Report BON8 input that ends after a UTF-8 lead byte as truncated

A lead byte (0xC2..0xF7) inside a string begins either another character
(if a continuation byte follows) or an integer (otherwise). When the input
ended right after the lead byte, the reader took the missing byte as "not
a continuation byte", ended the string before the lead byte, and treated
the lead byte as the start of the next value. With strict=false, a message
cut off there was therefore read as a shorter value: the 11 bytes of
"😀😀é" cut after 9 bytes gave "😀😀", and ["aé"] cut after 3 of its 5
bytes gave ["a"]. With strict=true, the input was rejected with a
misleading message ("expected end of input"), or, for a key, with
parse_error.112 instead of 110.

Either reading of the lead byte leaves the message incomplete: a string at
the end of a message must be terminated by 0xFF, so the lead byte cannot
belong to a following message. Report parse_error.110 (unexpected end of
input) for strings and keys, as the comment on get_bon8_string() already
requires and as the reference decoder (HikoGUI) does.

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

docs/mkdocs/docs/api/macros/json_brace_init_copy_semantics.md

@@ -115,4 +115,3 @@ The default value is `0` (disabled — existing behavior is preserved).
 ## 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/community/roadmap.md

@@ -14,7 +14,7 @@ work items are tracked in the [GitHub milestones](https://github.com/nlohmann/js
   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, see [Version 4.0](#version-40).
+  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
@@ -37,65 +37,7 @@ work items are tracked in the [GitHub milestones](https://github.com/nlohmann/js
 
