CBOR/MessagePack readers: leave one-element arrays to the general path

enter_cbor_array and enter_msgpack_array called the out-of-line scalar
run for every non-empty array. When the array's element is an object or
an array, as on every level of [[...]] or the single ring of a GeoJSON
polygon, the call reads nothing; it made 1,000 arrays nested in each
other cost 4 % more instructions (GCC 13) and up to 8 % more time (Apple
clang) than without the run. #5975 had fixed that with an inline
pre-check, which was later dropped after measuring canada only.

The run is now entered only for arrays of two or more elements: the same
comparison as before, so no other row changes. Arrays of one scalar,
which the run read faster, now take the general path; they are rare in
documents (one in citm_catalog, none in twitter or canada). The new
benchmark FromBinaryShape/singles measures them, next to
FromBinaryShape/nested.

Instructions per repetition, new/old, GCC 13 callgrind: nested arrays
0.973 (CBOR) and 0.971 (MessagePack), arrays of one scalar 1.037 and
1.021, every other row within 0.15 %.

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

.github/workflows/cancel_closed_pr_runs.yml

@@ -24,7 +24,7 @@ jobs:
     permissions:
       actions: write
 
-    runs-on: ubuntu-latest
+    runs-on: ubuntu-26.04
 
     steps:
       - name: Harden Runner

.github/workflows/check_amalgamation.yml

@@ -19,7 +19,7 @@ permissions:
 jobs:
   save:
     if: github.event_name == 'pull_request'
-    runs-on: ubuntu-latest
+    runs-on: ubuntu-26.04
     steps:
       - name: Harden Runner
         uses: step-security/harden-runner@dc2769abcc0fcfbadfc1df54081c1608579ef6b2 # v2.22.2

.github/workflows/check_docs_links.yml

@@ -16,7 +16,7 @@ concurrency:
 jobs:
   check_docs_links:
     if: github.repository == 'nlohmann/json'
-    runs-on: ubuntu-latest
+    runs-on: ubuntu-26.04
     timeout-minutes: 30
     steps:
       - name: Harden Runner

.github/workflows/comment_check_amalgamation.yml

@@ -12,7 +12,7 @@ jobs:
   comment:
     # push runs on develop have no PR to comment on (and no "pr" artifact)
     if: ${{ github.event.workflow_run.conclusion == 'failure' && github.event.workflow_run.event == 'pull_request' }}
-    runs-on: ubuntu-latest
+    runs-on: ubuntu-26.04
     permissions:
       contents: read
       actions: read

.github/workflows/dependency-review.yml

@@ -18,7 +18,7 @@ permissions:
 
 jobs:
   dependency-review:
-    runs-on: ubuntu-latest
+    runs-on: ubuntu-26.04
     steps:
       - name: Harden Runner
         uses: step-security/harden-runner@dc2769abcc0fcfbadfc1df54081c1608579ef6b2 # v2.22.2

.github/workflows/flawfinder.yml

@@ -24,7 +24,7 @@ on:
 jobs:
   flawfinder:
     name: Flawfinder
-    runs-on: ubuntu-latest
+    runs-on: ubuntu-26.04
     permissions:
       actions: read
       contents: read

.github/workflows/labeler.yml

@@ -19,7 +19,7 @@ jobs:
       contents: read
       pull-requests: write
 
-    runs-on: ubuntu-latest
+    runs-on: ubuntu-26.04
 
     steps:
       - name: Harden Runner

.github/workflows/publish_documentation.yml

@@ -34,7 +34,7 @@ jobs:
       contents: write
 
     if: github.repository == 'nlohmann/json'
-    runs-on: ubuntu-latest
+    runs-on: ubuntu-26.04
     steps:
       - name: Harden Runner
         uses: step-security/harden-runner@dc2769abcc0fcfbadfc1df54081c1608579ef6b2 # v2.22.2

.github/workflows/scorecards.yml

@@ -24,7 +24,7 @@ permissions:
 jobs:
   analysis:
     name: Scorecard analysis
-    runs-on: ubuntu-latest
+    runs-on: ubuntu-26.04
     permissions:
       # Needed to upload the results to code-scanning dashboard.
       security-events: write

.github/workflows/stale.yml

@@ -8,7 +8,7 @@ permissions:
 
 jobs:
   stale:
-    runs-on: ubuntu-latest
+    runs-on: ubuntu-26.04
 
     permissions:
       issues: write

.github/workflows/ubuntu.yml

@@ -18,7 +18,7 @@ concurrency:
 
 jobs:
   ci_test_gcc:
-    runs-on: ubuntu-latest
+    runs-on: ubuntu-26.04
     container: gcc:latest
     steps:
       - uses: actions/checkout@3d3c42e5aac5ba805825da76410c181273ba90b1 # v7.0.1
@@ -130,7 +130,7 @@ jobs:
         run: cmake --build build --target ${{ matrix.target }}
 
   ci_static_analysis_clang:
-    runs-on: ubuntu-latest
+    runs-on: ubuntu-26.04
     container: silkeh/clang:22
     strategy:
       matrix:
@@ -149,7 +149,7 @@ jobs:
         run: cmake --build build --target ${{ matrix.target }}
 
