Document why the digit-count fast path is safe regardless of number_unsigned_t/number_integer_t width

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

include/nlohmann/detail/input/lexer.hpp

@@ -149,10 +149,11 @@ class lexer : public lexer_base<BasicJsonType>
   public:
     using token_type = typename lexer_base<BasicJsonType>::token_type;
 
-    explicit lexer(InputAdapterType&& adapter, bool ignore_comments_ = false) noexcept
+    explicit lexer(InputAdapterType&& adapter, bool ignore_comments_ = false, bool discard_number_values_ = false) noexcept
         : ia(std::move(adapter))
         , ignore_comments(ignore_comments_)
         , decimal_point_char(static_cast<char_int_type>(get_decimal_point()))
+        , discard_number_values(discard_number_values_)
     {}
 
     // deleted because of pointer members
@@ -1279,6 +1280,58 @@ scan_number_done:
         // we are done scanning a number)
         unget();
 
+        // If the caller does not need the converted value (only whether the
+        // input is syntactically valid; see json_sax_acceptor/accept()), an
+        // unsigned/integer token can be reported without calling
+        // strtoull()/strtoll() at all, *provided* we can already tell from
+        // the digit count alone that the conversion cannot overflow 64 bits.
+        // Such tokens are always finite and are accepted unconditionally by
+        // the parser regardless of their actual value (parser::sax_parse_internal()
+        // never checks finiteness for value_unsigned/value_integer), so the
+        // classification below is all that is needed.
+        //
+        // A decimal number with up to 18 digits is always representable in
+        // both std::uint64_t and std::int64_t (18 nines is ~1e18, well below
+        // both UINT64_MAX ~1.8e19 and INT64_MAX ~9.2e18), so strtoull()/strtoll()
+        // could not have set errno to ERANGE for it. Numbers with more digits
+        // (rare in practice) fall through to the exact code below, unchanged,
+        // so their handling -- including reclassification to value_float when
+        // the value overflows 64 bits, and rejection when it is not even
+        // finite as a double -- is bit-for-bit identical to before this
+        // optimization.
+        //
+        // Note this reasons about std::uint64_t/std::int64_t, not about
+        // number_unsigned_t/number_integer_t (BasicJsonType's own, possibly
+        // narrower, template parameters -- e.g. std::uint32_t). That is fine
+        // *only* because discard_number_values is exclusively set by
+        // accept() (see json.hpp), and accept() always parses through the
+        // library's own json_sax_acceptor -- never a user-supplied SAX
+        // consumer -- whose number_unsigned()/number_integer()/number_float()
+        // callbacks unconditionally discard their argument and return true.
+        // So for every caller that can reach this branch, neither the token
+        // classification below nor the eventual (possibly narrowed, and on
+        // this fast path left stale/unset) value_unsigned/value_integer is
+        // ever consulted -- an unsigned/integer token is accepted outright,
+        // and even a >18-digit token that this fast path deliberately falls
+        // through for is, once reclassified to value_float, still finite
+        // (and thus accepted) for any digit count that fits in number_unsigned_t
+        // or number_integer_t regardless of that type's width. If this
+        // function is ever taught to run with discard_number_values true for
+        // a caller that *does* read the converted value, this reasoning (and
+        // the fast path below) would need to be revisited.
+        if (discard_number_values)
+        {
+            constexpr std::size_t safe_digit_count = 18;
+            if (number_type == token_type::value_unsigned && token_buffer.size() <= safe_digit_count)
+            {
+                return token_type::value_unsigned;
+            }
+            if (number_type == token_type::value_integer && token_buffer.size() - 1 <= safe_digit_count)
+            {
+                return token_type::value_integer;
+            }
+        }
+
         char* endptr = nullptr; // NOLINT(misc-const-correctness,cppcoreguidelines-pro-type-vararg,hicpp-vararg)
         errno = 0;
 
@@ -1754,6 +1807,13 @@ scan_number_done:
     const char_int_type decimal_point_char = '.';
     /// the position of the decimal point in the input
     std::size_t decimal_point_position = std::string::npos;
+
+    /// whether the caller (e.g. accept()/json_sax_acceptor) only needs the
+    /// token classification and never looks at the converted numeric value;
+    /// when set, scan_number() may skip strtoull()/strtoll() for
+    /// value_unsigned/value_integer tokens whose digit count guarantees they
+    /// fit into 64 bits (see scan_number())
+    const bool discard_number_values = false;
 };
 
