Replace use of locale dependent atoi(…) with locale-independent std::from_chars(…) (C++17)

test: Add test cases for LocaleIndependentAtoi

fuzz: Assert legacy atoi(s) == LocaleIndependentAtoi<int>(s)

fuzz: Assert legacy atoi64(s) == LocaleIndependentAtoi<int64_t>(s)
This commit is contained in:
practicalswift
2021-09-30 14:18:50 +00:00
parent 2d8e0c0c3c
commit 4343f114cc
15 changed files with 145 additions and 52 deletions

View File

@@ -50,8 +50,6 @@ FUZZ_TARGET(locale)
const bool parseint32_without_locale = ParseInt32(random_string, &parseint32_out_without_locale);
int64_t parseint64_out_without_locale;
const bool parseint64_without_locale = ParseInt64(random_string, &parseint64_out_without_locale);
const int64_t atoi64_without_locale = atoi64(random_string);
const int atoi_without_locale = atoi(random_string);
const int64_t random_int64 = fuzzed_data_provider.ConsumeIntegral<int64_t>();
const std::string tostring_without_locale = ToString(random_int64);
// The variable `random_int32` is no longer used, but the harness still needs to
@@ -77,10 +75,6 @@ FUZZ_TARGET(locale)
if (parseint64_without_locale) {
assert(parseint64_out_without_locale == parseint64_out_with_locale);
}
const int64_t atoi64_with_locale = atoi64(random_string);
assert(atoi64_without_locale == atoi64_with_locale);
const int atoi_with_locale = atoi(random_string);
assert(atoi_without_locale == atoi_with_locale);
const std::string tostring_with_locale = ToString(random_int64);
assert(tostring_without_locale == tostring_with_locale);
const std::string strprintf_int_with_locale = strprintf("%d", random_int64);

View File

@@ -25,13 +25,13 @@ FUZZ_TARGET(parse_numbers)
int32_t i32;
(void)ParseInt32(random_string, &i32);
(void)atoi(random_string);
(void)LocaleIndependentAtoi<int>(random_string);
uint32_t u32;
(void)ParseUInt32(random_string, &u32);
int64_t i64;
(void)atoi64(random_string);
(void)LocaleIndependentAtoi<int64_t>(random_string);
(void)ParseFixedPoint(random_string, 3, &i64);
(void)ParseInt64(random_string, &i64);

View File

@@ -122,6 +122,12 @@ bool LegacyParseUInt64(const std::string& str, uint64_t* out)
return endp && *endp == 0 && !errno &&
n <= std::numeric_limits<uint64_t>::max();
}
// For backwards compatibility checking.
int64_t atoi64_legacy(const std::string& str)
{
return strtoll(str.c_str(), nullptr, 10);
}
}; // namespace
FUZZ_TARGET(string)
@@ -268,4 +274,22 @@ FUZZ_TARGET(string)
assert(u8 == u8_legacy);
}
}
{
const int atoi_result = atoi(random_string_1.c_str());
const int locale_independent_atoi_result = LocaleIndependentAtoi<int>(random_string_1);
const int64_t atoi64_result = atoi64_legacy(random_string_1);
const bool out_of_range = atoi64_result < std::numeric_limits<int>::min() || atoi64_result > std::numeric_limits<int>::max();
if (out_of_range) {
assert(locale_independent_atoi_result == 0);
} else {
assert(atoi_result == locale_independent_atoi_result);
}
}
{
const int64_t atoi64_result = atoi64_legacy(random_string_1);
const int64_t locale_independent_atoi_result = LocaleIndependentAtoi<int64_t>(random_string_1);
assert(atoi64_result == locale_independent_atoi_result || locale_independent_atoi_result == 0);
}
}