   ci_cmake_options:
-    runs-on: ubuntu-latest
+    runs-on: ubuntu-26.04
     container: ubuntu:24.04
     strategy:
       matrix:
@@ -209,7 +209,7 @@ jobs:
           fail-on-error: false
 
   ci_test_compilers_gcc_old:
-    runs-on: ubuntu-latest
+    runs-on: ubuntu-26.04
     strategy:
       matrix:
         compiler: ['4.8', '4.9', '5', '6']
@@ -251,7 +251,7 @@ jobs:
         run: cmake --build build --target ci_test_compiler_g++-${{ matrix.compiler }}
 
   ci_test_compilers_gcc:
-    runs-on: ubuntu-latest
+    runs-on: ubuntu-26.04
     strategy:
       matrix:
         # older GCC docker images (4, 5, 6) fail to check out code
@@ -269,7 +269,7 @@ jobs:
         run: cmake --build build --target ci_test_compiler_default
 
   ci_test_compilers_clang:
-    runs-on: ubuntu-latest
+    runs-on: ubuntu-26.04
     strategy:
       matrix:
         compiler: ['3.4', '3.5', '3.6', '3.7', '3.8', '3.9', '4', '5', '6', '7', '8', '9', '10', '11', '12', '13', '14', '15-bullseye', '16', '17', '18', '19', '20', '21', '22', 'latest']
@@ -289,7 +289,7 @@ jobs:
         run: cmake --build build --target ci_test_compiler_default
 
   ci_test_standards_gcc:
-    runs-on: ubuntu-latest
+    runs-on: ubuntu-26.04
     container: gcc:latest
     strategy:
       matrix:
@@ -306,7 +306,7 @@ jobs:
         run: cmake --build build --target ci_test_gcc_cxx${{ matrix.standard }}
 
   ci_test_standards_clang:
-    runs-on: ubuntu-latest
+    runs-on: ubuntu-26.04
     container: silkeh/clang:latest
     strategy:
       matrix:
@@ -330,7 +330,7 @@ jobs:
         if: ${{ matrix.stdlib == 'libstdcxx' }}
 
   ci_cuda_example:
-    runs-on: ubuntu-latest
+    runs-on: ubuntu-26.04
     strategy:
       fail-fast: false
       matrix:
@@ -356,7 +356,7 @@ jobs:
     strategy:
       matrix:
         container: ['gcc:latest', 'silkeh/clang:latest']
-    runs-on: ubuntu-latest
+    runs-on: ubuntu-26.04
     container: ${{ matrix.container }}
     steps:
       - uses: actions/checkout@3d3c42e5aac5ba805825da76410c181273ba90b1 # v7.0.1
@@ -383,7 +383,7 @@ jobs:
         run: cmake --build build --target ci_module_cpp20
 
   ci_icpc:
-    runs-on: ubuntu-latest
+    runs-on: ubuntu-26.04
     # Intel discontinued the classic icc/icpc compiler in oneAPI 2024.0; this is
     # Intel's own last officially published image that still includes it.
     container: intel/oneapi-hpckit:2023.2.1-devel-ubuntu22.04
@@ -402,7 +402,7 @@ jobs:
         run: cmake --build build --target ci_icpc
 
   ci_icpx:
-    runs-on: ubuntu-latest
+    runs-on: ubuntu-26.04
     container: intel/oneapi-hpckit:latest
     steps:
       - uses: actions/checkout@3d3c42e5aac5ba805825da76410c181273ba90b1 # v7.0.1
@@ -416,7 +416,7 @@ jobs:
         run: cmake --build build --target ci_icpx
 
   ci_nvhpc:
-    runs-on: ubuntu-latest
+    runs-on: ubuntu-26.04
     container: nvcr.io/nvidia/nvhpc:25.5-devel-cuda12.9-ubuntu22.04
     steps:
       - uses: actions/checkout@3d3c42e5aac5ba805825da76410c181273ba90b1 # v7.0.1

docs/mkdocs/docs/api/basic_json/insert.md

@@ -65,17 +65,7 @@ the same index, which is now a different value.
 
 ## Exception safety
 
-1. Strong guarantee: if an exception is thrown, there are no changes in the JSON value.
-2. Basic guarantee: if an exception is thrown while the elements are copied (for example, `std::bad_alloc`), the JSON
-   value may be partially modified, but it stays a valid array.
-3. Basic guarantee: if an exception is thrown while the elements are copied (for example, `std::bad_alloc`), the JSON
-   value may be partially modified, but it stays a valid array.
-4. Basic guarantee: if an exception is thrown while the elements are copied (for example, `std::bad_alloc`), the JSON
-   value may be partially modified, but it stays a valid array.
-5. Basic guarantee: if an exception is thrown while the members are copied (for example, `std::bad_alloc`), the JSON
-   value may be partially modified, but it stays a valid object.
-
-The exceptions listed below are thrown before the JSON value is changed.
+Strong exception safety: if an exception occurs, the original value stays intact.
 