 ## Version 4.0
 
-There is no release date for version 4.0 yet. Proposals that need a major version, for instance stricter type
-conversions, are collected in issue [#3453](https://github.com/nlohmann/json/issues/3453).
-
-!!! note "Not final"
-
-    The plan for version 4.0 described below is not final and may still change: macros may be added to or removed from
-    the list, and planned defaults may be revised. Any such change will be documented on this page.
-
-### Trying out 4.0 today
-
-Version 4.0 will not be developed on a separate branch. Instead, every breaking change is first added to a 3.x release
-behind a macro whose default keeps the 3.x behavior. Version 4.0 then switches the defaults and removes the macros.
-Version 4.0 is therefore the sum of these macros: you can try it on the 3.x release train today by defining each macro
-to its 4.0 value and fixing what no longer compiles or behaves differently. Once your code works with all of them, it
-is ready for version 4.0.
-
-The following macros guard changes that are planned to become the default in version 4.0:
-
-| Macro                                                                                                            | 3.x default | 4.0 behavior                                                                                                                  | CMake option                                                                                                       | Added  |
-|------------------------------------------------------------------------------------------------------------------|-------------|-------------------------------------------------------------------------------------------------------------------------------|--------------------------------------------------------------------------------------------------------------------|--------|
-| [`JSON_USE_IMPLICIT_CONVERSIONS`](../api/macros/json_use_implicit_conversions.md)                                | `1`         | `0`: no implicit conversions from `basic_json` to other types; use [`get`](../api/basic_json/get.md) instead                  | [`JSON_ImplicitConversions`](../integration/cmake.md#json_implicitconversions)                                     | 3.9.0  |
-| [`JSON_USE_GLOBAL_UDLS`](../api/macros/json_use_global_udls.md)                                                  | `1`         | `0`: the string literals `_json` and `_json_pointer` are only available in namespace `nlohmann::literals`                    | [`JSON_GlobalUDLs`](../integration/cmake.md#json_globaludls)                                                       | 3.11.0 |
-| [`JSON_USE_LEGACY_DISCARDED_VALUE_COMPARISON`](../api/macros/json_use_legacy_discarded_value_comparison.md)      | `0`         | removed: the deprecated legacy comparison of discarded values can no longer be enabled                                        | [`JSON_LegacyDiscardedValueComparison`](../integration/cmake.md#json_legacydiscardedvaluecomparison)               | 3.11.0 |
-| [`JSON_BRACE_INIT_COPY_SEMANTICS`](../api/macros/json_brace_init_copy_semantics.md)                              | `0`         | `1`: single-element brace initialization such as `#!cpp json j{obj};` copies the element instead of creating an array         | –                                                                                                                  | 3.13.0 |
-| [`JSON_PRECISE_STREAM_POSITION`](../api/macros/json_precise_stream_position.md)                                  | `0`         | `1`: reading from a stream does not consume the character after a number                                                      | –                                                                                                                  | 3.13.0 |
-| [`JSON_STRICT_NUL_HANDLING`](../api/macros/json_strict_nul_handling.md)                                          | `0`         | `1`: a NUL byte in the input is a parse error instead of the end of input                                                     | [`JSON_StrictNulHandling`](../integration/cmake.md#json_strictnulhandling)                                         | 3.13.0 |
-
-For example, the following makes a 3.x release behave like version 4.0 with respect to these changes:
-
-```cpp
-#define JSON_USE_IMPLICIT_CONVERSIONS 0
-#define JSON_USE_GLOBAL_UDLS 0
-#define JSON_USE_LEGACY_DISCARDED_VALUE_COMPARISON 0
-#define JSON_BRACE_INIT_COPY_SEMANTICS 1
-#define JSON_PRECISE_STREAM_POSITION 1
-#define JSON_STRICT_NUL_HANDLING 1
-#include <nlohmann/json.hpp>
-```
-
-The macros must be defined before the library header is included; setting them once in the build system is the easiest
-way to achieve this.
-
-### Removal of deprecated functions
-
-Version 4.0 will remove all deprecated functions. Compiling with deprecation warnings enabled shows which of them your
-code still uses. The [migration guide](../integration/migration_guide.md#replace-deprecated-functions) shows how to
-replace each of them.
-
-| Deprecated                                                                                                                                                                                                                                         | Since  | Migration                                                                        |
-|----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------|--------|----------------------------------------------------------------------------------|
-| `#!cpp operator<<(basic_json&, std::istream&)`                                                                                                                                                                                                     | 3.0.0  | [Parsing](../integration/migration_guide.md#parsing)                             |
-| `#!cpp operator>>(const basic_json&, std::ostream&)`                                                                                                                                                                                               | 3.0.0  | [Miscellaneous functions](../integration/migration_guide.md#miscellaneous-functions) |
-| `iterator_wrapper`                                                                                                                                                                                                                                 | 3.1.0  | [Miscellaneous functions](../integration/migration_guide.md#miscellaneous-functions) |
-| [`parse`](../api/basic_json/parse.md), [`accept`](../api/basic_json/accept.md), and [`sax_parse`](../api/basic_json/sax_parse.md) with an initializer list `{ptr, len}` or `{first, last}`                                                         | 3.8.0  | [Parsing](../integration/migration_guide.md#parsing)                             |
-| [`from_bson`](../api/basic_json/from_bson.md), [`from_cbor`](../api/basic_json/from_cbor.md), [`from_msgpack`](../api/basic_json/from_msgpack.md), and [`from_ubjson`](../api/basic_json/from_ubjson.md) with `(ptr, len)` or an initializer list | 3.8.0  | [Parsing](../integration/migration_guide.md#parsing)                             |
-| [`json_pointer::operator string_t`](../api/json_pointer/operator_string_t.md)                                                                                                                                                                      | 3.11.0 | [JSON Pointers](../integration/migration_guide.md#json-pointers)                 |
-| [`json_pointer`](../api/json_pointer/index.md) with a `basic_json` type as template argument, and the overloads of `value`, `contains`, `operator[]`, and `at` accepting such a pointer                                                           | 3.11.0 | [JSON Pointers](../integration/migration_guide.md#json-pointers)                 |
-| Comparing a [`json_pointer`](../api/json_pointer/index.md) with a string via [`operator==`](../api/json_pointer/operator_eq.md) or [`operator!=`](../api/json_pointer/operator_ne.md)                                                              | 3.11.2 | [JSON Pointers](../integration/migration_guide.md#json-pointers)                 |
-
-The deprecated legacy comparison of discarded values is controlled by a macro and therefore listed in the table above.
-
-New breaking changes will follow the same path: they are added to these tables when they land in a 3.x release.
+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/integration/migration_guide.md

