json_view: skip the digits of a fraction 16 bytes at a time (SSE2)

The digits after the decimal point were skipped by skip_digits(), 16 compare-and-branch
steps in a row: about 3.5 instructions per digit, and an exit branch that is
mispredicted when the lengths of the fractions vary. On x86 with SSE2, the
fraction is now skipped by skip_fraction_digits(): 16 bytes are compared at
once and the end of the run is taken from the bit mask, which is about a third
of the instructions for a fraction of 12 digits. The last bytes of the input
are skipped one at a time as before. Where there is no SSE2 (AArch64, other
targets, JSON_NO_SIMD) skip_fraction_digits() is skip_digits(): nothing changes
there. Integer digits keep skip_digits(), whose exit is predictable.

GCC unrolls the loop over the last bytes in every inlined copy of the new
function, which made the parse loop 4 KiB larger and parse() of documents with
hardly any fractions about 4 % slower. For GCC that loop is therefore out of
line (skip_digits_tail()); Clang is faster with it inline.

json_document::parse(), A/B in one binary, 6 code layouts, x86-64 Haswell,
time after / before (GCC 16.2 / Clang 23):

- numbers.json (10,001 floats): 0.715 / 0.680
- canada.json: 0.778 / 0.790
- mesh.json: 0.908 / 0.883
- marine_ik.json: 0.937 / 0.926
- mesh.pretty.json: 0.923 / 0.960

Files with few or no fractions and files with fractions of a fixed length
(20,000 prices with two decimals, 60,000 coordinates with six): between 0.98 and
1.014 on both compilers, except citm_catalog.json with GCC (1.05: the
instruction count is +0.3 %, and that row varies by +-7 % between builds of
unrelated variants on this machine).

New test: fractions of every length from 0 to 40 digits followed by every kind of
byte that ends a number or lies next to the digits in ASCII, in several
positions, and cut in every place near the end of the text, against json::parse()
(the length of the number is part of what is compared), for all option
combinations and with and without a terminating NUL. Checked on x86-64 with GCC 16.2,
Clang 23 and Apple clang under Rosetta, and with a deliberate off-by-one at the 16th byte.

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

include/nlohmann/detail/input/binary_reader.hpp

@@ -465,17 +465,15 @@ 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
-            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(!get_bson_cstr(key)))
             {
                 return false;
             }
 
-            if (is_object && !sax->key(key))
+            // an array's elements are named "0", "1", ... in the wire format,
+            // and those names are not passed on
+            if (container_stack.back().is_object && !sax->key(key))
             {
                 return false;
             }
@@ -491,18 +489,14 @@ class binary_reader
     @brief Parses a C-style string from the BSON input.
     @param[in,out] result  A reference to the string variable where the read
                             string is to be stored.
-    @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
     @return `true` if the \\x00-byte indicating the end of the string was
              encountered before the EOF; false` indicates an unexpected EOF.
     */
-    bool get_bson_cstr(string_t& result, const bool store = true)
+    bool get_bson_cstr(string_t& result)
     {
-        if (get_bson_cstr_bulk(result, store, std::integral_constant<bool, bulk_scan> {}))
+        if (get_bson_cstr_bulk(result, std::integral_constant<bool, bulk_scan> {}))
         {
-            return !store || check_string_utf8(result, "key");
+            return check_string_utf8(result, "key");
         }
 
         auto out = std::back_inserter(result);
@@ -515,7 +509,7 @@ class binary_reader
             }
             if (current == 0x00)
             {
-                return !store || check_string_utf8(result, "key");
+                return check_string_utf8(result, "key");
             }
             *out++ = static_cast<typename string_t::value_type>(current);
         }
@@ -525,11 +519,10 @@ class binary_reader
     @brief read a C-style string from contiguous input in one step
 
