard;                                                         \
                 return (*lhs.m_data.m_value.object) op (*rhs.m_data.m_value.object);                                   \
-                \
+            }                                                                                            \
+            \
             case value_t::null:                                                                          \
                 return (null_result);                                                                    \
                 \
@@ -28736,7 +29230,8 @@ class basic_json // NOLINT(cppcoreguidelines-special-member-functions,hicpp-spec
         JSON_HEDLEY_PRAGMA(GCC diagnostic ignored "-Wfloat-equal")
 #endif
         const_reference lhs = *this;
-        JSON_IMPLEMENT_OPERATOR( ==, true, false, false)
+        JSON_IMPLEMENT_OPERATOR( ==, true, false, false,
+                                 compare_iteratively<false>(lhs, rhs, false) == compare_result::equal, true)
 #ifdef __GNUC__
         JSON_HEDLEY_DIAGNOSTIC_POP
 #endif
@@ -28761,7 +29256,8 @@ class basic_json // NOLINT(cppcoreguidelines-special-member-functions,hicpp-spec
         JSON_IMPLEMENT_OPERATOR(<=>, // *NOPAD*
                                 std::partial_ordering::equivalent,
                                 std::partial_ordering::unordered,
-                                lhs_type <=> rhs_type) // *NOPAD*
+                                lhs_type <=> rhs_type, // *NOPAD*
+                                to_partial_ordering(compare_iteratively<true>(lhs, rhs, false)), true)
     }
 
     /// @brief comparison: 3-way
@@ -28828,7 +29324,8 @@ class basic_json // NOLINT(cppcoreguidelines-special-member-functions,hicpp-spec
         JSON_HEDLEY_DIAGNOSTIC_PUSH
         JSON_HEDLEY_PRAGMA(GCC diagnostic ignored "-Wfloat-equal")
 #endif
-        JSON_IMPLEMENT_OPERATOR( ==, true, false, false)
+        JSON_IMPLEMENT_OPERATOR( ==, true, false, false,
+                                 compare_iteratively<false>(lhs, rhs, false) == compare_result::equal, true)
 #ifdef __GNUC__
         JSON_HEDLEY_DIAGNOSTIC_POP
 #endif
@@ -28884,7 +29381,8 @@ class basic_json // NOLINT(cppcoreguidelines-special-member-functions,hicpp-spec
         // default_result is used if we cannot compare values. In that case,
         // we compare types. Note we have to call the operator explicitly,
         // because MSVC has problems otherwise.
-        JSON_IMPLEMENT_OPERATOR( <, false, false, operator<(lhs_type, rhs_type))
+        JSON_IMPLEMENT_OPERATOR( <, false, false, operator<(lhs_type, rhs_type),
+                                 compare_iteratively<true>(lhs, rhs, true) == compare_result::less, false)
     }
 
     /// @brief comparison: less than
@@ -30747,6 +31245,7 @@ struct formatter<nlohmann::NLOHMANN_BASIC_JSON_TPL, char> // NOLINT(cert-dcl58-c
     #undef JSON_HAS_STATIC_RTTI
     #undef JSON_USE_LEGACY_DISCARDED_VALUE_COMPARISON
     #undef JSON_BRACE_INIT_COPY_SEMANTICS
+    #undef JSON_PRECISE_STREAM_POSITION
 #endif
 
 // #include <nlohmann/thirdparty/hedley/hedley_undef.hpp>

single_include/nlohmann/json_fwd.hpp

@@ -56,6 +56,10 @@
     #define JSON_BRACE_INIT_COPY_SEMANTICS 0
 #endif
 
+#ifndef JSON_PRECISE_STREAM_POSITION
+    #define JSON_PRECISE_STREAM_POSITION 0
+#endif
+
 #if JSON_DIAGNOSTICS
     #define NLOHMANN_JSON_ABI_TAG_DIAGNOSTICS _diag
 #else
@@ -80,21 +84,28 @@
     #define NLOHMANN_JSON_ABI_TAG_BRACE_INIT_COPY_SEMANTICS
 #endif
 
+#if JSON_PRECISE_STREAM_POSITION
+    #define NLOHMANN_JSON_ABI_TAG_PRECISE_STREAM_POSITION _psp
+#else
+    #define NLOHMANN_JSON_ABI_TAG_PRECISE_STREAM_POSITION
+#endif
+
 #ifndef NLOHMANN_JSON_NAMESPACE_NO_VERSION
     #define NLOHMANN_JSON_NAMESPACE_NO_VERSION 0
 #endif
 
 // Construct the namespace ABI tags component
-#define NLOHMANN_JSON_ABI_TAGS_CONCAT_EX(a, b, c, d) json_abi ## a ## b ## c ## d
-#define NLOHMANN_JSON_ABI_TAGS_CONCAT(a, b, c, d) \
-    NLOHMANN_JSON_ABI_TAGS_CONCAT_EX(a, b, c, d)
+#define NLOHMANN_JSON_ABI_TAGS_CONCAT_EX(a, b, c, d, e) json_abi ## a ## b ## c ## d ## e
+#define NLOHMANN_JSON_ABI_TAGS_CONCAT(a, b, c, d, e) \
+    NLOHMANN_JSON_ABI_TAGS_CONCAT_EX(a, b, c, d, e)
 
 #define NLOHMANN_JSON_ABI_TAGS                                       \
     NLOHMANN_JSON_ABI_TAGS_CONCAT(                                   \
             NLOHMANN_JSON_ABI_TAG_DIAGNOSTICS,                       \
             NLOHMANN_JSON_ABI_TAG_LEGACY_DISCARDED_VALUE_COMPARISON, \
             NLOHMANN_JSON_ABI_TAG_DIAGNOSTIC_POSITIONS,              \
-            NLOHMANN_JSON_ABI_TAG_BRACE_INIT_COPY_SEMANTICS)
+            NLOHMANN_JSON_ABI_TAG_BRACE_INIT_COPY_SEMANTICS,         \
+            NLOHMANN_JSON_ABI_TAG_PRECISE_STREAM_POSITION)
 