 }  // namespace detail

include/nlohmann/detail/input/parser.hpp

@@ -72,9 +72,10 @@ class parser
                     parser_callback_t<BasicJsonType> cb = nullptr,
                     const bool allow_exceptions_ = true,
                     const bool ignore_comments = false,
-                    const bool ignore_trailing_commas_ = false)
+                    const bool ignore_trailing_commas_ = false,
+                    const bool discard_number_values_ = false)
         : callback(std::move(cb))
-        , m_lexer(std::move(adapter), ignore_comments)
+        , m_lexer(std::move(adapter), ignore_comments, discard_number_values_)
         , allow_exceptions(allow_exceptions_)
         , ignore_trailing_commas(ignore_trailing_commas_)
     {

include/nlohmann/detail/output/binary_writer.hpp

@@ -1647,20 +1647,6 @@ class binary_writer
         return 'D';  // float 64
     }
 
-    /*!
-    @brief checks whether a JSON number fits into @a TargetType
-    @param[in] el a JSON number of either the signed or unsigned integer kind
-    @return whether @a el's value can be represented by @a TargetType without
-            wrapping, regardless of which of the two kinds it is stored as
-    */
-    template<typename TargetType>
-    static bool bjdata_ndarray_value_in_range(const BasicJsonType& el)
-    {
-        return el.is_number_unsigned()
-               ? value_in_range_of<TargetType>(el.template get<std::uint64_t>())
-               : value_in_range_of<TargetType>(el.template get<std::int64_t>());
-    }
-
     /*!
     @return false if the object is successfully converted to a bjdata ndarray, true if the type or size is invalid
     */
@@ -1672,16 +1658,6 @@ class binary_writer
         };
 
         string_t key = "_ArrayType_";
-        // the type name is looked up as a string below; a non-string
-        // annotation (e.g. a number, null, or an array) cannot name a known
-        // dtype, so it is treated the same as an unrecognized type name and
-        // falls back to a plain object encoding instead of throwing
-        // type_error.302 out of get<string_t>()
-        if (!value.at(key).is_string())
-        {
-            return true;
-        }
-
         // use get<string_t>() instead of static_cast<string_t> to avoid an
         // ambiguous conversion under explicit instantiation on C++17 (see #4825)
         auto it = bjdtype.find(value.at(key).template get<string_t>());
@@ -1691,16 +1667,6 @@ class binary_writer
         }
         CharType dtype = it->second;
 
-        // the 'B' (byte) marker is only defined by BJData Draft 3; emitting it
-        // under the default Draft 2 mode would produce a stream that Draft 2
-        // readers reject, so such an object falls back to a plain object
-        // encoding instead (see the "Binary values" section of the BJData
-        // documentation)
-        if (dtype == 'B' && bjdata_version != bjdata_version_t::draft3)
-        {
-            return true;
-        }
-
         key = "_ArraySize_";
         // the dimensions are written verbatim as the header length below, so a
         // value that is not an array cannot produce a valid one: null emits 'Z'
@@ -1765,60 +1731,6 @@ class binary_writer
             }
         }
 
-        // every element is cast to the (possibly narrower) C++ type matching
-        // dtype below; a value that does not fit that type would silently
-        // wrap (integers) or overflow to infinity (the "single" precision
-        // float) instead of being reported, so such an object falls back to
-        // a plain object encoding as well
-        for (const auto& el : value.at(key))
-        {
-            bool in_range = true;
-            switch (dtype)
-            {
-                case 'U':
-                case 'C':
-                case 'B':
-                    in_range = bjdata_ndarray_value_in_range<std::uint8_t>(el);
-                    break;
-                case 'i':
-                    in_range = bjdata_ndarray_value_in_range<std::int8_t>(el);
-                    break;
-                case 'u':
-                    in_range = bjdata_ndarray_value_in_range<std::uint16_t>(el);
-                    break;
-                case 'I':
-                    in_range = bjdata_ndarray_value_in_range<std::int16_t>(el);
-                    break;
-                case 'm':
-                    in_range = bjdata_ndarray_value_in_range<std::uint32_t>(el);
-                    break;
-                case 'l':
-                    in_range = bjdata_ndarray_value_in_range<std::int32_t>(el);
-                    break;
-                case 'M':
-                    in_range = bjdata_ndarray_value_in_range<std::uint64_t>(el);
-                    break;
-                case 'L':
-                    in_range = bjdata_ndarray_value_in_range<std::int64_t>(el);
-                    break;
-                case 'd':
-                {
-                    const auto dval = el.template get<double>();
-                    in_range = !std::isfinite(dval) ||
-                               (dval >= static_cast<double>(std::numeric_limits<float>::lowest()) &&
-                                dval <= static_cast<double>((std::numeric_limits<float>::max)()));
-                    break;
-                }
-                default:
-                    // 'D' (double) already spans the full range of number_float_t
-                    break;
-            }
-            if (!in_range)
-            {
-                return true;
-            }
-        }
-
         oa->write_character('[');
         oa->write_character('$');
         oa->write_character(dtype);