     @param[in,out] result  the string to append to
-    @param[in] store  whether to append the characters to @a result
     @return whether the string was read; if the input has no \\x00-byte, nothing
             is read, and @ref get_bson_cstr reports the end of the input
     */
-    bool get_bson_cstr_bulk(string_t& result, const bool store, std::true_type /*bulk*/)
+    bool get_bson_cstr_bulk(string_t& result, std::true_type /*bulk*/)
     {
         const std::size_t remaining = ia.bulk_remaining();
         if (remaining == 0)
@@ -537,15 +530,18 @@ class binary_reader
             return false;
         }
         const auto* const data = reinterpret_cast<const unsigned char*>(ia.bulk_data());
-        const std::size_t length = find_zero_byte(data, remaining);
+        // a plain loop rather than std::memchr: most keys are short (array
+        // indices are keys, too), and the call would cost more than it saves
+        std::size_t length = 0;
+        while (length < remaining && data[length] != 0x00)
+        {
+            ++length;
+        }
         if (length == remaining)
         {
             return false;
         }
-        if (store)
-        {
-            result.append(reinterpret_cast<const typename string_t::value_type*>(data), length);
-        }
+        result.append(reinterpret_cast<const typename string_t::value_type*>(data), length);
         // consume the string and its \x00-byte, as the byte-wise path does
         ia.bulk_skip(length + 1);
         chars_read += length + 1;
@@ -554,7 +550,7 @@ class binary_reader
     }
 
     /// 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
+    bool get_bson_cstr_bulk(string_t& /*result*/, std::false_type /*bulk*/) const noexcept
     {
         return false;
     }

include/nlohmann/detail/input/string_scan.hpp

@@ -90,34 +90,6 @@ JSON_HEDLEY_ALWAYS_INLINE std::size_t find_string_special(const unsigned char* d
     return n;
 }
 
-// return the index of the first 0x00 byte in [data, data+n), or n if there is
-// none; scans 8 bytes at a time (BSON's C-style strings)
-JSON_HEDLEY_ALWAYS_INLINE std::size_t find_zero_byte(const unsigned char* data, std::size_t n) noexcept
-{
-    constexpr std::uint64_t ones = 0x0101010101010101ull;
-    constexpr std::uint64_t high = 0x8080808080808080ull;
-    std::size_t i = 0;
-    for (; i + 8 <= n; i += 8)
-    {
-        const std::uint64_t v = read_eight_bytes(data + i);
-        const std::uint64_t zero = (v - ones) & ~v & high;
-        if (zero != 0)
-        {
-            // the lowest flagged byte is the first zero byte (see
-            // find_string_special())
-            return i + (static_cast<std::size_t>(count_trailing_zeros(zero)) / 8);
-        }
-    }
-    for (; i < n; ++i)
-    {
-        if (data[i] == 0x00)
-        {
-            return i;
-        }
-    }
-    return n;
-}
-
 // classify a byte as one the serializer must NOT copy verbatim when
 // ensure_ascii is requested: the closing quote, an escape, a control character
 // (< 0x20), DEL (0x7F), or any non-ASCII byte (>= 0x80). Everything else -

include/nlohmann/detail/view/builder.hpp

@@ -1082,7 +1082,7 @@ indent_done:
             {
                 ++p;
                 const unsigned char* const f0 = p;
-                p = skip_digits(p, e);
+                p = skip_fraction_digits(p, e);
                 if (NLOHMANN_VIEW_UNLIKELY(p == f0))
                 {
                     return fail(error_code::number_after_dot);

include/nlohmann/detail/view/scan.hpp

@@ -198,6 +198,57 @@ inline const unsigned char* find_block_comment_end(const unsigned char* p, const
     return p;
 }
 