 ## Exceptions
 

include/nlohmann/detail/input/binary_reader.hpp

@@ -467,11 +467,13 @@ class binary_reader
 
             const std::size_t element_type_parse_position = chars_read;
             // an array's elements are named "0", "1", ... in the wire format,
-            // and those names are not passed on: they are skipped unless
-            // error_handler_t::strict must reject ill-formed UTF-8 in them
+            // and those names are not passed on: they are skipped, and only
+            // checked if error_handler_t::strict must reject ill-formed UTF-8
             const bool is_object = container_stack.back().is_object;
             key.clear();
-            if (JSON_HEDLEY_UNLIKELY(!get_bson_cstr(key, is_object || error_handler == error_handler_t::strict)))
+            if (JSON_HEDLEY_UNLIKELY(!(is_object || error_handler != error_handler_t::strict
+                                       ? get_bson_cstr(key, is_object)
+                                       : check_bson_cstr(key, std::integral_constant<bool, bulk_scan> {}))))
             {
                 return false;
             }
@@ -495,7 +497,7 @@ class binary_reader
     @param[in] store  whether the characters are needed; if not (the names of
                       an array's elements), they are skipped and @a result may
                       be left unchanged, and they are not checked against
-                      @ref error_handler
+                      @ref error_handler (@ref check_bson_cstr checks them)
     @return `true` if the \\x00-byte indicating the end of the string was
              encountered before the EOF; false` indicates an unexpected EOF.
     */
@@ -554,6 +556,42 @@ class binary_reader
         return true;
     }
 
+    /*!
+    @brief skip a C-style string that error_handler_t::strict must check
+
+    The names of an array's elements are not passed on. On contiguous input,
+    a well-formed name is checked in place and skipped without being copied.
+    Otherwise (no \\x00-byte in the input that can be scanned, or ill-formed
+    UTF-8), it is read by @ref get_bson_cstr, which reports the error.
+
+    @param[in,out] result  an empty string, which may receive the name
+    @return whether the name was read and is well-formed
+    */
+    bool check_bson_cstr(string_t& result, std::true_type /*bulk*/)
+    {
+        const std::size_t remaining = ia.bulk_remaining();
+        if (remaining != 0)
+        {
+            const auto* const data = reinterpret_cast<const unsigned char*>(ia.bulk_data());
+            const std::size_t length = find_zero_byte(data, remaining);
+            if (length != remaining && JSON_HEDLEY_LIKELY(valid_utf8_string(data, length)))
+            {
+                // consume the string and its \x00-byte, as get_bson_cstr does
+                ia.bulk_skip(length + 1);
+                chars_read += length + 1;
+                current = 0x00;
+                return true;
+            }
+        }
+        return get_bson_cstr(result);
+    }
+
+    /// input that is not contiguous: the name is read and checked byte by byte
+    bool check_bson_cstr(string_t& result, std::false_type /*bulk*/)
+    {
+        return get_bson_cstr(result);
+    }
+
     /// input that is not contiguous: C-style strings are read byte by byte
     bool get_bson_cstr_bulk(string_t& /*result*/, const bool /*store*/, std::false_type /*bulk*/) const noexcept
     {
@@ -1818,14 +1856,18 @@ class binary_reader
 
     Only the start of an array is tried, which covers arrays of numbers
     (e.g. coordinates) but costs arrays of other values only one call, not one
-    per element.
+    per element. An array of one element is not tried at all: that element is
+    mostly an object or an array (`[[...]]`, the single ring of a GeoJSON
+    polygon), for which the call would find no scalar, while a single scalar,
+    which the run would read faster, is rare. The benchmarks
+    `FromBinaryShape/nested` and `FromBinaryShape/singles` show both sides.
 
     @param[in] len  the number of elements
     @return whether the SAX parser accepted the array and the values read
     */
     bool enter_cbor_array(const std::size_t len)
     {
-        return enter_array(len) && (len == 0 || parse_cbor_scalar_run(std::integral_constant<bool, bulk_scan> {}));
+        return enter_array(len) && (len <= 1 || parse_cbor_scalar_run(std::integral_constant<bool, bulk_scan> {}));
     }
 
     /// input that is not contiguous: every value is read by parse_cbor_value
@@ -2726,12 +2768,16 @@ class binary_reader
     /*!
     @brief open a MessagePack array and read the scalars it begins with (see
            @ref parse_msgpack_scalar_run)
+
+    An array of one element is not tried, for the reason given at
+    @ref enter_cbor_array.
+
     @param[in] len  the number of elements
     @return whether the SAX parser accepted the array and the values read
     */
     bool enter_msgpack_array(const std::size_t len)
     {
-        return enter_array(len) && (len == 0 || parse_msgpack_scalar_run(std::integral_constant<bool, bulk_scan> {}));
+        return enter_array(len) && (len <= 1 || parse_msgpack_scalar_run(std::integral_constant<bool, bulk_scan> {}));
     }
 
