#5597 was merged before all of its CI jobs had run, and two of them fail
on develop now, and so on every pull request:
- ci_clang_tidy: cert-err33-c for the two std::setlocale(LC_NUMERIC, "C")
calls whose result was discarded. Check the result, like the other
resets in the file.
- ci_test_standards_gcc (20) with GCC 16: -Wnoexcept for the two parser
callbacks, which cannot throw but were not declared noexcept.
Signed-off-by: Niels Lohmann <mail@nlohmann.me>
* Look up the locale decimal point at conversion time, not lexer construction
The lexer read localeconv()->decimal_point once in its constructor and wrote
that character into token_buffer in place of '.'. The strtod fallback then
used the locale current at conversion time, so an LC_NUMERIC change in
between (parser callback, SAX handler, another thread) truncated the value
in release builds and fired the endptr assertion in debug builds.
token_buffer now always holds '.'. Only the strtof/strtod/strtold fallback
depends on the locale: it looks up the decimal point right before the call,
restores '.' afterwards, and repeats the conversion if the locale changed in
between. As a side effect, std::from_chars and Clinger's fast path now also
apply under locales whose decimal point is not '.'.
Fixes#5198
Signed-off-by: Niels Lohmann <mail@nlohmann.me>
* Stop the strtod retry loop when the decimal point is unchanged
convert_float_locale_aware() repeated the conversion until strtod
consumed the whole token, assuming an early stop can only mean a locale
change. Under a locale whose decimal point is not a single character
(e.g. the two-byte U+066B of ar_EG.UTF-8, ar_SA.UTF-8, or fa_IR.UTF-8,
all available on macOS), the in-place substitution can never succeed,
so parsing any float that reaches the strtod fallback (for example
3.14159265358979323846 at C++11) hung forever. Before this branch, the
same input was truncated.
Retry only if the decimal point changed since the previous attempt;
otherwise keep the value strtod parsed so far, as before. Add a test
that parses such numbers under a multi-byte decimal point locale; it
hangs without this change.
Signed-off-by: Niels Lohmann <mail@nlohmann.me>
* Fix -Weffc++ errors in the #5198 locale test
GCC's -Weffc++ (an error in ci_test_gcc and ci_test_standards_gcc)
rejected LocaleSwitchingSax: it has a pointer data member but does not
declare its copy operations, and its vectors are not initialized in the
member initializer list. Store the locale name as a std::string and give
the vectors brace initializers, like SaxEventLogger in
unit-deserialization.cpp.
Signed-off-by: Niels Lohmann <mail@nlohmann.me>
---------
Signed-off-by: Niels Lohmann <mail@nlohmann.me>
* Add tests for uncovered code paths
Cover code the test suite did not reach, found from the Coveralls report
of develop and a local coverage run of HEAD:
- dump() of every kind of value below the bound of the recursive descent
(pretty-printed objects, binary values, discarded values, scalars), and
flushes of the escape and write buffers mid-string and mid-binary
- the iterative comparison: objects with different keys, containers that
are a prefix of each other, and elements that cannot be ordered, each
both at the top level and below the nesting bound
- SAX handlers that stop at any event, including the end of a nested
container, in the BSON, CBOR, MessagePack, UBJSON and BJData readers
- from_bson/cbor/msgpack/ubjson/bjdata returning a discarded value
through the iterator and pointer overloads
- JSON Patch, diff, merge_patch and update(..., true) on ordered_json
- smaller gaps: get_allocator(), to_ubjson/to_bjdata into a string,
value() with an unresolvable JSON pointer, integer/float comparison
below the integer range and with negative fractions, conversion to a
custom binary type, std::formatter::parse on a spec without '}',
unescape() of a lone '~', and the callback parser's start_array()
Signed-off-by: Niels Lohmann <mail@nlohmann.me>
* Cover more paths that were thought unreachable
- parse_float_fast() declining malformed or inexact input, called
directly since the lexer only passes well-formed numbers to it
- a UTF-16 high surrogate followed by a unit above the low surrogates
- self-assignment of a const_iterator
- a truncated CBOR string read through non-contiguous iterators
- serializing a long double under the de_DE locale, which undoes the
locale's decimal point and thousands separator
- values read from a binary format carrying no diagnostic positions,
with and without a parser callback
Signed-off-by: Niels Lohmann <mail@nlohmann.me>
* Fix the CI failures of the new coverage tests
- declare the self-assignment reference const (misc-const-correctness)
- expect the (/path) prefix that JSON_DIAGNOSTICS adds to the messages
of the failing ordered_json patch operations
Signed-off-by: Niels Lohmann <mail@nlohmann.me>
* Expect the byte range JSON_DIAGNOSTIC_POSITIONS adds to the patch errors
Signed-off-by: Niels Lohmann <mail@nlohmann.me>
* Build the expected dump of the nested-object test with +=
clang-tidy (performance-inefficient-string-concatenation) reported the
chain of operator+ calls that assembled the expected indented output.
Signed-off-by: Niels Lohmann <mail@nlohmann.me>
* Compare the BJData and UBJSON test outputs byte by byte
Building a std::string from the byte vector converts each byte
implicitly, which -fsanitize=integer reports for bytes of 0x80 and
above (ci_test_clang_sanitizer).
Signed-off-by: Niels Lohmann <mail@nlohmann.me>
---------
Signed-off-by: Niels Lohmann <mail@nlohmann.me>
Niels Lohmann