+/// The digits of a fraction, whose number varies from one number to the next: skip_digits() with 16 bytes compared at
+/// once and the end of the run taken from a bit mask (SSE2), instead of a compare and a branch for each digit. That
+/// is about a third of the instructions for a fraction of 12 digits, and no branch that is mispredicted when the
+/// lengths vary. Where there is no SSE2, skip_digits() is used.
+#if NLOHMANN_VIEW_SSE2
+// The last bytes of the input, one at a time. GCC unrolls this loop in every inlined copy of the function below, which
+// makes the parse loop 4 KiB larger and slows down documents that have hardly any fractions; out of line, it does
+// not. Clang does better with the loop inline (a call in the parse loop costs it more than the unrolling).
+#if defined(__GNUC__) && !defined(__clang__)
+inline NLOHMANN_VIEW_NOINLINE const unsigned char* skip_digits_tail(const unsigned char* p, const unsigned char* e) noexcept
+{
+    while (p != e && is_digit(*p))
+    {
+        ++p;
+    }
+    return p;
+}
+#else
+NLOHMANN_VIEW_ALWAYS_INLINE const unsigned char* skip_digits_tail(const unsigned char* p, const unsigned char* e) noexcept
+{
+    while (p != e && is_digit(*p))
+    {
+        ++p;
+    }
+    return p;
+}
+#endif
+
+NLOHMANN_VIEW_ALWAYS_INLINE const unsigned char* skip_fraction_digits(const unsigned char* p, const unsigned char* e) noexcept
+{
+    while (e - p >= 16)
+    {
+        const __m128i in = _mm_loadu_si128(static_cast<const __m128i*>(static_cast<const void*>(p))); // NOLINT(bugprone-casting-through-void)
+        // a digit is above '/' and below ':' (signed compares: a byte with the top bit set is negative, and neither)
+        const __m128i digits = _mm_and_si128(_mm_cmpgt_epi8(in, _mm_set1_epi8('/')), _mm_cmplt_epi8(in, _mm_set1_epi8(':')));
+        const auto not_digit = static_cast<std::uint64_t>(~static_cast<unsigned>(_mm_movemask_epi8(digits)) & 0xFFFFu);
+        if (not_digit != 0)
+        {
+            return p + count_trailing_zeros(not_digit);
+        }
+        p += 16;
+    }
+    return skip_digits_tail(p, e);
+}
+#else
+NLOHMANN_VIEW_ALWAYS_INLINE const unsigned char* skip_fraction_digits(const unsigned char* p, const unsigned char* e) noexcept
+{
+    return skip_digits(p, e);
+}
+#endif
+
 /// powers of ten up to 10^19 as integers
 inline std::uint64_t int_pow10(unsigned k) noexcept
 {

single_include/nlohmann/json.hpp

@@ -12458,34 +12458,6 @@ JSON_HEDLEY_ALWAYS_INLINE std::size_t find_string_special(const unsigned char* d
     return n;
 }
 
-// return the index of the first 0x00 byte in [data, data+n), or n if there is
-// none; scans 8 bytes at a time (BSON's C-style strings)
-JSON_HEDLEY_ALWAYS_INLINE std::size_t find_zero_byte(const unsigned char* data, std::size_t n) noexcept
-{
-    constexpr std::uint64_t ones = 0x0101010101010101ull;
-    constexpr std::uint64_t high = 0x8080808080808080ull;
-    std::size_t i = 0;
-    for (; i + 8 <= n; i += 8)
-    {
-        const std::uint64_t v = read_eight_bytes(data + i);
-        const std::uint64_t zero = (v - ones) & ~v & high;
-        if (zero != 0)
-        {
-            // the lowest flagged byte is the first zero byte (see
-            // find_string_special())
-            return i + (static_cast<std::size_t>(count_trailing_zeros(zero)) / 8);
-        }
-    }
-    for (; i < n; ++i)
-    {
-        if (data[i] == 0x00)
-        {
-            return i;
-        }
-    }
-    return n;
-}
-
 // classify a byte as one the serializer must NOT copy verbatim when
 // ensure_ascii is requested: the closing quote, an escape, a control character
 // (< 0x20), DEL (0x7F), or any non-ASCII byte (>= 0x80). Everything else -
@@ -18269,17 +18241,15 @@ 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
-            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(!get_bson_cstr(key)))
             {
                 return false;
             }
 