View File

@@ -1549,6 +1549,77 @@ BOOST_AUTO_TEST_CASE(test_ToIntegral)
BOOST_CHECK(!ToIntegral<uint8_t>("256"));
}
BOOST_AUTO_TEST_CASE(test_LocaleIndependentAtoi)
{
BOOST_CHECK_EQUAL(LocaleIndependentAtoi<int32_t>("1234"), 1'234);
BOOST_CHECK_EQUAL(LocaleIndependentAtoi<int32_t>("0"), 0);
BOOST_CHECK_EQUAL(LocaleIndependentAtoi<int32_t>("01234"), 1'234);
BOOST_CHECK_EQUAL(LocaleIndependentAtoi<int32_t>("-1234"), -1'234);
BOOST_CHECK_EQUAL(LocaleIndependentAtoi<int32_t>(" 1"), 1);
BOOST_CHECK_EQUAL(LocaleIndependentAtoi<int32_t>("1 "), 1);
BOOST_CHECK_EQUAL(LocaleIndependentAtoi<int32_t>("1a"), 1);
BOOST_CHECK_EQUAL(LocaleIndependentAtoi<int32_t>("1.1"), 1);
BOOST_CHECK_EQUAL(LocaleIndependentAtoi<int32_t>("1.9"), 1);
BOOST_CHECK_EQUAL(LocaleIndependentAtoi<int32_t>("+01.9"), 1);
BOOST_CHECK_EQUAL(LocaleIndependentAtoi<int32_t>("-1"), -1);
BOOST_CHECK_EQUAL(LocaleIndependentAtoi<int32_t>(" -1"), -1);
BOOST_CHECK_EQUAL(LocaleIndependentAtoi<int32_t>("-1 "), -1);
BOOST_CHECK_EQUAL(LocaleIndependentAtoi<int32_t>(" -1 "), -1);
BOOST_CHECK_EQUAL(LocaleIndependentAtoi<int32_t>("+1"), 1);
BOOST_CHECK_EQUAL(LocaleIndependentAtoi<int32_t>(" +1"), 1);
BOOST_CHECK_EQUAL(LocaleIndependentAtoi<int32_t>(" +1 "), 1);
BOOST_CHECK_EQUAL(LocaleIndependentAtoi<int32_t>("+-1"), 0);
BOOST_CHECK_EQUAL(LocaleIndependentAtoi<int32_t>("-+1"), 0);
BOOST_CHECK_EQUAL(LocaleIndependentAtoi<int32_t>("++1"), 0);
BOOST_CHECK_EQUAL(LocaleIndependentAtoi<int32_t>("--1"), 0);
BOOST_CHECK_EQUAL(LocaleIndependentAtoi<int32_t>(""), 0);
BOOST_CHECK_EQUAL(LocaleIndependentAtoi<int32_t>("aap"), 0);
BOOST_CHECK_EQUAL(LocaleIndependentAtoi<int32_t>("0x1"), 0);
BOOST_CHECK_EQUAL(LocaleIndependentAtoi<int32_t>("-32482348723847471234"), 0);
BOOST_CHECK_EQUAL(LocaleIndependentAtoi<int32_t>("32482348723847471234"), 0);
BOOST_CHECK_EQUAL(LocaleIndependentAtoi<int64_t>("-9223372036854775809"), 0);
BOOST_CHECK_EQUAL(LocaleIndependentAtoi<int64_t>("-9223372036854775808"), -9'223'372'036'854'775'807LL - 1LL);
BOOST_CHECK_EQUAL(LocaleIndependentAtoi<int64_t>("9223372036854775807"), 9'223'372'036'854'775'807);
BOOST_CHECK_EQUAL(LocaleIndependentAtoi<int64_t>("9223372036854775808"), 0);
BOOST_CHECK_EQUAL(LocaleIndependentAtoi<uint64_t>("-1"), 0U);
BOOST_CHECK_EQUAL(LocaleIndependentAtoi<uint64_t>("0"), 0U);
BOOST_CHECK_EQUAL(LocaleIndependentAtoi<uint64_t>("18446744073709551615"), 18'446'744'073'709'551'615ULL);
BOOST_CHECK_EQUAL(LocaleIndependentAtoi<uint64_t>("18446744073709551616"), 0U);
BOOST_CHECK_EQUAL(LocaleIndependentAtoi<int32_t>("-2147483649"), 0);
BOOST_CHECK_EQUAL(LocaleIndependentAtoi<int32_t>("-2147483648"), -2'147'483'648LL);
BOOST_CHECK_EQUAL(LocaleIndependentAtoi<int32_t>("2147483647"), 2'147'483'647);
BOOST_CHECK_EQUAL(LocaleIndependentAtoi<int32_t>("2147483648"), 0);
BOOST_CHECK_EQUAL(LocaleIndependentAtoi<uint32_t>("-1"), 0U);
BOOST_CHECK_EQUAL(LocaleIndependentAtoi<uint32_t>("0"), 0U);
BOOST_CHECK_EQUAL(LocaleIndependentAtoi<uint32_t>("4294967295"), 4'294'967'295U);
BOOST_CHECK_EQUAL(LocaleIndependentAtoi<uint32_t>("4294967296"), 0U);
BOOST_CHECK_EQUAL(LocaleIndependentAtoi<int16_t>("-32769"), 0);
BOOST_CHECK_EQUAL(LocaleIndependentAtoi<int16_t>("-32768"), -32'768);
BOOST_CHECK_EQUAL(LocaleIndependentAtoi<int16_t>("32767"), 32'767);
BOOST_CHECK_EQUAL(LocaleIndependentAtoi<int16_t>("32768"), 0);
BOOST_CHECK_EQUAL(LocaleIndependentAtoi<uint16_t>("-1"), 0U);
BOOST_CHECK_EQUAL(LocaleIndependentAtoi<uint16_t>("0"), 0U);
BOOST_CHECK_EQUAL(LocaleIndependentAtoi<uint16_t>("65535"), 65'535U);
BOOST_CHECK_EQUAL(LocaleIndependentAtoi<uint16_t>("65536"), 0U);
BOOST_CHECK_EQUAL(LocaleIndependentAtoi<int8_t>("-129"), 0);
BOOST_CHECK_EQUAL(LocaleIndependentAtoi<int8_t>("-128"), -128);
BOOST_CHECK_EQUAL(LocaleIndependentAtoi<int8_t>("127"), 127);
BOOST_CHECK_EQUAL(LocaleIndependentAtoi<int8_t>("128"), 0);
BOOST_CHECK_EQUAL(LocaleIndependentAtoi<uint8_t>("-1"), 0U);
BOOST_CHECK_EQUAL(LocaleIndependentAtoi<uint8_t>("0"), 0U);
BOOST_CHECK_EQUAL(LocaleIndependentAtoi<uint8_t>("255"), 255U);
BOOST_CHECK_EQUAL(LocaleIndependentAtoi<uint8_t>("256"), 0U);
}
BOOST_AUTO_TEST_CASE(test_ParseInt64)
{
int64_t n;