Compare commits

..
Author SHA1 Message Date
Niels Lohmann b0a4218e38 Report the same error position for bulk number runs as for byte-wise input
For contiguous input, the CBOR and MessagePack readers read the
scalars of an array in one loop, and the UBJSON/BJData readers the
numbers of an optimized array. These loops passed a value on before
they advanced the position and the last byte. With a number_float_t
narrower than the value (for example float), passing the value on
reports out_of_range.406, and SAX parse_error() then got a different
position and last token than for the same bytes read through a
non-contiguous adapter.

Advance before passing the value on in the CBOR and MessagePack loops.
The last byte is the head byte, as get_number() does not change it;
the MessagePack loop used the last byte of the value. In the typed
UBJSON/BJData loop, update the position per value only where passing
it on can report an error, which is decided at compile time and never
the case for the default number types, and do not change the last byte,
as reading the values one by one does not.

Instruction counts of SAX and DOM parsing of arrays of one million
numbers: CBOR -1 to -4 %, MessagePack -2 to -10 %, UBJSON/BJData typed
arrays -0.8 to +0.6 % (clang, -O2).

Signed-off-by: Niels Lohmann <mail@nlohmann.me>
2026-11-16 07:41:29 +01:00
5 changed files with 271 additions and 190 deletions

No files matched your search

+72 -43
View File
@@ -1927,6 +1927,21 @@ class binary_reader
return result;
}
/*!
@brief move past a value of @a size bytes, starting with @a head, that a
bulk run reads, as get() and get_number() would
Called before the value is passed on, so that an error reported while
passing it on (see @ref emit_float) has the position and the last byte
that reading the value byte by byte gives.
*/
void bulk_advance(const std::size_t size, const unsigned char head)
{
ia.bulk_skip(size);
chars_read += size;
current = head;
}
/*!
@brief read the integers, floats, booleans, and nulls that follow in a
definite-length array, without the per-element bookkeeping of
@@ -1952,71 +1967,70 @@ class binary_reader
{
const auto* const data = reinterpret_cast<const unsigned char*>(ia.bulk_data());
const unsigned char head = data[0];
std::size_t size = 0;
bool accepted = false;
switch (head)
{
case 0x18:
size = 2;
bulk_advance(2, head);
accepted = emit_unsigned(data[1]);
break;
case 0x19:
size = 3;
bulk_advance(3, head);
accepted = emit_unsigned(load_big_endian<std::uint16_t>(data + 1));
break;
case 0x1A:
size = 5;
bulk_advance(5, head);
accepted = emit_unsigned(load_big_endian<std::uint32_t>(data + 1));
break;
case 0x1B:
size = 9;
bulk_advance(9, head);
accepted = emit_unsigned(load_big_endian<std::uint64_t>(data + 1));
break;
case 0x38:
size = 2;
bulk_advance(2, head);
accepted = emit_cbor_negative_integer(data[1]);
break;
case 0x39:
size = 3;
bulk_advance(3, head);
accepted = emit_cbor_negative_integer(load_big_endian<std::uint16_t>(data + 1));
break;
case 0x3A:
size = 5;
bulk_advance(5, head);
accepted = emit_cbor_negative_integer(load_big_endian<std::uint32_t>(data + 1));
break;
case 0x3B:
size = 9;
bulk_advance(9, head);
accepted = emit_cbor_negative_integer(load_big_endian<std::uint64_t>(data + 1));
break;
case 0xF4:
size = 1;
bulk_advance(1, head);
accepted = sax->boolean(false);
break;
case 0xF5:
size = 1;
bulk_advance(1, head);
accepted = sax->boolean(true);
break;
case 0xF6:
size = 1;
bulk_advance(1, head);
accepted = sax->null();
break;
case 0xFA:
size = 5;
bulk_advance(5, head);
accepted = emit_float(load_big_endian<float>(data + 1));
break;
case 0xFB:
size = 9;
bulk_advance(9, head);
accepted = emit_float(load_big_endian<double>(data + 1));
break;
default:
if (head <= 0x17)
{
size = 1;