     /// input that is not contiguous: every value is read by parse_msgpack_value

include/nlohmann/json.hpp

@@ -4914,30 +4914,6 @@ public:
         return {it, res.second};
     }
 
-  private:
-    /// Sets the parents of a value when it goes out of scope, also when an
-    /// exception is thrown: a container insertion that throws may already
-    /// have added elements, whose parents must be set for the value to stay
-    /// valid (basic exception guarantee). Does nothing without JSON_DIAGNOSTICS.
-    class set_parents_on_exit
-    {
-      public:
-        explicit set_parents_on_exit(basic_json* v) noexcept : m_value(v) {}
-        set_parents_on_exit(const set_parents_on_exit&) = delete;
-        set_parents_on_exit& operator=(const set_parents_on_exit&) = delete;
-        set_parents_on_exit(set_parents_on_exit&&) = delete;
-        set_parents_on_exit& operator=(set_parents_on_exit&&) = delete;
-
-        ~set_parents_on_exit()
-        {
-            m_value->set_parents();
-        }
-
-      private:
-        basic_json* m_value;
-    };
-
-  public:
     /// Helper for insertion of an iterator
     /// @note: This uses std::distance to support GCC 4.8,
     ///        see https://github.com/nlohmann/json/pull/1257
@@ -4947,7 +4923,6 @@ public:
         iterator result(this);
         JSON_ASSERT(m_data.m_value.array != nullptr);
 
-        const set_parents_on_exit parents(this);
         auto insert_pos = std::distance(m_data.m_value.array->begin(), pos.m_it.array_iterator);
         m_data.m_value.array->insert(pos.m_it.array_iterator, std::forward<Args>(args)...);
         result.m_it.array_iterator = m_data.m_value.array->begin() + insert_pos;
@@ -4956,6 +4931,7 @@ public:
         // result.m_it.array_iterator = m_data.m_value.array->insert(pos.m_it.array_iterator, cnt, val);
         // but the return value of insert is missing in GCC 4.8, so it is written this way instead.
 
+        set_parents();
         return result;
     }
 
@@ -4975,16 +4951,8 @@ public:
             throw_iterator_from_other_value(this);
         }
 
-        // vector::insert has no effects if copying a single element to the
-        // end throws; elsewhere, it may already have shifted the elements, so
-        // copy first and insert by move, which cannot throw but by allocating
-        if (pos.m_it.array_iterator == m_data.m_value.array->end())
-        {
-            return insert_iterator(pos, val);
-        }
-
-        basic_json tmp(val);
-        return insert_iterator(pos, std::move(tmp));
+        // insert to array and return iterator
+        return insert_iterator(pos, val);
     }
 
     /// @brief inserts element into array
@@ -5116,8 +5084,8 @@ public:
             detail::throw_literal<invalid_iterator>(detail::exception_id::iterator_from_other_value, "iterators first and last must point to objects", this);
         }
 
-        const set_parents_on_exit parents(this);
         m_data.m_value.object->insert(first.m_it.object_iterator, last.m_it.object_iterator);
+        set_parents();
     }
 
     /// @brief updates a JSON object from another object, overwriting existing keys

single_include/nlohmann/json.hpp

@@ -18474,11 +18474,13 @@ class binary_reader
 
             const std::size_t element_type_parse_position = chars_read;
             // an array's elements are named "0", "1", ... in the wire format,
-            // and those names are not passed on: they are skipped unless
-            // error_handler_t::strict must reject ill-formed UTF-8 in them
+            // and those names are not passed on: they are skipped, and only
+            // checked if error_handler_t::strict must reject ill-formed UTF-8
             const bool is_object = container_stack.back().is_object;
             key.clear();
-            if (JSON_HEDLEY_UNLIKELY(!get_bson_cstr(key, is_object || error_handler == error_handler_t::strict)))
+            if (JSON_HEDLEY_UNLIKELY(!(is_object || error_handler != error_handler_t::strict
+                                       ? get_bson_cstr(key, is_object)
+                                       : check_bson_cstr(key, std::integral_constant<bool, bulk_scan> {}))))
             {
                 return false;
             }
@@ -18502,7 +18504,7 @@ class binary_reader
     @param[in] store  whether the characters are needed; if not (the names of
                       an array's elements), they are skipped and @a result may
                       be left unchanged, and they are not checked against
-                      @ref error_handler
+                      @ref error_handler (@ref check_bson_cstr checks them)
     @return `true` if the \\x00-byte indicating the end of the string was
              encountered before the EOF; false` indicates an unexpected EOF.
     */
@@ -18561,6 +18563,42 @@ class binary_reader
         return true;
     }
 
+    /*!
+    @brief skip a C-style string that error_handler_t::strict must check
+
+    The names of an array's elements are not passed on. On contiguous input,
+    a well-formed name is checked in place and skipped without being copied.
+    Otherwise (no \\x00-byte in the input that can be scanned, or ill-formed
+    UTF-8), it is read by @ref get_bson_cstr, which reports the error.
+
+    @param[in,out] result  an empty string, which may receive the name
+    @return whether the name was read and is well-formed
+    */
+    bool check_bson_cstr(string_t& result, std::true_type /*bulk*/)
+    {
+        const std::size_t remaining = ia.bulk_remaining();
+        if (remaining != 0)
+        {
+            const auto* const data = reinterpret_cast<const unsigned char*>(ia.bulk_data());
+            const std::size_t length = find_zero_byte(data, remaining);
+            if (length != remaining && JSON_HEDLEY_LIKELY(valid_utf8_string(data, length)))
+            {
+                // consume the string and its \x00-byte, as get_bson_cstr does
+                ia.bulk_skip(length + 1);
+                chars_read += length + 1;
+                current = 0x00;
+                return true;
+            }
+        }
+        return get_bson_cstr(result);
+    }
+
+    /// input that is not contiguous: the name is read and checked byte by byte
+    bool check_bson_cstr(string_t& result, std::false_type /*bulk*/)
+    {
+        return get_bson_cstr(result);
+    }
+
     /// input that is not contiguous: C-style strings are read byte by byte
     bool get_bson_cstr_bulk(string_t& /*result*/, const bool /*store*/, std::false_type /*bulk*/) const noexcept
     {
@@ -19825,14 +19863,18 @@ class binary_reader
 
