Compare commits

..
Author SHA1 Message Date
Niels Lohmann cc749ac699 Keep full byte2 x byte3 combinatorics in the wrong-fourth-byte sections
The maintainer wants exhaustive coverage of every byte combination here
rather than the representative-prefix reduction, matching the style of
the sibling "wrong second/third byte" sections in the same files.

Signed-off-by: Niels Lohmann <mail@nlohmann.me>
2026-09-05 23:18:51 +02:00
Niels Lohmann 1c6ca81b8a Fix dead ill-formed-fourth-byte UTF-8 test sections (byte3/byte4 typo)
The "ill-formed: wrong fourth byte" SECTIONs in unit-unicode3.cpp,
unit-unicode4.cpp, and unit-unicode5.cpp guarded their loop with a check
on byte3 instead of byte4. Since the enclosing loop already restricts
byte3 to its valid range, the guard was always true and the section's
"continue" fired unconditionally, so check_utf8string()/check_utf8dump()
were never actually invoked for a malformed fourth byte.

Fixing the guard naively (byte3 -> byte4) would also have swept the full
byte2 x byte3 combinatorics for every byte4 value, adding millions of
redundant iterations: the lexer validates continuation bytes strictly in
sequence with early exit (see next_byte_in_range() in lexer.hpp), so once
byte2/byte3 are within their valid range, the byte4 outcome does not
depend on which valid byte2/byte3 values were chosen. Instead, byte2 and
byte3 are now held to a small hedge of representative valid prefixes
(range corners plus a midpoint) while byte4 is still swept exhaustively
over its full 0x00-0xFF range, since that is the actual property under
test. Also fixed the garbled "skip fourth second byte" comment in
unit-unicode3.cpp.

Verified offline: before the fix, the "wrong fourth byte" subcase
executes 0 assertions in all three files (proving it was dead code);
after the fix, it executes 11520 (unicode3), 34560 (unicode4), and 11520
(unicode5) assertions, and a deliberately reintroduced bug in the
lexer's byte4 range check causes it to fail (proving it is now
meaningful). Total per-file assertion counts grow by the same small
amounts, not by millions, and all other sections in these files still
pass unchanged.

Fixes #5416

Signed-off-by: Niels Lohmann <mail@nlohmann.me>
2026-09-05 23:05:42 +02:00
6 changed files with 4 additions and 187 deletions
@@ -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
*/
@@ -1745,60 +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);
-68
View File
@@ -18655,20 +18655,6 @@ 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
*/
@@ -18753,60 +18739,6 @@ 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);
-47
View File
@@ -2776,53 +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}));
}
}
}
+2 -2
View File
@@ -304,8 +304,8 @@ TEST_CASE("Unicode (3/5)" * doctest::skip())
{
for (int byte4 = 0x00; byte4 <= 0xFF; ++byte4)
{
// skip fourth second byte
if (0x80 <= byte3 && byte3 <= 0xBF)
// skip correct fourth byte
if (0x80 <= byte4 && byte4 <= 0xBF)
{
continue;
}
+1 -1
View File
@@ -305,7 +305,7 @@ TEST_CASE("Unicode (4/5)" * doctest::skip())
for (int byte4 = 0x00; byte4 <= 0xFF; ++byte4)
{
// skip correct fourth byte
if (0x80 <= byte3 && byte3 <= 0xBF)
if (0x80 <= byte4 && byte4 <= 0xBF)
{
continue;
}
+1 -1
View File
@@ -305,7 +305,7 @@ TEST_CASE("Unicode (5/5)" * doctest::skip())
for (int byte4 = 0x00; byte4 <= 0xFF; ++byte4)
{
// skip correct fourth byte
if (0x80 <= byte3 && byte3 <= 0xBF)
if (0x80 <= byte4 && byte4 <= 0xBF)
{
continue;
}