bulk_advance(1, head);
accepted = sax->number_unsigned(static_cast<number_unsigned_t>(head));
}
else if (head >= 0x20 && head <= 0x37)
{
size = 1;
bulk_advance(1, head);
accepted = sax->number_integer(static_cast<std::int8_t>(0x20 - 1 - head));
}
else
@@ -2026,10 +2040,6 @@ class binary_reader
}
break;
}
// as if read by get() and get_number(): current is the head byte
ia.bulk_skip(size);
chars_read += size;
current = head;
--remaining;
if (JSON_HEDLEY_UNLIKELY(!accepted))
{
@@ -2867,73 +2877,72 @@ class binary_reader
{
const auto* const data = reinterpret_cast<const unsigned char*>(ia.bulk_data());
const unsigned char head = data[0];
std::size_t size = 0;
bool accepted = false;
switch (head)
{
case 0xC0: // nil
size = 1;
bulk_advance(1, head);
accepted = sax->null();
break;
case 0xC2: // false
size = 1;
bulk_advance(1, head);
accepted = sax->boolean(false);
break;
case 0xC3: // true
size = 1;
bulk_advance(1, head);
accepted = sax->boolean(true);
break;
case 0xCA: // float 32
size = 5;
bulk_advance(5, head);
accepted = emit_float(load_big_endian<float>(data + 1));
break;
case 0xCB: // float 64
size = 9;
bulk_advance(9, head);
accepted = emit_float(load_big_endian<double>(data + 1));
break;
case 0xCC: // uint 8
size = 2;
bulk_advance(2, head);
accepted = emit_unsigned(data[1]);
break;
case 0xCD: // uint 16
size = 3;
bulk_advance(3, head);
accepted = emit_unsigned(load_big_endian<std::uint16_t>(data + 1));
break;
case 0xCE: // uint 32
size = 5;
bulk_advance(5, head);
accepted = emit_unsigned(load_big_endian<std::uint32_t>(data + 1));
break;
case 0xCF: // uint 64
size = 9;
bulk_advance(9, head);
accepted = emit_unsigned(load_big_endian<std::uint64_t>(data + 1));
break;
case 0xD0: // int 8
size = 2;
bulk_advance(2, head);
accepted = emit_signed(static_cast<std::int8_t>(data[1]));
break;
case 0xD1: // int 16
size = 3;
bulk_advance(3, head);
accepted = emit_signed(load_big_endian<std::int16_t>(data + 1));
break;
case 0xD2: // int 32
size = 5;
bulk_advance(5, head);
accepted = emit_signed(load_big_endian<std::int32_t>(data + 1));
break;
case 0xD3: // int 64
size = 9;
bulk_advance(9, head);
accepted = emit_signed(load_big_endian<std::int64_t>(data + 1));
break;
default:
if (head <= 0x7F)
{
// positive fixint
size = 1;
bulk_advance(1, head);
accepted = sax->number_unsigned(static_cast<number_unsigned_t>(head));
}
else if (head >= 0xE0)
{
// negative fixint
size = 1;
bulk_advance(1, head);
accepted = sax->number_integer(static_cast<std::int8_t>(head));
}
else
@@ -2943,10 +2952,6 @@ class binary_reader
}
break;
}
// as if read by get() and get_number(): current is the last byte
ia.bulk_skip(size);
chars_read += size;
current = data[size - 1];
--remaining;
if (JSON_HEDLEY_UNLIKELY(!accepted))
{
@@ -3850,6 +3855,7 @@ class binary_reader
const auto* const data = reinterpret_cast<const unsigned char*>(ia.bulk_data());
// UBJSON stores numbers big-endian, BJData little-endian
const bool swap = is_little_endian != (input_format == input_format_t::bjdata);
const std::size_t start = chars_read;
std::size_t read = 0;
bool accepted = true;
while (read < count && accepted)
@@ -3861,16 +3867,39 @@ class binary_reader
byte_swap(number);
}
++read;
// where passing the value on can report an error (see
// emit_float), the position is that after the value, as if it
// was read by get_number()
set_chars_read(start + (read * size), std::integral_constant<bool, may_overflow_number_float<NumberType>()> {});
accepted = emit_ubjson_number(number);
}
// as if read by get_number(): current is the last byte read
// get_number() does not change current
ia.bulk_skip(read * size);
chars_read += read * size;
current = data[(read * size) - 1];
chars_read = start + (read * size);
remaining -= read;
return accepted;
}
/*!