     Only the start of an array is tried, which covers arrays of numbers
     (e.g. coordinates) but costs arrays of other values only one call, not one
-    per element.
+    per element. An array of one element is not tried at all: that element is
+    mostly an object or an array (`[[...]]`, the single ring of a GeoJSON
+    polygon), for which the call would find no scalar, while a single scalar,
+    which the run would read faster, is rare. The benchmarks
+    `FromBinaryShape/nested` and `FromBinaryShape/singles` show both sides.
 
     @param[in] len  the number of elements
     @return whether the SAX parser accepted the array and the values read
     */
     bool enter_cbor_array(const std::size_t len)
     {
-        return enter_array(len) && (len == 0 || parse_cbor_scalar_run(std::integral_constant<bool, bulk_scan> {}));
+        return enter_array(len) && (len <= 1 || parse_cbor_scalar_run(std::integral_constant<bool, bulk_scan> {}));
     }
 
     /// input that is not contiguous: every value is read by parse_cbor_value
@@ -20733,12 +20775,16 @@ class binary_reader
     /*!
     @brief open a MessagePack array and read the scalars it begins with (see
            @ref parse_msgpack_scalar_run)
+
+    An array of one element is not tried, for the reason given at
+    @ref enter_cbor_array.
+
     @param[in] len  the number of elements
     @return whether the SAX parser accepted the array and the values read
     */
     bool enter_msgpack_array(const std::size_t len)
     {
-        return enter_array(len) && (len == 0 || parse_msgpack_scalar_run(std::integral_constant<bool, bulk_scan> {}));
+        return enter_array(len) && (len <= 1 || parse_msgpack_scalar_run(std::integral_constant<bool, bulk_scan> {}));
     }
 
     /// input that is not contiguous: every value is read by parse_msgpack_value
@@ -42706,30 +42752,6 @@ public:
         return {it, res.second};
     }
 
-  private:
-    /// Sets the parents of a value when it goes out of scope, also when an
-    /// exception is thrown: a container insertion that throws may already
-    /// have added elements, whose parents must be set for the value to stay
-    /// valid (basic exception guarantee). Does nothing without JSON_DIAGNOSTICS.
-    class set_parents_on_exit
-    {
-      public:
-        explicit set_parents_on_exit(basic_json* v) noexcept : m_value(v) {}
-        set_parents_on_exit(const set_parents_on_exit&) = delete;
-        set_parents_on_exit& operator=(const set_parents_on_exit&) = delete;
-        set_parents_on_exit(set_parents_on_exit&&) = delete;
-        set_parents_on_exit& operator=(set_parents_on_exit&&) = delete;
-
-        ~set_parents_on_exit()
-        {
-            m_value->set_parents();
-        }
-
-      private:
-        basic_json* m_value;
-    };
-
-  public:
     /// Helper for insertion of an iterator
     /// @note: This uses std::distance to support GCC 4.8,
     ///        see https://github.com/nlohmann/json/pull/1257
@@ -42739,7 +42761,6 @@ public:
         iterator result(this);
         JSON_ASSERT(m_data.m_value.array != nullptr);
 
-        const set_parents_on_exit parents(this);
         auto insert_pos = std::distance(m_data.m_value.array->begin(), pos.m_it.array_iterator);
         m_data.m_value.array->insert(pos.m_it.array_iterator, std::forward<Args>(args)...);
         result.m_it.array_iterator = m_data.m_value.array->begin() + insert_pos;
@@ -42748,6 +42769,7 @@ public:
         // result.m_it.array_iterator = m_data.m_value.array->insert(pos.m_it.array_iterator, cnt, val);
         // but the return value of insert is missing in GCC 4.8, so it is written this way instead.
 
+        set_parents();
         return result;
     }
 
@@ -42767,16 +42789,8 @@ public:
             throw_iterator_from_other_value(this);
         }
 
-        // vector::insert has no effects if copying a single element to the
-        // end throws; elsewhere, it may already have shifted the elements, so
-        // copy first and insert by move, which cannot throw but by allocating
-        if (pos.m_it.array_iterator == m_data.m_value.array->end())
-        {
-            return insert_iterator(pos, val);
-        }
-
-        basic_json tmp(val);
-        return insert_iterator(pos, std::move(tmp));
+        // insert to array and return iterator
+        return insert_iterator(pos, val);
     }
 
     /// @brief inserts element into array
@@ -42908,8 +42922,8 @@ public:
             detail::throw_literal<invalid_iterator>(detail::exception_id::iterator_from_other_value, "iterators first and last must point to objects", this);
         }
 