@@ -1,13 +1,10 @@
 # Migration Guide
 
 This page collects some guidelines on how to future-proof your code for future versions of this library.
-The [roadmap](../community/roadmap.md#version-40) lists what will change in version 4.0, including the macros that let
-you try its behavior with a 3.x release; this page describes how to adjust your code.
 
 ## Replace deprecated functions
 
-The following functions have been deprecated and will be removed in the next major version (i.e., 4.0.0), see the
-[roadmap](../community/roadmap.md#removal-of-deprecated-functions) for an overview. All
+The following functions have been deprecated and will be removed in the next major version (i.e., 4.0.0). All
 deprecations are annotated with
 [`HEDLEY_DEPRECATED_FOR`](https://nemequ.github.io/hedley/api-reference.html#HEDLEY_DEPRECATED_FOR) to report which
 function to use instead.
@@ -37,9 +34,8 @@ function to use instead.
   [`accept`](../api/basic_json/accept.md), [`sax_parse`](../api/basic_json/sax_parse.md),
   [`from_cbor`](../api/basic_json/from_cbor.md), [`from_msgpack`](../api/basic_json/from_msgpack.md),
   [`from_ubjson`](../api/basic_json/from_ubjson.md), and [`from_bson`](../api/basic_json/from_bson.md) via initializer
-  lists is deprecated since 3.8.0. The same holds for passing a pointer and a length as two arguments to the `from_*`
-  functions. Instead, pass two iterators; for instance, call `from_cbor(ptr, ptr+len)` instead of
-  `from_cbor({ptr, len})` or `from_cbor(ptr, len)`.
+  lists is deprecated since 3.8.0. Instead, pass two iterators; for instance, call `from_cbor(ptr, ptr+len)` instead of
+  `from_cbor({ptr, len})`.
 
     === "Deprecated"
   

include/nlohmann/detail/input/binary_reader.hpp

@@ -3821,6 +3821,11 @@ class binary_reader
         if (0xC2 <= byte && byte <= 0xF7)
         {
             const auto second = get_bon8();
+            if (second == char_traits<char_type>::eof())
+            {
+                // the input ends inside a character or an integer
+                return unexpect_eof(input_format_t::bon8, "key");
+            }
             unget_bon8(second);
             if (is_bon8_continuation(second))
             {
@@ -3919,6 +3924,12 @@ class binary_reader
             // a lead byte ends the string if no continuation byte follows: it
             // is then the first byte of an integer
             const auto second = get_bon8();
+            if (second == char_traits<char_type>::eof())
+            {
+                // the input ends inside a character or an integer: either
+                // way, the message is incomplete
+                return unexpect_eof(input_format_t::bon8, "string");
+            }
             if (!is_bon8_continuation(second))
             {
                 unget_bon8(second);

single_include/nlohmann/json.hpp

@@ -16588,6 +16588,11 @@ class binary_reader
         if (0xC2 <= byte && byte <= 0xF7)
         {
             const auto second = get_bon8();
+            if (second == char_traits<char_type>::eof())
+            {
+                // the input ends inside a character or an integer
+                return unexpect_eof(input_format_t::bon8, "key");
+            }
             unget_bon8(second);
             if (is_bon8_continuation(second))
             {
@@ -16686,6 +16691,12 @@ class binary_reader
             // a lead byte ends the string if no continuation byte follows: it
             // is then the first byte of an integer
             const auto second = get_bon8();
+            if (second == char_traits<char_type>::eof())
+            {
+                // the input ends inside a character or an integer: either
+                // way, the message is incomplete
+                return unexpect_eof(input_format_t::bon8, "string");
+            }
             if (!is_bon8_continuation(second))
             {
                 unget_bon8(second);

tests/src/unit-bon8.cpp

@@ -531,6 +531,47 @@ TEST_CASE("BON8")
             CHECK_THROWS_WITH_AS(_ = json::from_bon8(bytes{0x87, 'a'}), "[json.exception.parse_error.110] parse error at byte 3: syntax error while parsing BON8 string: unexpected end of input", json::parse_error&);
         }
 