 // Construct the namespace version component
 #define NLOHMANN_JSON_NAMESPACE_VERSION_CONCAT_EX(major, minor, patch) \

tests/CMakeLists.txt

@@ -80,7 +80,10 @@ target_compile_options(test_main PUBLIC
     #                              std::allocator_traits<...>::construct for the custom-base-class
     #                              test's map type past VS2015's limit. The name is only used for
     #                              debug info, so truncation does not affect the build.
-    $<$<CXX_COMPILER_ID:MSVC>:/W4;/wd4566;/wd4996;/wd4702;/wd4503>
+    #       Disable warning C5285: cannot declare a specialization for 'std::tuple'; MSVC 19.51
+    #                              reports the forward declarations of standard library
+    #                              templates in the vendored doctest.h
+    $<$<CXX_COMPILER_ID:MSVC>:/W4;/wd4566;/wd4996;/wd4702;/wd4503;/wd5285>
     # https://github.com/nlohmann/json/issues/1114
     $<$<CXX_COMPILER_ID:MSVC>:/bigobj> $<$<BOOL:${MINGW}>:-Wa,-mbig-obj>
 

tests/abi/config/default.cpp

@@ -36,6 +36,10 @@ TEST_CASE("default namespace")
         expected += "_bics";
 #endif
 
+#if JSON_PRECISE_STREAM_POSITION
+        expected += "_psp";
+#endif
+
         expected += "_v" STRINGIZE(NLOHMANN_JSON_VERSION_MAJOR);
         expected += "_" STRINGIZE(NLOHMANN_JSON_VERSION_MINOR);
         expected += "_" STRINGIZE(NLOHMANN_JSON_VERSION_PATCH) "::basic_json";

tests/abi/config/noversion.cpp

@@ -37,6 +37,10 @@ TEST_CASE("default namespace without version component")
         expected += "_bics";
 #endif
 
+#if JSON_PRECISE_STREAM_POSITION
+        expected += "_psp";
+#endif
+
         expected += "::basic_json";
 
         // fallback for Clang

tests/fuzzing.md

@@ -79,3 +79,26 @@ the same `fuzzers` target as above and also relies on the `FUZZER_ENGINE` variab
 [build script](https://github.com/google/oss-fuzz/blob/master/projects/json/build.sh) for more information.
 
 In case the build at OSS-Fuzz fails, an issue will be created automatically.
+
+### Handling OSS-Fuzz reports
+
+OSS-Fuzz files the crashes it finds in its own [issue tracker](https://issues.oss-fuzz.com), not on GitHub. So that
+each report can be traced to the change that fixed it, and each fix to the report it answers, fixes follow these
+conventions:
+
+- **Reference the OSS-Fuzz issue in the pull request**, next to any GitHub issue it closes, as `OSS-Fuzz: <id>` (for
+  example, `OSS-Fuzz: 563659413`), and in the commit message. The ID alone does not disclose the crash. If the report
+  was triaged into a GitHub issue, link the OSS-Fuzz issue there too.
+- **Turn the reproducer into a unit test.** Download the testcase from the OSS-Fuzz report, reduce it if possible, and
+  add it as a regression test to the unit test of the affected format (e.g., `tests/src/unit-bjdata.cpp`), with a
+  comment naming the OSS-Fuzz issue. This way the input is checked by every CI run rather than only by OSS-Fuzz, and
+  it stays covered even if OSS-Fuzz later closes the report as not reproducible.
+- **Keep the fuzzer drivers and the unit tests in sync.** The round-trip checks of the UBJSON and BJData drivers are
+  also run on a fixed corpus in the unit tests (see `tests/src/round_trip_corpus.hpp` and the "round-trip invariants"
+  test cases), so a regression shows up in CI first. When a driver's checks change, change the unit tests with them.
+- **Record in the report whether the bug shipped.** OSS-Fuzz asks whether a crash was a short-lived regression or
+  affects a released version; answer it when the fix is merged, as it decides whether the fix needs a release note or
+  a security advisory (see the [security policy](../.github/SECURITY.md)).
+
+After the fix is merged, OSS-Fuzz re-runs the reproducer on its next build and marks the report as verified and
+closed. If it does not, the fix is incomplete.

tests/src/fuzzer-parse_bjdata.cpp

@@ -42,6 +42,9 @@ dump() serializes any non-finite double the same deterministic way (as JSON
 `null`, since JSON itself cannot represent NaN/Infinity), so comparing
 dumps is stable under exactly the same values that break operator==.
 
+The unit tests run the same checks on a fixed corpus (see the "BJData round-trip
+invariants" test case), so keep both in sync.
+
 The provided function `LLVMFuzzerTestOneInput` can be used in different fuzzer
 drivers.
 */

tests/src/fuzzer-parse_ubjson.cpp

@@ -21,6 +21,9 @@ array data, it performs the following steps:
 - j4 = from_ubjson(vec3)
 - assert(j1 == j4)
 
+The unit tests run the same checks on a fixed corpus (see the "UBJSON round-trip
+invariants" test case), so keep both in sync.
+
 The provided function `LLVMFuzzerTestOneInput` can be used in different fuzzer
 drivers.
 */