include/nlohmann/json.hpp

@@ -164,11 +164,12 @@ class basic_json // NOLINT(cppcoreguidelines-special-member-functions,hicpp-spec
         detail::parser_callback_t<basic_json>cb = nullptr,
         const bool allow_exceptions = true,
         const bool ignore_comments = false,
-        const bool ignore_trailing_commas = false
+        const bool ignore_trailing_commas = false,
+        const bool discard_number_values = false
     )
     {
         return ::nlohmann::detail::parser<basic_json, InputAdapterType>(std::move(adapter),
-               std::move(cb), allow_exceptions, ignore_comments, ignore_trailing_commas);
+               std::move(cb), allow_exceptions, ignore_comments, ignore_trailing_commas, discard_number_values);
     }
 
   private:
@@ -4133,7 +4134,7 @@ class basic_json // NOLINT(cppcoreguidelines-special-member-functions,hicpp-spec
                        const bool ignore_comments = false,
                        const bool ignore_trailing_commas = false)
     {
-        return parser(detail::input_adapter(std::forward<InputType>(i)), nullptr, false, ignore_comments, ignore_trailing_commas).accept(true);
+        return parser(detail::input_adapter(std::forward<InputType>(i)), nullptr, false, ignore_comments, ignore_trailing_commas, true).accept(true);
     }
 
     /// @brief check if the input is valid JSON (iterator pair, or iterator+sentinel pair for C++20 ranges support)
@@ -4144,7 +4145,7 @@ class basic_json // NOLINT(cppcoreguidelines-special-member-functions,hicpp-spec
                        const bool ignore_comments = false,
                        const bool ignore_trailing_commas = false)
     {
-        return parser(detail::input_adapter(std::move(first), std::move(last)), nullptr, false, ignore_comments, ignore_trailing_commas).accept(true);
+        return parser(detail::input_adapter(std::move(first), std::move(last)), nullptr, false, ignore_comments, ignore_trailing_commas, true).accept(true);
     }
 
     JSON_HEDLEY_WARN_UNUSED_RESULT
@@ -4153,7 +4154,7 @@ class basic_json // NOLINT(cppcoreguidelines-special-member-functions,hicpp-spec
                        const bool ignore_comments = false,
                        const bool ignore_trailing_commas = false)
     {
-        return parser(i.get(), nullptr, false, ignore_comments, ignore_trailing_commas).accept(true);
+        return parser(i.get(), nullptr, false, ignore_comments, ignore_trailing_commas, true).accept(true);
     }
 
     /// @brief generate SAX events

single_include/nlohmann/json.hpp

@@ -7932,10 +7932,11 @@ class lexer : public lexer_base<BasicJsonType>
   public:
     using token_type = typename lexer_base<BasicJsonType>::token_type;
 
-    explicit lexer(InputAdapterType&& adapter, bool ignore_comments_ = false) noexcept
+    explicit lexer(InputAdapterType&& adapter, bool ignore_comments_ = false, bool discard_number_values_ = false) noexcept
         : ia(std::move(adapter))
         , ignore_comments(ignore_comments_)
         , decimal_point_char(static_cast<char_int_type>(get_decimal_point()))
+        , discard_number_values(discard_number_values_)
     {}
 
     // deleted because of pointer members
@@ -9062,6 +9063,58 @@ scan_number_done:
         // we are done scanning a number)
         unget();
 
+        // If the caller does not need the converted value (only whether the
+        // input is syntactically valid; see json_sax_acceptor/accept()), an
+        // unsigned/integer token can be reported without calling
+        // strtoull()/strtoll() at all, *provided* we can already tell from
+        // the digit count alone that the conversion cannot overflow 64 bits.
+        // Such tokens are always finite and are accepted unconditionally by
+        // the parser regardless of their actual value (parser::sax_parse_internal()
+        // never checks finiteness for value_unsigned/value_integer), so the
+        // classification below is all that is needed.
+        //
+        // A decimal number with up to 18 digits is always representable in
+        // both std::uint64_t and std::int64_t (18 nines is ~1e18, well below
+        // both UINT64_MAX ~1.8e19 and INT64_MAX ~9.2e18), so strtoull()/strtoll()
+        // could not have set errno to ERANGE for it. Numbers with more digits
+        // (rare in practice) fall through to the exact code below, unchanged,
+        // so their handling -- including reclassification to value_float when
+        // the value overflows 64 bits, and rejection when it is not even
+        // finite as a double -- is bit-for-bit identical to before this
+        // optimization.
+        //
+        // Note this reasons about std::uint64_t/std::int64_t, not about
+        // number_unsigned_t/number_integer_t (BasicJsonType's own, possibly
+        // narrower, template parameters -- e.g. std::uint32_t). That is fine
+        // *only* because discard_number_values is exclusively set by
+        // accept() (see json.hpp), and accept() always parses through the
+        // library's own json_sax_acceptor -- never a user-supplied SAX
+        // consumer -- whose number_unsigned()/number_integer()/number_float()
+        // callbacks unconditionally discard their argument and return true.
+        // So for every caller that can reach this branch, neither the token
+        // classification below nor the eventual (possibly narrowed, and on
+        // this fast path left stale/unset) value_unsigned/value_integer is
+        // ever consulted -- an unsigned/integer token is accepted outright,
+        // and even a >18-digit token that this fast path deliberately falls
+        // through for is, once reclassified to value_float, still finite
+        // (and thus accepted) for any digit count that fits in number_unsigned_t
+        // or number_integer_t regardless of that type's width. If this
+        // function is ever taught to run with discard_number_values true for
+        // a caller that *does* read the converted value, this reasoning (and
+        // the fast path below) would need to be revisited.
+        if (discard_number_values)
+        {
+            constexpr std::size_t safe_digit_count = 18;
+            if (number_type == token_type::value_unsigned && token_buffer.size() <= safe_digit_count)
+            {
+                return token_type::value_unsigned;
+            }
+            if (number_type == token_type::value_integer && token_buffer.size() - 1 <= safe_digit_count)
+            {
+                return token_type::value_integer;
+            }
+        }
+
         char* endptr = nullptr; // NOLINT(misc-const-correctness,cppcoreguidelines-pro-type-vararg,hicpp-vararg)
         errno = 0;
 