@brief whether a value of type @a NumberType can be finite and still
overflow number_float_t, which @ref emit_float reports as an error
False for the default number types, so that the loop in @ref
parse_ubjson_typed_run need not track the position per value.
*/
template<typename NumberType>
static constexpr bool may_overflow_number_float() noexcept
{
return static_cast<long double>((std::numeric_limits<number_float_t>::max)()) < static_cast<long double>((std::numeric_limits<NumberType>::max)());
}
void set_chars_read(const std::size_t position, std::true_type /*track*/) noexcept
{
chars_read = position;
}
void set_chars_read(const std::size_t /*position*/, std::false_type /*track*/) const noexcept {}
template<typename NumberType, enable_if_t<std::is_floating_point<NumberType>::value, int> = 0>
bool emit_ubjson_number(const NumberType number)
{
+2 -41
View File
@@ -636,7 +636,7 @@ class serializer
put_literal("\"subtype\": ");
if (binary.has_subtype())
{
dump_subtype(binary.subtype());
dump_integer(binary.subtype());
}
else
{
@@ -662,7 +662,7 @@ class serializer
put_literal("],\"subtype\":");
if (binary.has_subtype())
{
dump_subtype(binary.subtype());
dump_integer(binary.subtype());
put_char('}');
}
else
@@ -1883,40 +1883,6 @@ class serializer
write_buffer_pos = pos;
}
/*!
@brief dump the subtype of a binary value
The subtype is a std::uint64_t, whatever the number types are. @ref
dump_integer converts its argument to the magnitude type of
number_unsigned_t, so it writes the subtype only if that type has at least
64 bits; otherwise (it would truncate the subtype) the digits are written
one by one.
@param[in] subtype the subtype to dump
*/
void dump_subtype(std::uint64_t subtype)
{
dump_subtype(subtype, std::integral_constant<bool, magnitude_has_at_least_64_bits> {});
}
void dump_subtype(std::uint64_t subtype, std::true_type /*magnitude has at least 64 bits*/)
{
dump_integer(subtype);
}
void dump_subtype(std::uint64_t subtype, std::false_type /*magnitude has less than 64 bits*/)
{
std::array<char, 20> digits{{}}; // 2^64 - 1 has 20 digits
std::size_t start = digits.size();
do
{
digits[--start] = static_cast<char>('0' + (subtype % 10));
subtype /= 10;
}
while (subtype != 0);
put_string(digits, start, digits.size());
}
/*!
@brief dump an integer
@@ -2170,11 +2136,6 @@ class serializer
static constexpr bool magnitude_has_at_most_64_bits = std::numeric_limits<number_magnitude_t>::is_specialized
&& std::numeric_limits<number_magnitude_t>::digits <= 64;
/// whether every std::uint64_t fits number_magnitude_t (then dump_integer
/// can write a binary subtype)
static constexpr bool magnitude_has_at_least_64_bits = std::numeric_limits<number_magnitude_t>::is_specialized
&& std::numeric_limits<number_magnitude_t>::digits >= 64;
/// whether @a x fits 64 bits: always if number_magnitude_t has at most
/// 64 bits (selected by type, as the comparison would be reported as
/// always true), else compared
+74 -84
View File
@@ -20605,6 +20605,21 @@ class binary_reader
return result;
}
/*!
@brief move past a value of @a size bytes, starting with @a head, that a
bulk run reads, as get() and get_number() would
Called before the value is passed on, so that an error reported while
passing it on (see @ref emit_float) has the position and the last byte
that reading the value byte by byte gives.
*/
void bulk_advance(const std::size_t size, const unsigned char head)
{
ia.bulk_skip(size);
chars_read += size;
current = head;
}
/*!