+        SECTION("input that ends after a UTF-8 lead byte")
+        {
+            // the lead byte begins either a character or an integer; both are
+            // incomplete, so the lead byte must not end the string before it
+            for (const bool strict :
+                    {
+                        true, false
+                    })
+            {
+                CAPTURE(strict)
+                CHECK_THROWS_WITH_AS(_ = json::from_bon8(bytes{'a', 0xC3}, strict), "[json.exception.parse_error.110] parse error at byte 3: syntax error while parsing BON8 string: unexpected end of input", json::parse_error&);
+                CHECK_THROWS_WITH_AS(_ = json::from_bon8(bytes{0x81, 'a', 0xC3}, strict), "[json.exception.parse_error.110] parse error at byte 4: syntax error while parsing BON8 string: unexpected end of input", json::parse_error&);
+                CHECK_THROWS_WITH_AS(_ = json::from_bon8(bytes{0x81, 'a', 0xE2}, strict), "[json.exception.parse_error.110] parse error at byte 4: syntax error while parsing BON8 string: unexpected end of input", json::parse_error&);
+                CHECK_THROWS_WITH_AS(_ = json::from_bon8(bytes{0x81, 'a', 0xF0}, strict), "[json.exception.parse_error.110] parse error at byte 4: syntax error while parsing BON8 string: unexpected end of input", json::parse_error&);
+                CHECK_THROWS_WITH_AS(_ = json::from_bon8(bytes{0x81, 0xC3, 0xA9, 0xC3}, strict), "[json.exception.parse_error.110] parse error at byte 5: syntax error while parsing BON8 string: unexpected end of input", json::parse_error&);
+                CHECK_THROWS_WITH_AS(_ = json::from_bon8(bytes{0x87, 'a', 0xC3}, strict), "[json.exception.parse_error.110] parse error at byte 4: syntax error while parsing BON8 string: unexpected end of input", json::parse_error&);
+                CHECK_THROWS_WITH_AS(_ = json::from_bon8(bytes{0x87, 0xC3}, strict), "[json.exception.parse_error.110] parse error at byte 3: syntax error while parsing BON8 key: unexpected end of input", json::parse_error&);
+                CHECK_THROWS_WITH_AS(_ = json::from_bon8(bytes{0x88, 'a', 0x91, 0xE2}, strict), "[json.exception.parse_error.110] parse error at byte 5: syntax error while parsing BON8 key: unexpected end of input", json::parse_error&);
+            }
+        }
+
+        SECTION("a message that is cut off is not read as a shorter value")
+        {
+            const json values = {"\xC3\xA9", "a\xE2\x82\xAC", "\xF0\x9F\x98\x80\xC3\xA9", {"a\xC3\xA9"}, {{"\xC3\xA9", "\xE2\x82\xAC"}}, {{"a", {"b\xC3\xA9", 1}}}};
+            for (const auto& j : values)
+            {
+                const bytes message = json::to_bon8(j);
+                for (std::size_t length = 0; length < message.size(); ++length)
+                {
+                    CAPTURE(j)
+                    CAPTURE(length)
+                    bytes prefix = message;
+                    prefix.resize(length);
+                    CHECK(json::from_bon8(prefix, false, false).is_discarded());
+                    // a stream is read byte by byte rather than in bulk
+                    std::istringstream stream(str(prefix));
+                    CHECK(json::from_bon8(stream, false, false).is_discarded());
+                }
+            }
+        }
+
         SECTION("invalid UTF-8")
         {
             // overlong