@@ -9537,6 +9590,13 @@ scan_number_done:
     const char_int_type decimal_point_char = '.';
     /// the position of the decimal point in the input
     std::size_t decimal_point_position = std::string::npos;
+
+    /// whether the caller (e.g. accept()/json_sax_acceptor) only needs the
+    /// token classification and never looks at the converted numeric value;
+    /// when set, scan_number() may skip strtoull()/strtoll() for
+    /// value_unsigned/value_integer tokens whose digit count guarantees they
+    /// fit into 64 bits (see scan_number())
+    const bool discard_number_values = false;
 };
 
 }  // namespace detail
@@ -14043,9 +14103,10 @@ class parser
                     parser_callback_t<BasicJsonType> cb = nullptr,
                     const bool allow_exceptions_ = true,
                     const bool ignore_comments = false,
-                    const bool ignore_trailing_commas_ = false)
+                    const bool ignore_trailing_commas_ = false,
+                    const bool discard_number_values_ = false)
         : callback(std::move(cb))
-        , m_lexer(std::move(adapter), ignore_comments)
+        , m_lexer(std::move(adapter), ignore_comments, discard_number_values_)
         , allow_exceptions(allow_exceptions_)
         , ignore_trailing_commas(ignore_trailing_commas_)
     {
@@ -18655,20 +18716,6 @@ class binary_writer
         return 'D';  // float 64
     }
 
-    /*!
-    @brief checks whether a JSON number fits into @a TargetType
-    @param[in] el a JSON number of either the signed or unsigned integer kind
-    @return whether @a el's value can be represented by @a TargetType without
-            wrapping, regardless of which of the two kinds it is stored as
-    */
-    template<typename TargetType>
-    static bool bjdata_ndarray_value_in_range(const BasicJsonType& el)
-    {
-        return el.is_number_unsigned()
-               ? value_in_range_of<TargetType>(el.template get<std::uint64_t>())
-               : value_in_range_of<TargetType>(el.template get<std::int64_t>());
-    }
-
     /*!
     @return false if the object is successfully converted to a bjdata ndarray, true if the type or size is invalid
     */
@@ -18680,16 +18727,6 @@ class binary_writer
         };
 
         string_t key = "_ArrayType_";
-        // the type name is looked up as a string below; a non-string
-        // annotation (e.g. a number, null, or an array) cannot name a known
-        // dtype, so it is treated the same as an unrecognized type name and
-        // falls back to a plain object encoding instead of throwing
-        // type_error.302 out of get<string_t>()
-        if (!value.at(key).is_string())
-        {
-            return true;
-        }
-
         // use get<string_t>() instead of static_cast<string_t> to avoid an
         // ambiguous conversion under explicit instantiation on C++17 (see #4825)
         auto it = bjdtype.find(value.at(key).template get<string_t>());
@@ -18699,16 +18736,6 @@ class binary_writer
         }
         CharType dtype = it->second;
 
-        // the 'B' (byte) marker is only defined by BJData Draft 3; emitting it
-        // under the default Draft 2 mode would produce a stream that Draft 2
-        // readers reject, so such an object falls back to a plain object
-        // encoding instead (see the "Binary values" section of the BJData
-        // documentation)
-        if (dtype == 'B' && bjdata_version != bjdata_version_t::draft3)
-        {
-            return true;
-        }
-
         key = "_ArraySize_";
         // the dimensions are written verbatim as the header length below, so a
         // value that is not an array cannot produce a valid one: null emits 'Z'
@@ -18773,60 +18800,6 @@ class binary_writer
             }
         }
 