@brief read the integers, floats, booleans, and nulls that follow in a
definite-length array, without the per-element bookkeeping of
@@ -20630,71 +20645,70 @@ class binary_reader
{
const auto* const data = reinterpret_cast<const unsigned char*>(ia.bulk_data());
const unsigned char head = data[0];
std::size_t size = 0;
bool accepted = false;
switch (head)
{
case 0x18:
size = 2;
bulk_advance(2, head);
accepted = emit_unsigned(data[1]);
break;
case 0x19:
size = 3;
bulk_advance(3, head);
accepted = emit_unsigned(load_big_endian<std::uint16_t>(data + 1));
break;
case 0x1A:
size = 5;
bulk_advance(5, head);
accepted = emit_unsigned(load_big_endian<std::uint32_t>(data + 1));
break;
case 0x1B:
size = 9;
bulk_advance(9, head);
accepted = emit_unsigned(load_big_endian<std::uint64_t>(data + 1));
break;
case 0x38:
size = 2;
bulk_advance(2, head);
accepted = emit_cbor_negative_integer(data[1]);
break;
case 0x39:
size = 3;
bulk_advance(3, head);
accepted = emit_cbor_negative_integer(load_big_endian<std::uint16_t>(data + 1));
break;
case 0x3A:
size = 5;
bulk_advance(5, head);
accepted = emit_cbor_negative_integer(load_big_endian<std::uint32_t>(data + 1));
break;
case 0x3B:
size = 9;
bulk_advance(9, head);
accepted = emit_cbor_negative_integer(load_big_endian<std::uint64_t>(data + 1));
break;
case 0xF4:
size = 1;
bulk_advance(1, head);
accepted = sax->boolean(false);
break;
case 0xF5:
size = 1;
bulk_advance(1, head);
accepted = sax->boolean(true);
break;
case 0xF6:
size = 1;
bulk_advance(1, head);
accepted = sax->null();
break;
case 0xFA:
size = 5;
bulk_advance(5, head);
accepted = emit_float(load_big_endian<float>(data + 1));
break;
case 0xFB:
size = 9;
bulk_advance(9, head);
accepted = emit_float(load_big_endian<double>(data + 1));
break;
default:
if (head <= 0x17)
{
size = 1;
bulk_advance(1, head);
accepted = sax->number_unsigned(static_cast<number_unsigned_t>(head));
}
else if (head >= 0x20 && head <= 0x37)
{
size = 1;
bulk_advance(1, head);
accepted = sax->number_integer(static_cast<std::int8_t>(0x20 - 1 - head));
}
else
@@ -20704,10 +20718,6 @@ class binary_reader
}
break;
}
// as if read by get() and get_number(): current is the head byte
ia.bulk_skip(size);
chars_read += size;
current = head;
--remaining;
if (JSON_HEDLEY_UNLIKELY(!accepted))
{
@@ -21545,73 +21555,72 @@ class binary_reader
{
const auto* const data = reinterpret_cast<const unsigned char*>(ia.bulk_data());
const unsigned char head = data[0];
std::size_t size = 0;
bool accepted = false;
switch (head)
{
case 0xC0: // nil
size = 1;
bulk_advance(1, head);
accepted = sax->null();
break;
case 0xC2: // false
size = 1;
bulk_advance(1, head);
accepted = sax->boolean(false);
break;
case 0xC3: // true
size = 1;
bulk_advance(1, head);
accepted = sax->boolean(true);
break;
case 0xCA: // float 32
size = 5;
bulk_advance(5, head);
accepted = emit_float(load_big_endian<float>(data + 1));
break;
case 0xCB: // float 64
size = 9;
bulk_advance(9, head);
accepted = emit_float(load_big_endian<double>(data + 1));
break;
case 0xCC: // uint 8
size = 2;
bulk_advance(2, head);
accepted = emit_unsigned(data[1]);
break;
case 0xCD: // uint 16
size = 3;
bulk_advance(3, head);
accepted = emit_unsigned(load_big_endian<std::uint16_t>(data + 1));
break;
case 0xCE: // uint 32
size = 5;
bulk_advance(5, head);
accepted = emit_unsigned(load_big_endian<std::uint32_t>(data + 1));
break;
case 0xCF: // uint 64
size = 9;
bulk_advance(9, head);