-        const set_parents_on_exit parents(this);
         m_data.m_value.object->insert(first.m_it.object_iterator, last.m_it.object_iterator);
+        set_parents();
     }
 
     /// @brief updates a JSON object from another object, overwriting existing keys

tests/benchmarks/src/benchmarks.cpp

@@ -629,6 +629,22 @@ static json make_containers()
     return j;
 }
 
+/// many arrays of one number each: one-element arrays are left to the general
+/// path (CBOR and MessagePack), which reads a lone number slower than the loop
+/// for arrays of scalars would
+static json make_singles()
+{
+    json data = json::array();
+    for (std::size_t i = 0; i < 100000; ++i)
+    {
+        data.push_back(json::array({i}));
+    }
+
+    json j = json::object();
+    j["data"] = std::move(data);
+    return j;
+}
+
 /// one flat array of numbers: the scalar decoding path, which must not move
 static json make_scalars()
 {
@@ -676,6 +692,8 @@ BENCHMARK_CAPTURE(FromBinaryShape, containers / ubjson_optimized, make_container
 BENCHMARK_CAPTURE(FromBinaryShape, containers / bjdata, make_containers, binary_format::bjdata);
 BENCHMARK_CAPTURE(FromBinaryShape, containers / bson, make_containers, binary_format::bson);
 BENCHMARK_CAPTURE(FromBinaryShape, containers / bon8, make_containers, binary_format::bon8);
+BENCHMARK_CAPTURE(FromBinaryShape, singles / cbor, make_singles, binary_format::cbor);
+BENCHMARK_CAPTURE(FromBinaryShape, singles / msgpack, make_singles, binary_format::msgpack);
 // BSON names every array element, so a large array measures key generation
 // rather than scalar decoding and is left out here
 BENCHMARK_CAPTURE(FromBinaryShape, scalars / cbor, make_scalars, binary_format::cbor);

tests/src/unit-allocator.cpp

@@ -12,7 +12,6 @@
 #include <nlohmann/json.hpp>
 using nlohmann::json;
 