-        // every element is cast to the (possibly narrower) C++ type matching
-        // dtype below; a value that does not fit that type would silently
-        // wrap (integers) or overflow to infinity (the "single" precision
-        // float) instead of being reported, so such an object falls back to
-        // a plain object encoding as well
-        for (const auto& el : value.at(key))
-        {
-            bool in_range = true;
-            switch (dtype)
-            {
-                case 'U':
-                case 'C':
-                case 'B':
-                    in_range = bjdata_ndarray_value_in_range<std::uint8_t>(el);
-                    break;
-                case 'i':
-                    in_range = bjdata_ndarray_value_in_range<std::int8_t>(el);
-                    break;
-                case 'u':
-                    in_range = bjdata_ndarray_value_in_range<std::uint16_t>(el);
-                    break;
-                case 'I':
-                    in_range = bjdata_ndarray_value_in_range<std::int16_t>(el);
-                    break;
-                case 'm':
-                    in_range = bjdata_ndarray_value_in_range<std::uint32_t>(el);
-                    break;
-                case 'l':
-                    in_range = bjdata_ndarray_value_in_range<std::int32_t>(el);
-                    break;
-                case 'M':
-                    in_range = bjdata_ndarray_value_in_range<std::uint64_t>(el);
-                    break;
-                case 'L':
-                    in_range = bjdata_ndarray_value_in_range<std::int64_t>(el);
-                    break;
-                case 'd':
-                {
-                    const auto dval = el.template get<double>();
-                    in_range = !std::isfinite(dval) ||
-                               (dval >= static_cast<double>(std::numeric_limits<float>::lowest()) &&
-                                dval <= static_cast<double>((std::numeric_limits<float>::max)()));
-                    break;
-                }
-                default:
-                    // 'D' (double) already spans the full range of number_float_t
-                    break;
-            }
-            if (!in_range)
-            {
-                return true;
-            }
-        }
-
         oa->write_character('[');
         oa->write_character('$');
         oa->write_character(dtype);
@@ -21680,11 +21653,12 @@ class basic_json // NOLINT(cppcoreguidelines-special-member-functions,hicpp-spec
         detail::parser_callback_t<basic_json>cb = nullptr,
         const bool allow_exceptions = true,
         const bool ignore_comments = false,
-        const bool ignore_trailing_commas = false
+        const bool ignore_trailing_commas = false,
+        const bool discard_number_values = false
                                  )
     {
         return ::nlohmann::detail::parser<basic_json, InputAdapterType>(std::move(adapter),
-            std::move(cb), allow_exceptions, ignore_comments, ignore_trailing_commas);
+            std::move(cb), allow_exceptions, ignore_comments, ignore_trailing_commas, discard_number_values);
     }
 
   private:
@@ -25649,7 +25623,7 @@ class basic_json // NOLINT(cppcoreguidelines-special-member-functions,hicpp-spec
                        const bool ignore_comments = false,
                        const bool ignore_trailing_commas = false)
     {
-        return parser(detail::input_adapter(std::forward<InputType>(i)), nullptr, false, ignore_comments, ignore_trailing_commas).accept(true);
+        return parser(detail::input_adapter(std::forward<InputType>(i)), nullptr, false, ignore_comments, ignore_trailing_commas, true).accept(true);
     }
 
     /// @brief check if the input is valid JSON (iterator pair, or iterator+sentinel pair for C++20 ranges support)
@@ -25660,7 +25634,7 @@ class basic_json // NOLINT(cppcoreguidelines-special-member-functions,hicpp-spec
                        const bool ignore_comments = false,
                        const bool ignore_trailing_commas = false)
     {
-        return parser(detail::input_adapter(std::move(first), std::move(last)), nullptr, false, ignore_comments, ignore_trailing_commas).accept(true);
+        return parser(detail::input_adapter(std::move(first), std::move(last)), nullptr, false, ignore_comments, ignore_trailing_commas, true).accept(true);
     }
 
     JSON_HEDLEY_WARN_UNUSED_RESULT
@@ -25669,7 +25643,7 @@ class basic_json // NOLINT(cppcoreguidelines-special-member-functions,hicpp-spec
                        const bool ignore_comments = false,
                        const bool ignore_trailing_commas = false)
     {
-        return parser(i.get(), nullptr, false, ignore_comments, ignore_trailing_commas).accept(true);
+        return parser(i.get(), nullptr, false, ignore_comments, ignore_trailing_commas, true).accept(true);
     }
 