-            if (is_object && !sax->key(key))
+            // an array's elements are named "0", "1", ... in the wire format,
+            // and those names are not passed on
+            if (container_stack.back().is_object && !sax->key(key))
             {
                 return false;
             }
@@ -18295,18 +18265,14 @@ class binary_reader
     @brief Parses a C-style string from the BSON input.
     @param[in,out] result  A reference to the string variable where the read
                             string is to be stored.
-    @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
     @return `true` if the \\x00-byte indicating the end of the string was
              encountered before the EOF; false` indicates an unexpected EOF.
     */
-    bool get_bson_cstr(string_t& result, const bool store = true)
+    bool get_bson_cstr(string_t& result)
     {
-        if (get_bson_cstr_bulk(result, store, std::integral_constant<bool, bulk_scan> {}))
+        if (get_bson_cstr_bulk(result, std::integral_constant<bool, bulk_scan> {}))
         {
-            return !store || check_string_utf8(result, "key");
+            return check_string_utf8(result, "key");
         }
 
         auto out = std::back_inserter(result);
@@ -18319,7 +18285,7 @@ class binary_reader
             }
             if (current == 0x00)
             {
-                return !store || check_string_utf8(result, "key");
+                return check_string_utf8(result, "key");
             }
             *out++ = static_cast<typename string_t::value_type>(current);
         }
@@ -18329,11 +18295,10 @@ class binary_reader
     @brief read a C-style string from contiguous input in one step
 
     @param[in,out] result  the string to append to
-    @param[in] store  whether to append the characters to @a result
     @return whether the string was read; if the input has no \\x00-byte, nothing
             is read, and @ref get_bson_cstr reports the end of the input
     */
-    bool get_bson_cstr_bulk(string_t& result, const bool store, std::true_type /*bulk*/)
+    bool get_bson_cstr_bulk(string_t& result, std::true_type /*bulk*/)
     {
         const std::size_t remaining = ia.bulk_remaining();
         if (remaining == 0)
@@ -18341,15 +18306,18 @@ class binary_reader
             return false;
         }
         const auto* const data = reinterpret_cast<const unsigned char*>(ia.bulk_data());
-        const std::size_t length = find_zero_byte(data, remaining);
+        // a plain loop rather than std::memchr: most keys are short (array
+        // indices are keys, too), and the call would cost more than it saves
+        std::size_t length = 0;
+        while (length < remaining && data[length] != 0x00)
+        {
+            ++length;
+        }
         if (length == remaining)
         {
             return false;
         }
-        if (store)
-        {
-            result.append(reinterpret_cast<const typename string_t::value_type*>(data), length);
-        }
+        result.append(reinterpret_cast<const typename string_t::value_type*>(data), length);
         // consume the string and its \x00-byte, as the byte-wise path does
         ia.bulk_skip(length + 1);
         chars_read += length + 1;
@@ -18358,7 +18326,7 @@ class binary_reader
     }
 
     /// 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
+    bool get_bson_cstr_bulk(string_t& /*result*/, std::false_type /*bulk*/) const noexcept
     {
         return false;
     }

single_include/nlohmann/json_view.hpp

@@ -1218,6 +1218,57 @@ inline const unsigned char* find_block_comment_end(const unsigned char* p, const
     return p;
 }
 