tests/src/round_trip_corpus.hpp

@@ -0,0 +1,213 @@
+//     __ _____ _____ _____
+//  __|  |   __|     |   | |  JSON for Modern C++ (supporting code)
+// |  |  |__   |  |  | | | |  version 3.12.0
+// |_____|_____|_____|_|___|  https://github.com/nlohmann/json
+//
+// SPDX-FileCopyrightText: 2013-2026 Niels Lohmann <https://nlohmann.me>
+// SPDX-License-Identifier: MIT
+
+#pragma once
+
+#include <cmath> // nan
+#include <cstddef> // size_t
+#include <cstdint> // int32_t, int64_t, uint32_t, uint64_t
+#include <limits> // numeric_limits
+#include <random> // mt19937
+#include <string> // string, to_string
+#include <utility> // move
+#include <vector> // vector
+
+#include <nlohmann/json.hpp>
+
+// Values for the round-trip property tests of the UBJSON and BJData writers.
+//
+// The fuzzer drivers (tests/src/fuzzer-parse_ubjson.cpp and
+// fuzzer-parse_bjdata.cpp) check that anything the library parses can be
+// serialized, parsed back, and serialized again without loss. Those checks
+// only run at OSS-Fuzz, so a regression used to surface days later as an
+// external report. The unit tests run the same checks on this corpus in CI.
+//
+// The corpus is deterministic: std::mt19937's output sequence is fixed by
+// the standard, and it is used directly rather than through a distribution
+// (whose results are implementation-defined).
+namespace utils
+{
+
+class round_trip_corpus
+{
+  public:
+    using json = nlohmann::json;
+
+    static std::vector<json> values()
+    {
+        round_trip_corpus corpus;
+        return corpus.build();
+    }
+
+    // whether a value contains a binary value, which a BJData or UBJSON round
+    // trip may turn into an array of integers
+    static bool contains_binary(const json& j)
+    {
+        if (j.is_binary())
+        {
+            return true;
+        }
+        if (j.is_structured())
+        {
+            for (const auto& element : j)
+            {
+                if (contains_binary(element))
+                {
+                    return true;
+                }
+            }
+        }
+        return false;
+    }
+
+  private:
+    std::vector<json> atoms;
+    // a fixed seed is the point: the corpus must be the same in every run
+    std::mt19937 generator{42}; // NOLINT(cert-msc32-c,cert-msc51-cpp,bugprone-random-generator-seed)
+
+    round_trip_corpus()
+        : atoms
+    {
+        nullptr, true, false,
+        // integers at the boundaries of every UBJSON/BJData integer type
+        0, 1, -1, 127, 128, 255, 256, -128, -129,
+        32767, 32768, 65535, 65536, -32768, -32769,
+        (std::numeric_limits<std::int32_t>::min)(), (std::numeric_limits<std::int32_t>::max)(),
+        (std::numeric_limits<std::uint32_t>::max)(),
+        (std::numeric_limits<std::int64_t>::min)(), (std::numeric_limits<std::int64_t>::max)(),
+        static_cast<std::uint64_t>((std::numeric_limits<std::int64_t>::max)()) + 1u,
+        (std::numeric_limits<std::uint64_t>::max)(),
+        // floating-point numbers, including non-finite ones
+        0.0, -0.0, 1.5, -2.25, 3.4e38, (std::numeric_limits<double>::max)(),
+        std::nan(""), std::numeric_limits<double>::infinity(), -std::numeric_limits<double>::infinity(),
+        // strings, including a non-ASCII one and one longer than 255 bytes
+        "", "a", "\xC3\xA4", std::string(300, 'x'),
+        // binary values with and without subtype
+        json::binary({}), json::binary({1, 2, 255}), json::binary({0x80, 0x7F}, 42), json::binary({1}, 0)
+    }
+    {}
+
+    std::vector<json> build()
+    {
+        std::vector<json> result = atoms;
+
+        // each atom inside containers, including homogeneous ones that the
+        // writers encode as optimized (typed) containers
+        result.emplace_back(json::array());
+        result.emplace_back(json::object());
+        for (const auto& atom : atoms)
+        {
+            result.push_back(json::array({atom}));
+            result.push_back(json::array({atom, atom, atom}));
+            result.push_back(json::array({json::array({atom})}));
+            result.push_back(json::object({{"key", atom}}));
+        }
+        result.push_back(json::array({1, 1.5}));
+        result.push_back(json::array({-1, 255}));
+        result.push_back(json::array({"a", "b"}));
+
+        // deep, but well below any recursion or depth limit
+        json nested_array = 1;
+        json nested_object = 1;
+        for (int i = 0; i < 300; ++i)
+        {
+            nested_array = json::array({nested_array});
+            nested_object = json::object({{"key", nested_object}});
+        }
+        result.push_back(nested_array);
+        result.push_back(nested_object);
+
+        add_annotated_arrays(result);
+        add_random_values(result);
+        return result;
+    }
+
+    // objects in the JData annotated array format, which the BJData writer
+    // encodes as ND-arrays when the annotation describes a packed array, and
+    // as plain objects otherwise (see #5398, #5399, #5403, #5404, and #5542)
+    static void add_annotated_arrays(std::vector<json>& result)
+    {
+        const std::vector<json> types =
+        {
+            "uint8", "int8", "uint16", "int16", "uint32", "int32", "uint64", "int64",
+            "single", "double", "char", "byte", "bool", "unknown", 5, nullptr
+        };
+        const std::vector<json> sizes =
+        {
+            json::array(), {3}, {1, 3}, {3, 1}, {2, 3}, {2, 0}, {0, 2}, {2, 2, 2}, {-1, 2}, {2, 1.5},
+            "3", 3, nullptr, json::binary({})
+        };
+        const std::vector<json> data =
+        {
+            nullptr, 5, "s", json::object({{"a", 1}}), json::array(),
+            {1, 2, 3}, {1, 2, 3, 4, 5, 6}, {1, 2, 3, 4, 5, 6, 7, 8},
+            {1.5, 2.5, 3.5, 4.5, 5.5, 6.5}, {300, -300, 70000, -70000, 1, 2},
+            {"a", "b", "c", "d", "e", "f"}, {json::array({1, 2, 3}), json::array({4, 5, 6})}
+        };
+
+        for (const auto& type : types)
+        {
+            for (const auto& size : sizes)
+            {
+                for (const auto& d : data)
+                {
+                    result.push_back({{"_ArrayType_", type}, {"_ArraySize_", size}, {"_ArrayData_", d}});
+                }
+            }
+        }
+
+        // incomplete annotations and annotations with an extra key
+        result.push_back({{"_ArraySize_", {2, 3}}, {"_ArrayData_", {1, 2, 3, 4, 5, 6}}});
+        result.push_back({{"_ArrayType_", "uint8"}, {"_ArrayData_", {1, 2, 3, 4, 5, 6}}});
+        result.push_back({{"_ArrayType_", "uint8"}, {"_ArraySize_", {2, 3}}});
+        result.push_back({{"_ArrayType_", "uint8"}, {"_ArraySize_", {2, 3}}, {"_ArrayData_", {1, 2, 3, 4, 5, 6}}, {"extra", 1}});
+    }
+
+    // random containers of atoms, both homogeneous and mixed
+    void add_random_values(std::vector<json>& result)
+    {
+        for (int i = 0; i < 1000; ++i)
+        {
+            result.push_back(random_value(0));
+        }
+    }
+
+    std::size_t random_below(std::size_t bound)
+    {
+        return generator() % bound;
+    }
+
+    json random_value(int depth)
+    {
+        const auto kind = random_below(10);
+        if (depth > 3 || kind < 5)
+        {
+            return atoms[random_below(atoms.size())];
+        }
+
+        json result = kind < 8 ? json::array() : json::object();
+        const auto count = random_below(5);
+        const bool homogeneous = random_below(2) == 0;
+        const json fixed = atoms[random_below(atoms.size())];
+        for (std::size_t i = 0; i < count; ++i)
+        {
+            json element = homogeneous ? fixed : random_value(depth + 1);
+            if (result.is_array())
+            {
+                result.push_back(std::move(element));
+            }
+            else
+            {
+                result[std::to_string(i)] = std::move(element);
+            }
+        }
+        return result;
+    }
+};
+
+} // namespace utils

