Binary readers: swap the bytes of floats with a single byteswap (#5976)

* Binary readers: swap the bytes of floats with a single byteswap

byte_swap reversed float and double values byte by byte. GCC does not
merge that loop into one bswap, and C++23's std::byteswap accepts only
integers. Overloads for float and double now copy the bits into a
uint32_t/uint64_t, call detail::byteswap, and copy them back. This
applies to every binary reader that uses get_number.

In the CBOR reader, the single- and double-precision float cases call a
non-inlined helper. Without it, GCC stopped inlining the stream
adapter's get_character() into parse_cbor_internal, which made reading
from a std::istream about 0.5 % slower.

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

* CBOR reader: inline get_cbor_float() and always inline get() instead

The never-inlined get_cbor_float() costs a call for every float read
from input without bulk access: reading canada.json from CBOR through a
FILE* became 5.6 % slower on Apple clang in the benchmark suite. The
helper existed only because GCC 13 stopped inlining the stream adapter's
get_character() into parse_cbor_internal. Always inlining
binary_reader::get() fixes that as well.

callgrind, all binary formats on buffer, std::istringstream and FILE*
input (canada and citm_catalog): GCC 13 0.986-1.003, GCC 16 0.994-1.006,
Clang 23 identical. Apple clang suite: FromBinaryFile/cbor/canada 0.939,
geomean of the 53 binary reader rows 0.998.

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

---------

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

include/nlohmann/detail/input/binary_reader.hpp

@@ -22,6 +22,7 @@
     #include <bit>  //byteswap
 #endif
 
+#include <nlohmann/detail/bit_ops.hpp>
 #include <nlohmann/detail/exceptions.hpp>
 #include <nlohmann/detail/input/input_adapters.hpp>
 #include <nlohmann/detail/input/json_sax.hpp>
@@ -1139,16 +1140,10 @@ class binary_reader
                 return get_half_float(false);
 
             case 0xFA: // Single-Precision Float (four-byte IEEE 754)
-            {
-                float number{};
-                return get_number(number) && emit_float(number);
-            }
+                return get_cbor_float<float>();
 
             case 0xFB: // Double-Precision Float (eight-byte IEEE 754)
-            {
-                double number{};
-                return get_number(number) && emit_float(number);
-            }
+                return get_cbor_float<double>();
 
             default: // anything else (0xFF is handled inside the other types)
             {
@@ -1157,6 +1152,14 @@ class binary_reader
         }
     }
 
+    /// read a single- or double-precision float
+    template<typename NumberType>
+    bool get_cbor_float()
+    {
+        NumberType number{};
+        return get_number(number) && emit_float(number);
+    }
+
     /*!
     @brief reports a nested indefinite-length CBOR string or byte array
     @param[in] type_name  name of the rejected string type
@@ -4176,7 +4179,12 @@ class binary_reader
     `char_traits<char_type>::eof()` in that case.
 
     @return character read from the input
+
+    @note always inlined: GCC otherwise stops inlining the stream adapter's
+          get_character() into parse_cbor_internal once that function grows
+          (+0.45 % instructions for CBOR from a std::istringstream with GCC 13)
     */
+    JSON_HEDLEY_ALWAYS_INLINE
     char_int_type get()
     {
         ++chars_read;
@@ -4222,6 +4230,27 @@ class binary_reader
         return current;
     }
 
+    /// reverse the byte order of a double with a single byteswap(): GCC does
+    /// not merge the byte-wise swap of the generic version into one
+    static void byte_swap(double& number) noexcept
+    {
+        static_assert(sizeof(double) == sizeof(std::uint64_t), "double must have 64 bits");
+        std::uint64_t bits{};
+        std::memcpy(&bits, &number, sizeof(bits));
+        bits = byteswap(bits);
+        std::memcpy(&number, &bits, sizeof(bits));
+    }
+
+    /// reverse the byte order of a float with a single byteswap()
+    static void byte_swap(float& number) noexcept
+    {
+        static_assert(sizeof(float) == sizeof(std::uint32_t), "float must have 32 bits");
+        std::uint32_t bits{};
+        std::memcpy(&bits, &number, sizeof(bits));
+        bits = byteswap(bits);
+        std::memcpy(&number, &bits, sizeof(bits));
+    }
+
     template<class NumberType>
     static void byte_swap(NumberType& number)
     {