+/// The digits of a fraction, whose number varies from one number to the next: skip_digits() with 16 bytes compared at
+/// once and the end of the run taken from a bit mask (SSE2), instead of a compare and a branch for each digit. That
+/// is about a third of the instructions for a fraction of 12 digits, and no branch that is mispredicted when the
+/// lengths vary. Where there is no SSE2, skip_digits() is used.
+#if NLOHMANN_VIEW_SSE2
+// The last bytes of the input, one at a time. GCC unrolls this loop in every inlined copy of the function below, which
+// makes the parse loop 4 KiB larger and slows down documents that have hardly any fractions; out of line, it does
+// not. Clang does better with the loop inline (a call in the parse loop costs it more than the unrolling).
+#if defined(__GNUC__) && !defined(__clang__)
+inline NLOHMANN_VIEW_NOINLINE const unsigned char* skip_digits_tail(const unsigned char* p, const unsigned char* e) noexcept
+{
+    while (p != e && is_digit(*p))
+    {
+        ++p;
+    }
+    return p;
+}
+#else
+NLOHMANN_VIEW_ALWAYS_INLINE const unsigned char* skip_digits_tail(const unsigned char* p, const unsigned char* e) noexcept
+{
+    while (p != e && is_digit(*p))
+    {
+        ++p;
+    }
+    return p;
+}
+#endif
+
+NLOHMANN_VIEW_ALWAYS_INLINE const unsigned char* skip_fraction_digits(const unsigned char* p, const unsigned char* e) noexcept
+{
+    while (e - p >= 16)
+    {
+        const __m128i in = _mm_loadu_si128(static_cast<const __m128i*>(static_cast<const void*>(p))); // NOLINT(bugprone-casting-through-void)
+        // a digit is above '/' and below ':' (signed compares: a byte with the top bit set is negative, and neither)
+        const __m128i digits = _mm_and_si128(_mm_cmpgt_epi8(in, _mm_set1_epi8('/')), _mm_cmplt_epi8(in, _mm_set1_epi8(':')));
+        const auto not_digit = static_cast<std::uint64_t>(~static_cast<unsigned>(_mm_movemask_epi8(digits)) & 0xFFFFu);
+        if (not_digit != 0)
+        {
+            return p + count_trailing_zeros(not_digit);
+        }
+        p += 16;
+    }
+    return skip_digits_tail(p, e);
+}
+#else
+NLOHMANN_VIEW_ALWAYS_INLINE const unsigned char* skip_fraction_digits(const unsigned char* p, const unsigned char* e) noexcept
+{
+    return skip_digits(p, e);
+}
+#endif
+
 /// powers of ten up to 10^19 as integers
 inline std::uint64_t int_pow10(unsigned k) noexcept
 {
@@ -2324,7 +2375,7 @@ indent_done:
             {
                 ++p;
                 const unsigned char* const f0 = p;
-                p = skip_digits(p, e);
+                p = skip_fraction_digits(p, e);
                 if (NLOHMANN_VIEW_UNLIKELY(p == f0))
                 {
                     return fail(error_code::number_after_dot);

tests/src/unit-bson.cpp

@@ -1164,13 +1164,7 @@ TEST_CASE("BSON keys from contiguous and stream input")
     // contiguous input reads a key up to its \x00-byte in one step, a stream
     // reads it byte by byte; both must give the same value or error for the
     // complete document and for every truncation of it
-    json j = {{"", true}, {"k", {1, 2, 3}}, {std::string(40, 'x'), {{"nested key", "value"}}}};
-    // keys and array element names ("0" to "11") that end at every position of
-    // an 8-byte word, which contiguous input scans at a time
-    for (std::size_t length = 1; length <= 17; ++length)
-    {
-        j[std::string(length, static_cast<char>('a' + length))] = json::array({0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, {11}});
-    }
+    const json j = {{"", true}, {"k", {1, 2, 3}}, {std::string(40, 'x'), {{"nested key", "value"}}}};
     const std::vector<std::uint8_t> bson = json::to_bson(j);
     CHECK(json::from_bson(bson) == j);
 