tests/src/unit-allocator.cpp

@@ -270,6 +270,82 @@ TEST_CASE("controlled bad_alloc")
     }
 }
 
+namespace
+{
+// counts the allocations of pairs with a non-const first member: the object
+// types store std::pair<const Key, T>, so only the scratch space of the
+// iterative deep copy allocates std::pair<Key, T>
+std::size_t scratch_pair_allocations = 0;
+
+template<class T>
+struct is_scratch_pair : std::false_type {};
+
+template<class K, class V>
+struct is_scratch_pair<std::pair<K, V>> : std::integral_constant < bool, !std::is_const<K>::value > {};
+
+template<class T>
+struct scratch_counting_allocator : std::allocator<T>
+{
+    using std::allocator<T>::allocator;
+
+    T* allocate(std::size_t n)
+    {
+        if (is_scratch_pair<T>::value)
+        {
+            ++scratch_pair_allocations;
+        }
+        return std::allocator<T>::allocate(n);
+    }
+
+#ifdef __cpp_lib_allocate_at_least
+    // std::allocator<T>::allocate_at_least would bypass the counting, and
+    // libc++'s containers prefer it over allocate from C++23 on
+    auto allocate_at_least(std::size_t n)
+    {
+        if (is_scratch_pair<T>::value)
+        {
+            ++scratch_pair_allocations;
+        }
+        return std::allocator<T>::allocate_at_least(n);
+    }
+#endif
+
+    template <class U>
+    struct rebind
+    {
+        using other = scratch_counting_allocator<U>;
+    };
+};
+} // namespace
+
+TEST_CASE("deep copy uses the provided allocator")
+{
+    using counting_json = nlohmann::basic_json<std::map,
+          std::vector,
+          std::string,
+          bool,
+          std::int64_t,
+          std::uint64_t,
+          double,
+          scratch_counting_allocator>;
+
+    // deeper than the 128 levels the copy constructor descends into, so the
+    // innermost objects are copied by the iterative deep copy
+    counting_json j = 1;
+    for (std::size_t i = 0; i < 300; ++i)
+    {
+        counting_json wrapper = counting_json::object();
+        wrapper["a"] = std::move(j);
+        j = std::move(wrapper);
+    }
+
+    scratch_pair_allocations = 0;
+    // NOLINTNEXTLINE(performance-unnecessary-copy-initialization): the copy is what is tested
+    const counting_json copy(j);
+    CHECK(scratch_pair_allocations > 0);
+    CHECK(copy == j);
+}
+
 namespace
 {
 template<class T>

tests/src/unit-bjdata.cpp

@@ -19,6 +19,7 @@ using nlohmann::json;
 #include <fstream>
 #include <set>
 #include "make_test_data_available.hpp"
+#include "round_trip_corpus.hpp"
 #include "test_utils.hpp"
 
 namespace
@@ -2867,6 +2868,21 @@ TEST_CASE("BJData")
                 const auto out_num = json::to_bjdata(j_num);
                 CHECK(out_num.at(0) == '{');
                 CHECK(json::from_bjdata(out_num) == j_num);
+
+                // OSS-Fuzz issue 474400817: an empty object _ArraySize_ was
+                // written as the ND-array header length, which from_bjdata()
+                // could not read back
+                const std::vector<uint8_t> input =
+                {
+                    '[', '{', 'U', 11, '_', 'A', 'r', 'r', 'a', 'y', 'D', 'a', 't', 'a', '_', 'Z',
+                    'U', 11, '_', 'A', 'r', 'r', 'a', 'y', 'T', 'y', 'p', 'e', '_', 'S', 'i', 5, 'i', 'n', 't', '1', '6',
+                    'U', 11, '_', 'A', 'r', 'r', 'a', 'y', 'S', 'i', 'z', 'e', '_', '{', '}', '}', ']'
+                };
+                const json j1 = json::from_bjdata(input);
+                CHECK(j1 == json::parse(R"([{"_ArrayType_":"int16","_ArraySize_":{},"_ArrayData_":null}])"));
+                json j2;
+                CHECK_NOTHROW(j2 = json::from_bjdata(json::to_bjdata(j1, false, false)));
+                CHECK(j2 == j1);
             }
 
             SECTION("ndarray with out-of-range _ArrayData_ elements stays as object")
@@ -4273,6 +4289,93 @@ TEST_CASE("BJData use_type requires use_size")
     }
 }
 