accepted = emit_unsigned(load_big_endian<std::uint64_t>(data + 1));
break;
case 0xD0: // int 8
size = 2;
bulk_advance(2, head);
accepted = emit_signed(static_cast<std::int8_t>(data[1]));
break;
case 0xD1: // int 16
size = 3;
bulk_advance(3, head);
accepted = emit_signed(load_big_endian<std::int16_t>(data + 1));
break;
case 0xD2: // int 32
size = 5;
bulk_advance(5, head);
accepted = emit_signed(load_big_endian<std::int32_t>(data + 1));
break;
case 0xD3: // int 64
size = 9;
bulk_advance(9, head);
accepted = emit_signed(load_big_endian<std::int64_t>(data + 1));
break;
default:
if (head <= 0x7F)
{
// positive fixint
size = 1;
bulk_advance(1, head);
accepted = sax->number_unsigned(static_cast<number_unsigned_t>(head));
}
else if (head >= 0xE0)
{
// negative fixint
size = 1;
bulk_advance(1, head);
accepted = sax->number_integer(static_cast<std::int8_t>(head));
}
else
@@ -21621,10 +21630,6 @@ class binary_reader
}
break;
}
// as if read by get() and get_number(): current is the last byte
ia.bulk_skip(size);
chars_read += size;
current = data[size - 1];
--remaining;
if (JSON_HEDLEY_UNLIKELY(!accepted))
{
@@ -22528,6 +22533,7 @@ class binary_reader
const auto* const data = reinterpret_cast<const unsigned char*>(ia.bulk_data());
// UBJSON stores numbers big-endian, BJData little-endian
const bool swap = is_little_endian != (input_format == input_format_t::bjdata);
const std::size_t start = chars_read;
std::size_t read = 0;
bool accepted = true;
while (read < count && accepted)
@@ -22539,16 +22545,39 @@ class binary_reader
byte_swap(number);
}
++read;
// where passing the value on can report an error (see
// emit_float), the position is that after the value, as if it
// was read by get_number()
set_chars_read(start + (read * size), std::integral_constant<bool, may_overflow_number_float<NumberType>()> {});
accepted = emit_ubjson_number(number);
}
// as if read by get_number(): current is the last byte read
// get_number() does not change current
ia.bulk_skip(read * size);
chars_read += read * size;
current = data[(read * size) - 1];
chars_read = start + (read * size);
remaining -= read;
return accepted;
}
/*!
@brief whether a value of type @a NumberType can be finite and still
overflow number_float_t, which @ref emit_float reports as an error
False for the default number types, so that the loop in @ref
parse_ubjson_typed_run need not track the position per value.
*/
template<typename NumberType>
static constexpr bool may_overflow_number_float() noexcept
{
return static_cast<long double>((std::numeric_limits<number_float_t>::max)()) < static_cast<long double>((std::numeric_limits<NumberType>::max)());
}
void set_chars_read(const std::size_t position, std::true_type /*track*/) noexcept
{
chars_read = position;
}
void set_chars_read(const std::size_t /*position*/, std::false_type /*track*/) const noexcept {}
template<typename NumberType, enable_if_t<std::is_floating_point<NumberType>::value, int> = 0>
bool emit_ubjson_number(const NumberType number)
{
@@ -38122,7 +38151,7 @@ class serializer
put_literal("\"subtype\": ");
if (binary.has_subtype())
{
dump_subtype(binary.subtype());
dump_integer(binary.subtype());
}
else
{
@@ -38148,7 +38177,7 @@ class serializer
put_literal("],\"subtype\":");
if (binary.has_subtype())
{
dump_subtype(binary.subtype());
dump_integer(binary.subtype());
put_char('}');
}
else
@@ -39369,40 +39398,6 @@ class serializer
write_buffer_pos = pos;
}
/*!