@@ -1190,7 +1184,7 @@ TEST_CASE("BSON keys from contiguous and stream input")
         }
         try
         {
-            std::istringstream stream(std::string(reinterpret_cast<const char*>(input.data()), input.size()));
+            std::istringstream stream(std::string(input.begin(), input.end()));
             from_stream = json::from_bson(stream).dump();
         }
         catch (const json::parse_error& e)
@@ -1200,33 +1194,6 @@ TEST_CASE("BSON keys from contiguous and stream input")
         CHECK(from_vector == from_stream);
     }
 }
-
-TEST_CASE("BSON array element names")
-{
-    // an array's element names are not part of the value, so they are not
-    // passed on; error_handler_t::strict still rejects ill-formed UTF-8 in them
-    const std::vector<std::uint8_t> bson =
-    {
-        0x1D, 0x00, 0x00, 0x00, // size (little endian)
-        0x04, 'a', 0x00,        // entry: array "a"
-        0x15, 0x00, 0x00, 0x00, // size (little endian)
-        0x08, 0xFF, 0x00, 0x01, // entry: boolean named "\xFF", true
-        0x08, '1', '2', '3', '4', '5', '6', '7', '8', '9', 0x00, 0x00, // entry: boolean named "123456789", false
-        0x00,                   // end of the array
-        0x00                    // end of the document
-    };
-    const json expected = {{"a", {true, false}}};
-
-    CHECK(json::from_bson(bson) == expected);
-    CHECK(json::from_bson(bson, true, true, json::error_handler_t::replace) == expected);
-    CHECK(json::from_bson(bson, true, true, json::error_handler_t::ignore) == expected);
-
-    json _;
-    CHECK_THROWS_WITH_AS(_ = json::from_bson(bson, 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&);
-
-    std::istringstream stream(std::string(reinterpret_cast<const char*>(bson.data()), bson.size()));
-    CHECK(json::from_bson(stream) == expected);
-}
 #endif
 
 TEST_CASE("Negative size of binary value")

tests/src/unit-json_view_builder.cpp

@@ -523,6 +523,47 @@ TEST_CASE("json_view builder: strings across vector blocks")
     }
 }
 
+TEST_CASE("json_view builder: fraction digits")
+{
+    // The digits of a fraction are skipped 16 bytes at a time where there are
+    // that many left (SSE2), and one at a time after that. Fractions of every
+    // length from 0 to 40 digits, followed by every kind of byte that ends a
+    // number or lies next to the digits in ASCII ('/' and ':'), as values, as
+    // keys' neighbours, and at the very end of the text (a vector without a
+    // terminating NUL). The length of the number is part of the value that
+    // the test compares with json::parse().
+    const std::string followers = ",]}/:.eE+-x \n\t\"09";
+    std::mt19937 rng(2028); // NOLINT(cert-msc32-c,cert-msc51-cpp,bugprone-random-generator-seed): reproducible
+    for (std::size_t digits = 0; digits <= 40; ++digits)
+    {
+        for (int i = 0; i < 3; ++i)
+        {
+            std::string fraction;
+            for (std::size_t k = 0; k < digits; ++k)
+            {
+                fraction += static_cast<char>('0' + (rng() % 10));
+            }
+            const std::string number = std::string(rng() % 2 == 0 ? "-" : "") + std::to_string(rng() % 100000) + "." + fraction;
+            check_same(number);
+            check_same("[" + number + "]");
+            check_same("[" + number + "," + number + "]");
+            check_same("{\"a\":" + number + "}");
+            for (const char follower : followers)
+            {
+                check_same(number + follower);
+                check_same("[" + number + follower + "]");
+                check_same("[" + number + follower + std::string(rng() % 20, '1'));
+            }
+            // everything near the end of the text: the number cut in every place
+            for (std::size_t cut = 1; cut <= 18 && cut <= number.size(); ++cut)
+            {
+                check_same(number.substr(0, number.size() - cut));
+                check_same("[" + number.substr(0, number.size() - cut));
+            }
+        }
+    }
+}
+
 TEST_CASE("json_view builder: escapes between runs of text")
 {
     // The text between escapes is copied 16 bytes at a time where there are