     /// @brief generate SAX events

tests/src/unit-bjdata.cpp

@@ -2586,12 +2586,7 @@ TEST_CASE("BJData")
                 CHECK(json::to_bjdata(json::from_bjdata(v_d), true, true) == v_d);
                 CHECK(json::to_bjdata(json::from_bjdata(v_D), true, true) == v_D);
                 CHECK(json::to_bjdata(json::from_bjdata(v_C), true, true) == v_C);
-                // v_B uses the Draft-3-only 'B' marker, so it round-trips only when
-                // Draft 3 is explicitly selected (see GitHub issue #5404); the
-                // default Draft 2 falls back to a plain object instead, covered by
-                // the "ndarray with _ArrayType_ "byte" is gated by the BJData draft
-                // version" section below
-                CHECK(json::to_bjdata(json::from_bjdata(v_B), true, true, json::bjdata_version_t::draft3) == v_B);
+                CHECK(json::to_bjdata(json::from_bjdata(v_B), true, true) == v_B);
             }
 
             SECTION("ndarray with data not matching _ArrayType_ is written as an object")
@@ -2634,10 +2629,8 @@ TEST_CASE("BJData")
                 // the C++ API stores an int literal as number_integer, so _ArrayType_
                 // names the wire type rather than the storage. Both storages have to
                 // produce the same typed array for every type.
-                // "byte" is checked separately below since it additionally requires
-                // BJData Draft 3 to be selected explicitly (see GitHub issue #5404).
                 for (const char* type :
-                        {"uint8", "int8", "uint16", "int16", "uint32", "int32", "uint64", "int64", "char"
+                        {"uint8", "int8", "uint16", "int16", "uint32", "int32", "uint64", "int64", "char", "byte"
                         })
                 {
                     CAPTURE(type);
@@ -2648,14 +2641,6 @@ TEST_CASE("BJData")
                     CHECK(from_text == json::to_bjdata(json({{"_ArrayType_", type}, {"_ArraySize_", {2, 3}}, {"_ArrayData_", {1, 2, 3, 4, 5, 6}}})));
                 }
 
-                {
-                    const std::string text = R"({"_ArrayType_":"byte","_ArraySize_":[2,3],"_ArrayData_":[1,2,3,4,5,6]})";
-                    const auto from_text = json::to_bjdata(json::parse(text), true, true, json::bjdata_version_t::draft3);
-                    CHECK(from_text.at(0) == '[');
-                    CHECK(from_text == json::to_bjdata(json({{"_ArrayType_", "byte"}, {"_ArraySize_", {2, 3}}, {"_ArrayData_", {1, 2, 3, 4, 5, 6}}}),
-                    true, true, json::bjdata_version_t::draft3));
-                }
-
                 // negative values under a signed type behave the same way
                 const auto from_neg = json::to_bjdata(json::parse(R"({"_ArrayType_":"int32","_ArraySize_":[2],"_ArrayData_":[-5,7]})"));
                 CHECK(from_neg.at(0) == '[');
@@ -2746,39 +2731,6 @@ TEST_CASE("BJData")
                 CHECK(json::from_bjdata(json::to_bjdata(j_size), true, true) == j_size);
             }
 