+TEST_CASE("BJData round-trip invariants")
+{
+    // This checks what the parse_bjdata_fuzzer driver checks (see
+    // tests/src/fuzzer-parse_bjdata.cpp), so that a regression shows up in CI
+    // rather than as an OSS-Fuzz report: every value from_bjdata() returns
+    // (j1) can be serialized with any combination of options, the result can
+    // be parsed back (j2), and serializing j2 again with the same options
+    // yields a value-equal result.
+    //
+    // Beyond the driver, this also checks that j2 equals j1 and that
+    // serializing j2 reproduces the exact bytes, both except for values that
+    // contain a binary value: a binary value is only written as a binary
+    // value with Draft 3's optimized binary array, and otherwise read back as
+    // an array of integers, for which the writer may choose different (but
+    // equally valid) type markers when it is serialized again (see #5494).
+    //
+    // Values are compared with dump() rather than operator==, because a NaN
+    // never compares equal to itself.
+    struct options
+    {
+        bool use_size;
+        bool use_type;
+        json::bjdata_version_t version;
+    };
+    const std::vector<options> all_options =
+    {
+        {false, false, json::bjdata_version_t::draft2},
+        {true, false, json::bjdata_version_t::draft2},
+        {true, true, json::bjdata_version_t::draft2},
+        {false, false, json::bjdata_version_t::draft3},
+        {true, false, json::bjdata_version_t::draft3},
+        {true, true, json::bjdata_version_t::draft3},
+    };
+
+    for (const auto& j0 : utils::round_trip_corpus::values())
+    {
+        // turn the corpus value into a value as from_bjdata() returns it
+        for (const auto& initial : all_options)
+        {
+            const json j1 = json::from_bjdata(json::to_bjdata(j0, initial.use_size, initial.use_type, initial.version));
+            const bool has_binary = utils::round_trip_corpus::contains_binary(j1);
+
+            for (const auto& o : all_options)
+            {
+                INFO("j1 = " << j1.dump() << ", use_size = " << o.use_size << ", use_type = " << o.use_type
+                     << ", draft3 = " << (o.version == json::bjdata_version_t::draft3));
+
+                const std::vector<std::uint8_t> vec = json::to_bjdata(j1, o.use_size, o.use_type, o.version);
+                json j2;
+                // anything the library writes must be parsable by the library
+                REQUIRE_NOTHROW(j2 = json::from_bjdata(vec));
+                const std::vector<std::uint8_t> vec2 = json::to_bjdata(j2, o.use_size, o.use_type, o.version);
+                CHECK(json::from_bjdata(vec2).dump() == j2.dump());
+
+                if (!has_binary)
+                {
+                    CHECK(j2.dump() == j1.dump());
+                    CHECK(vec2 == vec);
+                }
+            }
+        }
+    }
+}
+
+TEST_CASE("BJData round trip of a binary value is value-stable, not byte-stable")
+{
+    // OSS-Fuzz issue 474480402: a Draft 3 optimized binary array is read as a
+    // binary value, which to_bjdata() writes in the default Draft 2 mode as a
+    // plain array of uint8 numbers. That is read back as an array of numbers,
+    // for which the writer then picks the smallest type marker, int8 ('i'),
+    // so re-serializing changes the bytes, but not the value. This is the
+    // exception described in the "Round trips" note of the BJData
+    // documentation, and why the fuzzer checks value stability (see #5494).
+    const std::vector<uint8_t> input = {'[', '$', 'B', '#', 'U', 1, 0x20};
+    const json j1 = json::from_bjdata(input);
+    CHECK(j1 == json::binary({0x20}));
+
+    const std::vector<uint8_t> vec = json::to_bjdata(j1, false, false);
+    CHECK(vec == std::vector<uint8_t>({'[', 'U', 0x20, ']'}));
+    const json j2 = json::from_bjdata(vec);
+    CHECK(j2 == json::array({0x20}));
+
+    const std::vector<uint8_t> vec2 = json::to_bjdata(j2, false, false);
+    CHECK(vec2 == std::vector<uint8_t>({'[', 'i', 0x20, ']'}));
+    CHECK(json::from_bjdata(vec2) == j2);
+}
+
 TEST_CASE("BJData roundtrips" * doctest::skip())
 {
     SECTION("input from self-generated BJData files")

tests/src/unit-deserialization.cpp

@@ -25,6 +25,7 @@ using nlohmann::json;
 #include <iostream>
 #include <iterator>
 #include <sstream>
+#include <string>
 #include <valarray>
 
 #if defined(_WIN32)
@@ -1233,6 +1234,57 @@ TEST_CASE("deserialization")
         }
     }
 
+    SECTION("stream position after extraction without JSON_PRECISE_STREAM_POSITION (#5340)")
+    {
+        // By default, the character that terminates a number is consumed, so
+        // the stream is left one byte too far after a number (and only after a
+        // number). JSON_PRECISE_STREAM_POSITION changes this; see
+        // unit-precise-stream-position.cpp. These checks pin the default.
+        const auto remaining = [](std::istream & is)
+        {
+            return std::string(std::istreambuf_iterator<char>(is), std::istreambuf_iterator<char>());
+        };
+
+        SECTION("the character after a number is consumed")
+        {
+            std::istringstream ss("1true");
+            json j;
+            ss >> j;
+            CHECK(j == 1);
+            CHECK(remaining(ss) == "rue");
+        }
+
+        SECTION("the character after other values is not consumed")
+        {
+            std::istringstream ss("[1]true");
+            json j;
+            ss >> j;
+            CHECK(j == json::parse("[1]"));
+            CHECK(remaining(ss) == "true");
+        }
+
+        SECTION("comma-separated numbers can be read one by one")
+        {
+            std::istringstream ss("1,2,3");
+            json j1, j2, j3;
+            ss >> j1 >> j2 >> j3;
+            CHECK(j1 == 1);
+            CHECK(j2 == 2);
+            CHECK(j3 == 3);
+        }
+
+        SECTION("std::getline after a number skips the line break")
+        {
+            std::istringstream ss("42\nfoo");
+            json j;
+            std::string line;
+            ss >> j;
+            std::getline(ss, line);
+            CHECK(j == 42);
+            CHECK(line == "foo");
+        }
+    }
+
     // build with C++20
     // JSON_HAS_CPP_20
 #if defined(__cpp_char8_t)