@brief dump the subtype of a binary value
The subtype is a std::uint64_t, whatever the number types are. @ref
dump_integer converts its argument to the magnitude type of
number_unsigned_t, so it writes the subtype only if that type has at least
64 bits; otherwise (it would truncate the subtype) the digits are written
one by one.
@param[in] subtype the subtype to dump
*/
void dump_subtype(std::uint64_t subtype)
{
dump_subtype(subtype, std::integral_constant<bool, magnitude_has_at_least_64_bits> {});
}
void dump_subtype(std::uint64_t subtype, std::true_type /*magnitude has at least 64 bits*/)
{
dump_integer(subtype);
}
void dump_subtype(std::uint64_t subtype, std::false_type /*magnitude has less than 64 bits*/)
{
std::array<char, 20> digits{{}}; // 2^64 - 1 has 20 digits
std::size_t start = digits.size();
do
{
digits[--start] = static_cast<char>('0' + (subtype % 10));
subtype /= 10;
}
while (subtype != 0);
put_string(digits, start, digits.size());
}
/*!
@brief dump an integer
@@ -39656,11 +39651,6 @@ class serializer
static constexpr bool magnitude_has_at_most_64_bits = std::numeric_limits<number_magnitude_t>::is_specialized
&& std::numeric_limits<number_magnitude_t>::digits <= 64;
/// whether every std::uint64_t fits number_magnitude_t (then dump_integer
/// can write a binary subtype)
static constexpr bool magnitude_has_at_least_64_bits = std::numeric_limits<number_magnitude_t>::is_specialized
&& std::numeric_limits<number_magnitude_t>::digits >= 64;
/// whether @a x fits 64 bits: always if number_magnitude_t has at most
/// 64 bits (selected by type, as the comparison would be reported as
/// always true), else compared
+123
View File
@@ -14,6 +14,7 @@ using nlohmann::json;
#include <cmath>
#include <fstream>
#include <limits>
#include <list>
#include <map>
#include <string>
#include <vector>
@@ -367,3 +368,125 @@ TEST_CASE("Binary formats with narrow number types")
CHECK(format.decode(format.encode(json(1e300)), false).is_discarded());
}
}
namespace
{
// records the SAX events of narrow_json, including the arguments of parse_error
struct narrow_event_recorder
{
using number_integer_t = narrow_json::number_integer_t;
using number_unsigned_t = narrow_json::number_unsigned_t;
using number_float_t = narrow_json::number_float_t;
using string_t = narrow_json::string_t;
using binary_t = narrow_json::binary_t;
bool null()
{
events.emplace_back("null");
return true;
}
bool boolean(bool val)
{
events.emplace_back(val ? "true" : "false");
return true;
}
bool number_integer(number_integer_t val)
{
events.push_back(std::to_string(val));
return true;
}
bool number_unsigned(number_unsigned_t val)
{
events.push_back(std::to_string(val));
return true;
}
bool number_float(number_float_t val, const string_t& /*unused*/)
{
events.push_back(std::to_string(val));
return true;
}
bool string(string_t& val)
{
events.push_back(val);
return true;
}
bool binary(binary_t& /*unused*/)
{
events.emplace_back("binary");
return true;
}
bool start_object(std::size_t /*unused*/)
{
events.emplace_back("{");
return true;
}
bool key(string_t& val)
{
events.push_back(val);
return true;
}
bool end_object()
{
events.emplace_back("}");
return true;
}
bool start_array(std::size_t /*unused*/)
{
events.emplace_back("[");
return true;
}
bool end_array()
{
events.emplace_back("]");
return true;
}
bool parse_error(std::size_t position, const std::string& last_token, const nlohmann::detail::exception& ex)
{
events.push_back("parse_error(" + std::to_string(position) + ", " + last_token + ", " + ex.what() + ")");
return false;
}