-            SECTION("ndarray whose _ArrayType_ is not a string stays as object")
-            {
-                // the type name is looked up as a string below the annotation
-                // check; a non-string _ArrayType_ cannot name a known dtype,
-                // so calling get<string_t>() on it would throw type_error.302
-                // instead of falling back like an unrecognized type name
-                // already does (see GitHub issue #5398)
-                json const j_number = json({{"_ArrayType_", 1}, {"_ArraySize_", {2}}, {"_ArrayData_", {1, 2}}});
-                const auto out_number = json::to_bjdata(j_number);
-                CHECK(out_number.at(0) == '{');
-                CHECK(json::from_bjdata(out_number) == j_number);
-
-                json const j_null = json({{"_ArrayType_", nullptr}, {"_ArraySize_", {2}}, {"_ArrayData_", {1, 2}}});
-                const auto out_null = json::to_bjdata(j_null);
-                CHECK(out_null.at(0) == '{');
-                CHECK(json::from_bjdata(out_null) == j_null);
-
-                json const j_bool = json({{"_ArrayType_", true}, {"_ArraySize_", {2}}, {"_ArrayData_", {1, 2}}});
-                const auto out_bool = json::to_bjdata(j_bool);
-                CHECK(out_bool.at(0) == '{');
-                CHECK(json::from_bjdata(out_bool) == j_bool);
-
-                json const j_array = json({{"_ArrayType_", {"uint8"}}, {"_ArraySize_", {2}}, {"_ArrayData_", {1, 2}}});
-                const auto out_array = json::to_bjdata(j_array);
-                CHECK(out_array.at(0) == '{');
-                CHECK(json::from_bjdata(out_array) == j_array);
-
-                json const j_object = json({{"_ArrayType_", {{"a", 1}}}, {"_ArraySize_", {2}}, {"_ArrayData_", {1, 2}}});
-                const auto out_object = json::to_bjdata(j_object);
-                CHECK(out_object.at(0) == '{');
-                CHECK(json::from_bjdata(out_object) == j_object);
-            }
-
             SECTION("ndarray whose dimensions overflow stays as object")
             {
                 // the product of the dimensions wraps around std::size_t to 0
@@ -2824,83 +2776,6 @@ TEST_CASE("BJData")
                 CHECK(out_num.at(0) == '{');
                 CHECK(json::from_bjdata(out_num) == j_num);
             }
-
-            SECTION("ndarray with out-of-range _ArrayData_ elements stays as object")
-            {
-                // each element is cast to the (possibly narrower) C++ type
-                // named by _ArrayType_ before being written; a value that
-                // does not fit that type would silently wrap instead of
-                // being reported, so such an object falls back to a plain
-                // object encoding that still round-trips (see GitHub issue #5403)
-
-                // an unsigned element that does not fit uint8
-                json const j_uint8 = json({{"_ArrayType_", "uint8"}, {"_ArraySize_", {2}}, {"_ArrayData_", {1, 256}}});
-                const auto out_uint8 = json::to_bjdata(j_uint8);
-                CHECK(out_uint8.at(0) == '{');
-                CHECK(json::from_bjdata(out_uint8) == j_uint8);
-
-                // a signed element that does not fit int8
-                json const j_int8 = json({{"_ArrayType_", "int8"}, {"_ArraySize_", {2}}, {"_ArrayData_", {1, 200}}});
-                const auto out_int8 = json::to_bjdata(j_int8);
-                CHECK(out_int8.at(0) == '{');
-                CHECK(json::from_bjdata(out_int8) == j_int8);
-
-                // a negative element is likewise out of range for an
-                // unsigned _ArrayType_
-                json const j_uint16_neg = json({{"_ArrayType_", "uint16"}, {"_ArraySize_", {2}}, {"_ArrayData_", {1, -1}}});
-                const auto out_uint16_neg = json::to_bjdata(j_uint16_neg);
-                CHECK(out_uint16_neg.at(0) == '{');
-                CHECK(json::from_bjdata(out_uint16_neg) == j_uint16_neg);
-
-                // a double element that overflows to infinity when narrowed
-                // to the "single" (float) precision named by _ArrayType_
-                json const j_single = json({{"_ArrayType_", "single"}, {"_ArraySize_", {2}}, {"_ArrayData_", {1.5, 1e40}}});
-                const auto out_single = json::to_bjdata(j_single);
-                CHECK(out_single.at(0) == '{');
-                CHECK(json::from_bjdata(out_single) == j_single);
-
-                // in-range boundary values still use the compact ndarray encoding
-                json const j_uint8_ok = json({{"_ArrayType_", "uint8"}, {"_ArraySize_", {2}}, {"_ArrayData_", {0, 255}}});
-                CHECK(json::to_bjdata(j_uint8_ok) == std::vector<uint8_t>({'[', '$', 'U', '#', '[', 'i', 2, ']', 0, 255}));
-
-                json const j_int8_ok = json({{"_ArrayType_", "int8"}, {"_ArraySize_", {2}}, {"_ArrayData_", {-128, 127}}});
-                CHECK(json::to_bjdata(j_int8_ok) == std::vector<uint8_t>({'[', '$', 'i', '#', '[', 'i', 2, ']', 0x80, 0x7F}));
-
-                json const j_single_ok = json({{"_ArrayType_", "single"}, {"_ArraySize_", {1}}, {"_ArrayData_", {1.5}}});
-                const auto out_single_ok = json::to_bjdata(j_single_ok);
-                CHECK(out_single_ok.at(0) == '[');
-                CHECK(json::from_bjdata(out_single_ok) == json({1.5f}));
-            }
-
-            SECTION("ndarray with _ArrayType_ \"byte\" is gated by the BJData draft version")
-            {
-                // the 'B' (byte) marker used by _ArrayType_ "byte" is only defined
-                // by BJData Draft 3; Draft 2 (the default) has no such marker, so
-                // emitting it unconditionally produced a stream that a Draft 2
-                // reader could not parse as intended (see GitHub issue #5404).
-                // Two dimensions are used so that a successfully written ndarray
-                // round-trips back into the annotated object (a single dimension
-                // is, by the BJData ndarray convention, read back as a plain
-                // binary value rather than the annotated object, same as every
-                // other single-dimension ndarray of a non-"byte" type is read
-                // back as a plain array instead of the annotated object).
-                json const j_byte = json({{"_ArrayType_", "byte"}, {"_ArraySize_", {2, 3}}, {"_ArrayData_", {1, 2, 3, 4, 5, 6}}});
-
-                // default (Draft 2): falls back to a plain object and round-trips
-                const auto out_draft2 = json::to_bjdata(j_byte);
-                CHECK(out_draft2.at(0) == '{');
-                CHECK(json::from_bjdata(out_draft2) == j_byte);
-
-                // explicit Draft 2: same as the default
-                const auto out_draft2_explicit = json::to_bjdata(j_byte, true, true, json::bjdata_version_t::draft2);
-                CHECK(out_draft2_explicit.at(0) == '{');
-                CHECK(json::from_bjdata(out_draft2_explicit) == j_byte);
-
-                // Draft 3 explicitly selected: still uses the compact 'B' ndarray encoding
-                const auto out_draft3 = json::to_bjdata(j_byte, true, true, json::bjdata_version_t::draft3);
-                CHECK(out_draft3 == std::vector<uint8_t>({'[', '$', 'B', '#', '[', '$', 'i', '#', 'i', 2, 2, 3, 1, 2, 3, 4, 5, 6}));
-                CHECK(json::from_bjdata(out_draft3) == j_byte);
-            }
         }
     }
 