tests/src/unit-diagnostics.cpp

@@ -361,6 +361,106 @@ TEST_CASE("Regression tests for extended diagnostics")
             CHECK(p == o);
         }
     }
+
+    SECTION("Regression test - erase() and update() must keep JSON_DIAGNOSTICS parent pointers of ordered_json members")
+    {
+        // ordered_json keeps its members in a vector: erasing a member
+        // re-constructs all members after it in place, and adding a key may
+        // reallocate the vector; both reset the parent pointers of the members
+        // that were moved
+        using nlohmann::ordered_json;
+
+        const auto check_parents = [](const ordered_json & j)
+        {
+            // const access, so operator[] cannot repair the parent pointers
+            CHECK_THROWS_WITH_AS(j["z"]["x"].at(0), "[json.exception.type_error.304] (/z/x) cannot use at() with number", ordered_json::type_error);
+
+            // must not trigger assert_invariant() in a debug/assert-enabled build
+            ordered_json const copy = j; // NOLINT(performance-unnecessary-copy-initialization)
+            CHECK(copy == j);
+        };
+
+        // erase(key)
+        {
+            ordered_json j = {{"a", 1}, {"z", {{"x", 1}}}};
+            CHECK(j.erase("a") == 1);
+            check_parents(j);
+        }
+
+        // erase(iterator)
+        {
+            ordered_json j = {{"a", 1}, {"z", {{"x", 1}}}};
+            j.erase(j.begin());
+            check_parents(j);
+        }
+
+        // erase(iterator, iterator)
+        {
+            ordered_json j = {{"a", 1}, {"b", 2}, {"z", {{"x", 1}}}};
+            j.erase(j.begin(), j.find("z"));
+            check_parents(j);
+        }
+
+        // patch() removes via erase(iterator)
+        {
+            ordered_json j = {{"a", 1}, {"z", {{"x", 1}}}};
+            j.patch_inplace(ordered_json::parse(R"([{"op": "remove", "path": "/a"}])"));
+            check_parents(j);
+        }
+
+        // update(j)
+        {
+            ordered_json j = {{"z", {{"x", 1}}}};
+            j.update({{"a", 1}, {"b", 2}});
+            check_parents(j);
+        }
+
+        // update(j, true), the outer and the nested vector both grow
+        {
+            ordered_json j = {{"z", {{"x", 1}}}};
+            j.update({{"z", {{"y", 2}}}, {"a", 1}}, true);
+            check_parents(j);
+        }
+
+        // update(j, true) around its descent bound, where the nested vectors
+        // grow while the objects are merged without recursing
+        for (const std::size_t depth :
+                {
+                    nlohmann::detail::recursion_depth_limit() - 1, nlohmann::detail::recursion_depth_limit(), nlohmann::detail::recursion_depth_limit() + 2
+                })
+        {
+            ordered_json j = {{"z", {{"x", 1}}}};
+            ordered_json patch = {{"a", 1}, {"b", 2}, {"c", {{"d", 3}}}};
+            for (std::size_t i = 0; i < depth; ++i)
+            {
+                j = ordered_json{{"k", 0}, {"n", std::move(j)}};
+                patch = ordered_json{{"n", std::move(patch)}, {"l", 1}, {"m", 2}};
+            }
+            j.update(patch, true);
+
+            // must not trigger assert_invariant() on any level in a
+            // debug/assert-enabled build
+            ordered_json const copy = j; // NOLINT(performance-unnecessary-copy-initialization)
+            CHECK(copy == j);
+        }
+
+        // merge_patch() inserts "c" and removes "d" at /a/c, then inserts "e"
+        // at /a, which copies /a/c
+        {
+            auto j = ordered_json::parse(R"({"a": {"c": {"d": {}}}})");
+            j.merge_patch(ordered_json::parse(R"({"a": {"c": {"c": "s", "d": null}, "e": "s"}})"));
+            CHECK(j.dump() == R"({"a":{"c":{"c":"s"},"e":"s"}})");
+
+            auto const& constJ = j;
+#if JSON_DIAGNOSTIC_POSITIONS
+            CHECK_THROWS_WITH_AS(constJ["a"]["c"]["c"].at(0), "[json.exception.type_error.304] (/a/c/c) (bytes 18-21) cannot use at() with string", ordered_json::type_error);
+#else
+            CHECK_THROWS_WITH_AS(constJ["a"]["c"]["c"].at(0), "[json.exception.type_error.304] (/a/c/c) cannot use at() with string", ordered_json::type_error);
+#endif
+            ordered_json const copy = j;
+            CHECK(copy == j);
+        }
+    }
 }
 
 TEST_CASE("Better diagnostics past the descent bound of update() and merge_patch()")

tests/src/unit-large_json.cpp

@@ -157,6 +157,54 @@ TEST_CASE("tests on deeply nested JSONs")
             CHECK(deep_depth == depth);
         }
 
+        SECTION("comparing")
+        {
+            // Comparing used to descend once per level, and an ordered
+            // comparison used to compare every pair of elements twice, once in
+            // each direction, which took exponentially long in the nesting
+            // depth. Both are gone: these finish in milliseconds, where the
+            // second used to take longer than anyone would wait even for a
+            // value nested only a few dozen levels deep.
+            const std::string text = std::string(depth, '[') + '0' + std::string(depth, ']');
+            const json j = json::parse(text);
+            const json same = json::parse(text);
+            const json larger = json::parse(std::string(depth, '[') + '1' + std::string(depth, ']'));
+
+            CHECK(j == same);
+            CHECK_FALSE(j == larger);
+            CHECK(j != larger);
+
+            CHECK(j < larger);
+            CHECK_FALSE(larger < j);
+            CHECK(larger > j);
+            CHECK(j <= same);
+            CHECK(j >= same);
+
+            // a value that ends earlier is the smaller one
+            const json shorter = json::parse(std::string(depth - 1, '[') + '0' + std::string(depth - 1, ']'));
+            CHECK_FALSE(j == shorter);
+        }
+
+        SECTION("comparing objects")
+        {
+            std::string text;
+            text.reserve((6 * depth) + 1);
+            for (std::size_t i = 0; i < depth; ++i)
+            {
+                text += "{\"a\":";
+            }
+            text += '1';
+            text.append(depth, '}');
+
+            const json j = json::parse(text);
+            const json same = json::parse(text);
+
+            CHECK(j == same);
+            CHECK_FALSE(j != same);
+            CHECK(j <= same);
+            CHECK(j >= same);
+        }
+
         SECTION("the copy is independent of the original")
         {
             const json j = json::parse(std::string(depth, '[') + '0' + std::string(depth, ']'));