std::vector<std::string> events {}; // NOLINT(readability-redundant-member-init)
};
} // namespace
TEST_CASE("Binary formats with narrow number types from contiguous and non-contiguous input")
{
// contiguous input reads runs of numbers in one loop; a number that
// overflows number_float_t must be reported with the same position and
// last byte as when the input is read byte by byte
const json values = {1, 70000, -70000, 1.5, 2.5, 1e300, 3.5, 4.5, 5.5, 6.5, 7.5, 8.5};
const json doubles = {1.5, 2.5, 3.5, 1e300, 4.5, 5.5, 6.5, 7.5, 8.5};
struct binary_input
{
const char* name;
json::input_format_t format;
bytes data;
};
const std::vector<binary_input> inputs =
{
{"CBOR", json::input_format_t::cbor, json::to_cbor(values)},
{"CBOR doubles", json::input_format_t::cbor, json::to_cbor(doubles)},
{"MessagePack", json::input_format_t::msgpack, json::to_msgpack(values)},
{"MessagePack doubles", json::input_format_t::msgpack, json::to_msgpack(doubles)},
{"UBJSON", json::input_format_t::ubjson, json::to_ubjson(values, true)},
{"UBJSON typed doubles", json::input_format_t::ubjson, json::to_ubjson(doubles, true, true)},
{"BJData", json::input_format_t::bjdata, json::to_bjdata(values, true)},
{"BJData typed doubles", json::input_format_t::bjdata, json::to_bjdata(doubles, true, true)},
};
for (const auto& input : inputs)
{
const std::string name = input.name;
CAPTURE(name)
const std::list<std::uint8_t> bytewise(input.data.begin(), input.data.end());
narrow_event_recorder from_contiguous;
narrow_event_recorder from_bytewise;
CHECK_FALSE(narrow_json::sax_parse(input.data, &from_contiguous, input.format));
CHECK_FALSE(narrow_json::sax_parse(bytewise.begin(), bytewise.end(), &from_bytewise, input.format));
CHECK(from_contiguous.events == from_bytewise.events);
REQUIRE(!from_contiguous.events.empty());
CHECK(from_contiguous.events.back().find("out_of_range.406") != std::string::npos);
}
}
-22
View File
@@ -483,25 +483,3 @@ TEST_CASE("custom number types: 32-bit integers in the binary formats")
CHECK(json32::to_msgpack(values) == nlohmann::json::to_msgpack(values64));
CHECK(json32::to_ubjson(values) == nlohmann::json::to_ubjson(values64));
}
TEST_CASE("custom number types: binary subtypes wider than the integer types")
{
// the subtype of a binary value is a std::uint64_t, whatever the number types are
using json8 = nlohmann::json::with_integers_t<std::int8_t, std::uint8_t>;
using json16 = nlohmann::json::with_integers_t<std::int16_t, std::uint16_t>;
using json32 = nlohmann::json::with_integers_t<std::int32_t, std::uint32_t>;
const std::uint64_t max_uint64 = (std::numeric_limits<std::uint64_t>::max)();
CHECK(json8::binary({1}, 300).dump() == "{\"bytes\":[1],\"subtype\":300}");
CHECK(json8::binary({200}, 255).dump() == "{\"bytes\":[200],\"subtype\":255}");
CHECK(json16::binary({1}, 65543).dump() == "{\"bytes\":[1],\"subtype\":65543}");
CHECK(json32::binary({1}, 4294967301u).dump() == "{\"bytes\":[1],\"subtype\":4294967301}");
CHECK(json32::binary({1}, max_uint64).dump() == "{\"bytes\":[1],\"subtype\":18446744073709551615}");
CHECK(json32::binary({1}, 0).dump() == "{\"bytes\":[1],\"subtype\":0}");
CHECK(json32::binary({1}, 4294967301u).dump(1) == "{\n \"bytes\": [1],\n \"subtype\": 4294967301\n}");
CHECK(json8::binary({1}).dump() == "{\"bytes\":[1],\"subtype\":null}");
// unchanged for the default number types
CHECK(nlohmann::json::binary({1}, 4294967301u).dump() == "{\"bytes\":[1],\"subtype\":4294967301}");
CHECK(nlohmann::json::binary({1}, max_uint64).dump(1) == "{\n \"bytes\": [1],\n \"subtype\": 18446744073709551615\n}");
}