tests/src/unit-class_parser.cpp

@@ -930,6 +930,94 @@ TEST_CASE("parser class")
                 CHECK(accept_helper("+1") == false);
                 CHECK(accept_helper("+0") == false);
             }
+
+            SECTION("issue #5411 - skip conversion when accept() does not need the numeric value")
+            {
+                // lexer::scan_number() may skip strtoull()/strtoll() for
+                // value_unsigned/value_integer tokens when the caller (e.g.
+                // json::accept()) does not need the converted value, as long
+                // as the digit count alone guarantees no 64-bit overflow (see
+                // the "safe_digit_count" fast path in scan_number()). This
+                // differential test checks that json::accept() (which enables
+                // the fast path) and json::parse() (which never does) always
+                // agree, over a corpus that exercises both the fast path
+                // (<=18 digits) and the untouched, exact fallback path (>=19
+                // digits) -- including reclassification of huge digit-only
+                // integers to a (possibly non-finite) floating-point value.
+                const std::vector<std::pair<std::string, bool>> cases =
+                {
+                    // normal small/large integers, both signs
+                    {"0", true}, {"1", true}, {"-1", true}, {"42", true}, {"-42", true},
+                    {"123456789", true}, {"-123456789", true},
+
+                    // digit-count boundary around the 18-digit safe cutoff (both signs)
+                    {std::string(17, '9'), true},
+                    {std::string(18, '9'), true},
+                    {std::string(19, '9'), true},
+                    {std::string(20, '9'), true},
+                    {"-" + std::string(17, '9'), true},
+                    {"-" + std::string(18, '9'), true},
+                    {"-" + std::string(19, '9'), true},
+                    {"-" + std::string(20, '9'), true},
+
+                    // 64-bit boundaries
+                    {"9223372036854775807", true},    // INT64_MAX
+                    {"-9223372036854775808", true},   // INT64_MIN
+                    {"18446744073709551615", true},   // UINT64_MAX
+                    {"18446744073709551616", true},   // UINT64_MAX + 1 (overflows uint64_t, finite double)
+
+                    // the 28-digit example from the issue: overflows uint64_t
+                    // but is finite as a double, so the scanner reclassifies
+                    // it to value_float and it is accepted
+                    {"9999999999999999999999999999", true},
+
+                    // huge digit-only integers that overflow even a double -> rejected
+                    {std::string(309, '9'), false},
+                    {std::string(400, '9'), false},
+                    {"1" + std::string(400, '0'), false},
+
+                    // 1e999 / 1e400 style overflow -> rejected
+                    {"1e999", false},
+                    {"1e400", false},
+                    {"-1e999", false},
+                    {"1E999", false},
+
+                    // values straddling DBL_MAX
+                    {"1.7976931348623157e308", true},   // <= DBL_MAX, finite
+                    {"1.7976931348623159e308", false},  // > DBL_MAX, overflows to inf
+
+                    // a mix of other valid/invalid numeric syntax
+                    {"3.14159", true},
+                    {"-0.0", true},
+                    {"1.0e10", true},
+                    {"01", false},
+                    {"-", false},
+                    {"1.", false},
+                    {"1e", false},
+                    {"+1", false},
+                };
+
+                for (const auto& c : cases)
+                {
+                    const std::string& number = c.first;
+                    const bool expected = c.second;
+                    CAPTURE(number)
+                    CAPTURE(expected)
+
+                    // accept() takes the fast path (skips conversion when possible)
+                    CHECK(json::accept(number) == expected);
+
+                    // parse() always performs the full conversion; it must agree
+                    json j;
+                    CHECK_NOTHROW(json::parser(nlohmann::detail::input_adapter(number), nullptr, false).parse(true, j));
+                    CHECK(!j.is_discarded() == expected);
+
+                    // wrap in an array so get_token() is exercised beyond the
+                    // very first (constructor-time) scan as well
+                    const std::string wrapped = "[" + number + "," + number + "]";
+                    CHECK(json::accept(wrapped) == expected);
+                }
+            }
         }
     }