tests/src/unit-merge_patch.cpp

@@ -62,7 +62,7 @@ std::string nested_objects(const std::size_t depth, const int variant)
         }
         text += "\"a\":";
     }
-    text += variant == 1 ? "{\"x\":1,\"y\":null}" : "{\"y\":2}";
+    text += variant == 1 ? R"({"x":1,"y":null})" : "{\"y\":2}";
     text.append(depth, '}');
     return text;
 }

tests/src/unit-modifiers.cpp

@@ -48,7 +48,7 @@ std::string nested_objects(const std::size_t depth, const int variant)
         if (variant == 2 && i % 5 == 0)
         {
             // an object replacing a primitive, which is not merged
-            text += "\"s0\":{\"o\":1},";
+            text += R"("s0":{"o":1},)";
         }
         text += "\"a\":";
     }

tests/src/unit-precise-stream-position.cpp

@@ -0,0 +1,237 @@
+//     __ _____ _____ _____
+//  __|  |   __|     |   | |  JSON for Modern C++ (supporting code)
+// |  |  |__   |  |  | | | |  version 3.12.0
+// |_____|_____|_____|_|___|  https://github.com/nlohmann/json
+//
+// SPDX-FileCopyrightText: 2013-2026 Niels Lohmann <https://nlohmann.me>
+// SPDX-License-Identifier: MIT
+
+#include "doctest_compatibility.h"
+
+// This file tests the opt-in JSON_PRECISE_STREAM_POSITION, so it defines the
+// macro itself rather than relying on a -D flag, and runs in every build. The
+// default behavior is pinned in unit-deserialization.cpp.
+#ifdef JSON_PRECISE_STREAM_POSITION
+    #undef JSON_PRECISE_STREAM_POSITION
+#endif
+
+#define JSON_PRECISE_STREAM_POSITION 1
+
+#include <nlohmann/json.hpp>
+using nlohmann::json;
+
+#include <cstddef>
+#include <sstream>
+#include <streambuf>
+#include <string>
+#include <utility>
+#include <vector>
+
+#define STRINGIZE_EX(x) #x
+#define STRINGIZE(x) STRINGIZE_EX(x)
+
+namespace
+{
+// A streambuf that keeps no get area at all and refuses every putback: with an
+// empty get area, sungetc() always ends up in pbackfail(). Used to check that
+// the character terminating a number is left in the input without relying on
+// the streambuf being able to put a consumed character back.
+class no_putback_streambuf : public std::streambuf
+{
+  public:
+    explicit no_putback_streambuf(std::string s) : m_data(std::move(s)) {}
+
+  protected:
+    // peek at the next character without consuming it
+    int_type underflow() override
+    {
+        if (m_pos >= m_data.size())
+        {
+            return traits_type::eof();
+        }
+        return traits_type::to_int_type(m_data[m_pos]);
+    }
+
+    // consume the next character
+    int_type uflow() override
+    {
+        if (m_pos >= m_data.size())
+        {
+            return traits_type::eof();
+        }
+        return traits_type::to_int_type(m_data[m_pos++]);
+    }
+
+    int_type pbackfail(int_type /*c*/) override
+    {
+        return traits_type::eof();
+    }
+
+  private:
+    std::string m_data;
+    std::size_t m_pos = 0;
+};
+
+// read the characters that are left in a stream
+std::string remaining(std::istream& is)
+{
+    std::string result;
+    char c = 0;
+    while (is.get(c))
+    {
+        result += c;
+    }
+    return result;
+}
+}  // namespace
+
+TEST_CASE("JSON_PRECISE_STREAM_POSITION")
+{
+    SECTION("the macro is part of the ABI tag")
+    {
+        const std::string ns = STRINGIZE(NLOHMANN_JSON_NAMESPACE);
+        // other tags may come before it, e.g. json_abi_diag_psp
+        CHECK(ns.find("_psp") != std::string::npos);
+    }
+
+    SECTION("a number does not consume the character that terminates it")
+    {
+        // a number is only terminated by the character following it; that
+        // character must be given back so the stream is positioned right
+        // after the value
+        const std::vector<std::pair<std::string, std::string>> tests =
+        {
+            {"1true", "true"},
+            {"1[2]", "[2]"},
+            {"1{}", "{}"},
+            {R"(1"a")", R"("a")"},
+            {"1 true", " true"},
+            {"12,", ","},
+            {"-0.5e3x", "x"},
+            {"1null", "null"}
+        };
+
+        for (const auto& test : tests)
+        {
+            CAPTURE(test.first);
+            std::istringstream ss(test.first);
+            json j;
+            ss >> j;
+            CHECK(j == json::parse(test.first.substr(0, test.first.size() - test.second.size())));
+            CHECK(remaining(ss) == test.second);
+        }
+    }
+
+    SECTION("values that are self-delimiting are unaffected")
+    {
+        const std::vector<std::pair<std::string, std::string>> tests =
+        {
+            {"truefalse", "false"},
+            {"[1][2]", "[2]"},
+            {R"({"a":1}{"b":2})", R"({"b":2})"},
+            {R"("a""b")", R"("b")"},
+            {"null null", " null"}
+        };
+
+        for (const auto& test : tests)
+        {
+            CAPTURE(test.first);
+            std::istringstream ss(test.first);
+            json j;
+            ss >> j;
+            CHECK(remaining(ss) == test.second);
+        }
+    }
+
+    SECTION("a number at the end of the input leaves nothing behind")
+    {
+        for (const std::string s :
+                {"1", "12", "-3.5e2", " 7 "
+                })
+        {
+            CAPTURE(s);
+            std::istringstream ss(s);
+            json j;
+            ss >> j;
+            CHECK(remaining(ss).find_first_not_of(" \t\n\r") == std::string::npos);
+        }
+    }
+
+    SECTION("repeated extraction of concatenated values")
+    {
+        std::istringstream ss(R"(1true[2]3"x"{"a":4}5)");
+        const std::vector<json> expected =
+        {
+            json(1), json(true), json::parse("[2]"), json(3),
+            json("x"), json::parse(R"({"a":4})"), json(5)
+        };
+
+        for (const auto& e : expected)
+        {
+            json j;
+            ss >> j;
+            CHECK(j == e);
+        }
+    }
+
+    SECTION("differences to the default behavior")
+    {
+        // both of these work by accident without the macro, because the
+        // character after a number is swallowed; see unit-deserialization.cpp
+
+        SECTION("a separator after a number is not skipped")
+        {
+            std::istringstream ss("1,2");
+            json j;
+            ss >> j;
+            CHECK(j == 1);
+            CHECK_THROWS_AS(ss >> j, json::parse_error&);
+        }
+
+        SECTION("std::getline after a number sees the line break")
+        {
+            std::istringstream ss("42\nfoo");
+            json j;
+            std::string line;
+            ss >> j;
+            std::getline(ss, line);
+            CHECK(j == 42);
+            CHECK(line.empty());
+            std::getline(ss, line);
+            CHECK(line == "foo");
+        }
+    }
+
+    SECTION("sax_parse with strict == false")
+    {
+        std::istringstream ss("1true");
+        json j;
+        nlohmann::detail::json_sax_dom_parser<json, nlohmann::detail::input_stream_adapter> sdp(j, true);
+        CHECK(json::sax_parse(ss, &sdp, nlohmann::detail::input_format_t::json, false));
+        CHECK(j == 1);
+        CHECK(remaining(ss) == "true");
+    }
+
+    SECTION("strict parsing still rejects trailing data")
+    {
+        std::istringstream ss("1true");
+        json _;
+        CHECK_THROWS_WITH_AS(_ = json::parse(ss),
+                             "[json.exception.parse_error.101] parse error at line 1, column 5: syntax error while parsing value - unexpected true literal; expected end of input", json::parse_error&);
+
+        std::istringstream ss2("1true");
+        CHECK_FALSE(json::accept(ss2));
+    }
+
+    SECTION("a streambuf that cannot put back is not needed")
+    {
+        // the terminating character is never consumed, so no putback
+        // position is required
+        no_putback_streambuf buf("1true");
+        std::istream is(&buf);
+        json j;
+        is >> j;
+        CHECK(j == json(1));
+        CHECK(remaining(is) == "true");
+    }
+}