-#include <functional>
 #include <valarray>
 #if JSON_HAS_RANGES
     #include <ranges>
@@ -805,222 +804,3 @@ TEST_CASE("a failed allocation leaves the value unchanged")
 #endif
 }
 #endif
-
-// the no-exceptions CI job has no std::bad_alloc to catch; with iterator
-// debugging (MSVC STL debug builds) containers allocate inside noexcept
-// functions, so failing an allocation there terminates the program
-#if !defined(JSON_NOEXCEPTION) && !(defined(_ITERATOR_DEBUG_LEVEL) && _ITERATOR_DEBUG_LEVEL > 0)
-namespace
-{
-// whether every child of j names j as its parent (always true without
-// JSON_DIAGNOSTICS, where no parents are stored)
-template<class BasicJsonType>
-bool children_name_their_parent(const BasicJsonType& j)
-{
-#if JSON_DIAGNOSTICS
-    if (j.is_array())
-    {
-        for (const auto& element : *j.m_data.m_value.array)
-        {
-            if (element.m_parent != &j)
-            {
-                return false;
-            }
-        }
-    }
-    else if (j.is_object())
-    {
-        for (const auto& element : *j.m_data.m_value.object)
-        {
-            if (element.second.m_parent != &j)
-            {
-                return false;
-            }
-        }
-    }
-#else
-    static_cast<void>(j);
-#endif
-    return true;
-}
-
-enum class guarantee
-{
-    basic, // the value stays valid and keeps its type
-    strong // the value is unchanged
-};
-
-// Runs op on a fresh value make() with the 0th, 1st, 2nd, ... allocation made
-// by the library failing in turn, until op completes. Every failed attempt
-// must have thrown std::bad_alloc and left the value valid, with all parent
-// pointers intact (copying a value asserts the invariant), and for the strong
-// guarantee exactly as it was.
-template<class BasicJsonType>
-void check_exception_guarantee(guarantee g,
-                               const std::function<BasicJsonType()>& make,
-                               const std::function<void(BasicJsonType&)>& op)
-{
-    fail_at_alloc_call = -1;
-
-    bool completed = false;
-    std::size_t failures = 0;
-    for (std::size_t n = 0; !completed; ++n)
-    {
-        CAPTURE(n)
-        REQUIRE(n < 1000);
-
-        BasicJsonType j = make();
-        const std::string before = j.dump();
-        const auto type_before = j.type();
-
-        alloc_call_count = 0;
-        fail_at_alloc_call = static_cast<long>(n);
-        try
-        {
-            op(j);
-            completed = true;
-        }
-        catch (const std::bad_alloc&)
-        {
-            ++failures;
-        }
-        fail_at_alloc_call = -1;
-
-        if (!completed)
-        {
-            CHECK((j.type() == type_before));
-            CHECK(children_name_their_parent(j));
-            const BasicJsonType copy(j); // NOLINT(performance-unnecessary-copy-initialization): the copy is the point, it allocates
-            CHECK(copy.dump() == j.dump());
-            if (g == guarantee::strong)
-            {
-                CHECK(j.dump() == before);
-            }
-        }
-    }
-
-    // the operation allocates, so at least one attempt must have failed
-    CHECK(failures > 0);
-}
-
-template<class BasicJsonType>
-void check_insert_exception_guarantees()
-{
-    using json_t = BasicJsonType;
-    using make_t = std::function<json_t()>;
-
-    // strings of this length do not fit into the small string buffer, so
-    // copying them allocates
-    const json_t val = std::string(64, 'x');
-    const json_t source = json_t::array({std::string(64, 'p'), std::string(64, 'q'), std::string(64, 'r')});
-
-    for (const bool spare_capacity :
-            {
-                false, true
-            })
-    {
-        CAPTURE(spare_capacity)
-
-        // [ "aaa...", "bbb...", "ccc..." ], with a spare slot if requested
-        const make_t make = [spare_capacity]()
-        {
-            json_t j = json_t::array({std::string(64, 'a'), std::string(64, 'b'), std::string(64, 'c')});
-            if (spare_capacity)
-            {
-                j.push_back(std::string(64, 'd'));
-                j.erase(3);
-            }
-            return json_t(std::move(j));
-        };
-
-        for (const bool at_end :
-                {
-                    false, true
-                })
-        {
-            CAPTURE(at_end)
-
-            SECTION("insert(pos, const basic_json&)")
-            {
-                check_exception_guarantee<json_t>(guarantee::strong, make, [&](json_t& j)
-                {
-                    j.insert(at_end ? j.end() : j.begin() + 1, val);
-                });
-            }
-
-            SECTION("insert(pos, basic_json&&)")
-            {
-                check_exception_guarantee<json_t>(guarantee::strong, make, [&](json_t& j)
-                {
-                    j.insert(at_end ? j.end() : j.begin() + 1, json_t(val));
-                });
-            }
-
-            SECTION("insert(pos, size_type, const basic_json&)")
-            {
-                check_exception_guarantee<json_t>(guarantee::basic, make, [&](json_t& j)
-                {
-                    j.insert(at_end ? j.end() : j.begin() + 1, 3, val);
-                });
-            }
-
-            SECTION("insert(pos, const_iterator, const_iterator)")
-            {
-                check_exception_guarantee<json_t>(guarantee::basic, make, [&](json_t& j)
-                {
-                    j.insert(at_end ? j.end() : j.begin() + 1, source.begin(), source.end());
-                });
-            }
-
-            SECTION("insert(pos, initializer_list_t)")
-            {
-                check_exception_guarantee<json_t>(guarantee::basic, make, [&](json_t& j)
-                {
-                    j.insert(at_end ? j.end() : j.begin() + 1, {val, val});
-                });
-            }
-        }
-    }
-
-    SECTION("insert(const_iterator, const_iterator) for objects")
-    {
-        const json_t members = {{"m", std::string(64, 'm')}, {"n", std::string(64, 'n')}, {"o", std::string(64, 'o')}};
-
-        SECTION("no member exists yet")
-        {
-            check_exception_guarantee<json_t>(guarantee::basic, []()
-            {
-                return json_t({{"a", 1}});
-            }, [&](json_t& j)
-            {
-                j.insert(members.begin(), members.end());
-            });
-        }
-
-        SECTION("some members exist already")
-        {
-            check_exception_guarantee<json_t>(guarantee::basic, []()
-            {
-                return json_t({{"a", 1}, {"n", std::string(64, 'x')}});
-            }, [&](json_t& j)
-            {
-                j.insert(members.begin(), members.end());
-            });
-        }
-    }
-}
-} // namespace
-
-TEST_CASE("insert() provides its documented exception guarantees (#5986)")
-{
-    SECTION("std::map-backed object_t")
-    {
-        check_insert_exception_guarantees<nlohmann::json::with_allocator_t<nth_alloc_fails_allocator>>();
-    }
-
-    SECTION("ordered_map-backed object_t")
-    {
-        check_insert_exception_guarantees<nlohmann::ordered_json::with_allocator_t<nth_alloc_fails_allocator>>();
-    }
-}
-#endif

tests/src/unit-bson.cpp

@@ -23,6 +23,7 @@ using nlohmann::json;
 #include <cstdint>
 #include <fstream>
 #include <limits>
+#include <list>
 #include <sstream>
 #include <vector>
 #include "make_test_data_available.hpp"
@@ -1199,6 +1200,31 @@ TEST_CASE("BSON keys from contiguous and stream input")
             from_stream = e.what();
         }
         CHECK(from_vector == from_stream);
+
+        // error_handler_t::strict checks the array element names in place on
+        // contiguous input and byte by byte otherwise; the names are
+        // well-formed, so both give the same value or error as above
+        std::string strict_from_vector;
+        std::string strict_from_list;
+        try
+        {
+            strict_from_vector = json::from_bson(input, true, true, json::error_handler_t::strict).dump();
+        }
+        catch (const json::parse_error& e)
+        {
+            strict_from_vector = e.what();
+        }
+        try
+        {
+            const std::list<std::uint8_t> list(input.begin(), input.end());
+            strict_from_list = json::from_bson(list.begin(), list.end(), true, true, json::error_handler_t::strict).dump();
+        }
+        catch (const json::parse_error& e)
+        {
+            strict_from_list = e.what();
+        }
+        CHECK(strict_from_vector == from_vector);
+        CHECK(strict_from_list == from_vector);
     }
 }
 