tests/src/unit-ubjson.cpp

@@ -15,6 +15,7 @@ using nlohmann::json;
 #include <fstream>
 #include <set>
 #include "make_test_data_available.hpp"
+#include "round_trip_corpus.hpp"
 #include "test_utils.hpp"
 
 namespace
@@ -2265,7 +2266,9 @@ TEST_CASE("UBJSON optimized arrays of a valueless type are bounded")
 
     SECTION("an excessive count is rejected")
     {
-        // 'l' is a big-endian int32: 0x7FFFFFFF elements, about 34 GB of value
+        // 'l' is a big-endian int32: 0x7FFFFFFF elements, about 34 GB of value;
+        // OSS-Fuzz reported this shape as a parse_ubjson_fuzzer timeout
+        // (testcase 6347769435193344, no issue filed)
         for (const auto marker :
                 {'Z', 'T', 'F'
                 })
@@ -2817,6 +2820,51 @@ TEST_CASE("UBJSON use_type requires use_size")
     }
 }
 
+TEST_CASE("UBJSON round-trip invariants")
+{
+    // This checks what the parse_ubjson_fuzzer driver checks (see
+    // tests/src/fuzzer-parse_ubjson.cpp), so that a regression shows up in CI
+    // rather than as an OSS-Fuzz report: every value from_ubjson() returns
+    // (j1) can be serialized with any combination of options, the result can
+    // be parsed back (j2), and serializing j2 again with the same options
+    // reproduces the exact bytes. Beyond the driver, this also checks that j2
+    // equals j1. Values are compared with dump() rather than operator==,
+    // because a NaN never compares equal to itself.
+    struct options
+    {
+        bool use_size;
+        bool use_type;
+    };
+    const std::vector<options> all_options =
+    {
+        {false, false},
+        {true, false},
+        {true, true},
+    };
+
+    for (const auto& j0 : utils::round_trip_corpus::values())
+    {
+        // turn the corpus value into a value as from_ubjson() returns it; this
+        // has no binary values, as UBJSON writes them as arrays of integers
+        for (const auto& initial : all_options)
+        {
+            const json j1 = json::from_ubjson(json::to_ubjson(j0, initial.use_size, initial.use_type));
+
+            for (const auto& o : all_options)
+            {
+                INFO("j1 = " << j1.dump() << ", use_size = " << o.use_size << ", use_type = " << o.use_type);
+
+                const std::vector<std::uint8_t> vec = json::to_ubjson(j1, o.use_size, o.use_type);
+                json j2;
+                // anything the library writes must be parsable by the library
+                REQUIRE_NOTHROW(j2 = json::from_ubjson(vec));
+                CHECK(j2.dump() == j1.dump());
+                CHECK(json::to_ubjson(j2, o.use_size, o.use_type) == vec);
+            }
+        }
+    }
+}
+
 TEST_CASE("UBJSON roundtrips" * doctest::skip())
 {
     SECTION("input from self-generated UBJSON files")

tools/generate_natvis/generate_natvis.py

@@ -20,7 +20,7 @@ if __name__ == '__main__':
 
     namespaces = ['nlohmann']
     abi_prefix = 'json_abi'
-    abi_tags = ['_diag', '_ldvcmp', '_dp', '_bics']
+    abi_tags = ['_diag', '_ldvcmp', '_dp', '_bics', '_psp']
     version = '_v' + args.version.replace('.', '_')
     inline_namespaces = []