@@ -1227,6 +1253,18 @@ TEST_CASE("BSON array element names")
 
     std::istringstream stream(std::string(reinterpret_cast<const char*>(bson.data()), bson.size()));
     CHECK(json::from_bson(stream) == expected);
+
+    // input that is not contiguous reads the names byte by byte and must reject
+    // the same name
+    const std::list<std::uint8_t> list(bson.begin(), bson.end());
+    CHECK_THROWS_WITH_AS(_ = json::from_bson(list.begin(), list.end(), true, true, json::error_handler_t::strict), "[json.exception.parse_error.113] parse error at byte 14: syntax error while parsing BSON key: invalid string: ill-formed UTF-8 byte", json::parse_error&);
+
+    // without exceptions, the rejected name ends the parse
+    CHECK(json::from_bson(bson, true, false, json::error_handler_t::strict).is_discarded());
+
+    // well-formed names pass the check without being stored
+    const json valid = {{"a", {true, false, {{"k", json::array({1, 2})}}}}};
+    CHECK(json::from_bson(json::to_bson(valid), true, true, json::error_handler_t::strict) == valid);
 }
 #endif
 

tests/src/unit-cbor.cpp

@@ -2520,6 +2520,34 @@ TEST_CASE("CBOR arrays of scalars from contiguous and non-contiguous input")
     }
 }
 
+TEST_CASE("CBOR arrays of one element from contiguous and non-contiguous input")
+{
+    // contiguous input leaves an array of one element to the general path
+    // rather than to the loop that reads arrays of scalars; both inputs must
+    // still give the same events, value, or error for the complete document
+    // and for every truncation of it, with enough bytes after each array that
+    // the loop could have been entered
+    const json j = json::parse(R"({"a": [[[[1.5]]]], "b": [7], "c": [{"x": 2.5}], "d": ["s"], "e": [[1, 2]], "f": [null], "g": 1.25})");
+    const std::vector<std::uint8_t> bytes = json::to_cbor(j);
+
+    CHECK(json::from_cbor(bytes) == j);
+
+    for (std::size_t length = 0; length <= bytes.size(); ++length)
+    {
+        CAPTURE(length)
+        const std::vector<std::uint8_t> contiguous(bytes.begin(), bytes.begin() + static_cast<std::ptrdiff_t>(length));
+        const std::list<std::uint8_t> bytewise(contiguous.begin(), contiguous.end());
+
+        CHECK(json::from_cbor(contiguous, true, false).dump() == json::from_cbor(bytewise.begin(), bytewise.end(), true, false).dump());
+
+        SaxEventLogger from_contiguous;
+        SaxEventLogger from_bytewise;
+        CHECK(json::sax_parse(contiguous, &from_contiguous, json::input_format_t::cbor, true, false)
+              == json::sax_parse(bytewise.begin(), bytewise.end(), &from_bytewise, json::input_format_t::cbor, true, false));
+        CHECK(from_contiguous.events == from_bytewise.events);
+    }
+}
+
 TEST_CASE("CBOR indefinite-length strings do not recurse per chunk")
 {
     // Reading an indefinite-length string or byte array used to call itself

tests/src/unit-msgpack.cpp

@@ -2311,3 +2311,31 @@ TEST_CASE("MessagePack arrays of scalars from contiguous and non-contiguous inpu
               == json::sax_parse(bytewise.begin(), bytewise.end(), &from_bytewise, json::input_format_t::msgpack));
     }
 }
+
+TEST_CASE("MessagePack arrays of one element from contiguous and non-contiguous input")
+{
+    // contiguous input leaves an array of one element to the general path
+    // rather than to the loop that reads arrays of scalars; both inputs must
+    // still give the same events, value, or error for the complete document
+    // and for every truncation of it, with enough bytes after each array that
+    // the loop could have been entered
+    const json j = json::parse(R"({"a": [[[[1.5]]]], "b": [7], "c": [{"x": 2.5}], "d": ["s"], "e": [[1, 2]], "f": [null], "g": 1.25})");
+    const std::vector<std::uint8_t> bytes = json::to_msgpack(j);
+
+    CHECK(json::from_msgpack(bytes) == j);
+
+    for (std::size_t length = 0; length <= bytes.size(); ++length)
+    {
+        CAPTURE(length)
+        const std::vector<std::uint8_t> contiguous(bytes.begin(), bytes.begin() + static_cast<std::ptrdiff_t>(length));
+        const std::list<std::uint8_t> bytewise(contiguous.begin(), contiguous.end());
+
+        CHECK(json::from_msgpack(contiguous, true, false).dump() == json::from_msgpack(bytewise.begin(), bytewise.end(), true, false).dump());
+
+        SaxEventLogger from_contiguous;
+        SaxEventLogger from_bytewise;
+        CHECK(json::sax_parse(contiguous, &from_contiguous, json::input_format_t::msgpack, true, false)
+              == json::sax_parse(bytewise.begin(), bytewise.end(), &from_bytewise, json::input_format_t::msgpack, true, false));
+        CHECK(from_contiguous.events == from_bytewise.events);
+    }
+}