shylie/llvm-project
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity
llvm-project/libcxx/test/std/utilities/charconv/charconv.msvc/test.cpp
Mark de Wever a8025e06fc Microsoft's floating-point to_chars powered by Ryu and Ryu Printf 
Microsoft would like to contribute its implementation of floating-point to_chars to libc++. This uses the impossibly fast Ryu and Ryu Printf algorithms invented by Ulf Adams at Google. Upstream repos: https://github.com/microsoft/STL and https://github.com/ulfjack/ryu .

Licensing notes: MSVC's STL is available under the Apache License v2.0 with LLVM Exception, intentionally chosen to match libc++. We've used Ryu under the Boost Software License.

This patch contains minor changes from Jorg Brown at Google, to adapt the code to libc++. He verified that it works in Google's Linux-based environment, but then I applied more changes on top of his, so any compiler errors are my fault. (I haven't tried to build and test libc++ yet.) Please tell me if we need to do anything else in order to follow https://llvm.org/docs/DeveloperPolicy.html#attribution-of-changes .

Notes:

* libc++'s integer charconv is unchanged (except for a small refactoring). MSVC's integer charconv hasn't been tuned for performance yet, so you're not missing anything.
* Floating-point from_chars isn't part of this patch because Jorg found that MSVC's implementation (derived from our CRT's strtod) was slower than Abseil's. If you're unable to use Abseil or another implementation due to licensing or technical considerations, Microsoft would be delighted if you used MSVC's from_chars (and you can just take it, or ask us to provide a patch like this). Ulf is also working on a novel algorithm for from_chars.
* This assumes that float is IEEE 32-bit, double is IEEE 64-bit, and long double is also IEEE 64-bit.
* I have added MSVC's charconv tests (the whole thing: integer/floating from_chars/to_chars), but haven't adapted them to libcxx's harness at all. (These tests will be available in the microsoft/STL repo soon.)
* Jorg added int128 codepaths. These were originally present in upstream Ryu, and I removed them from microsoft/STL purely for performance reasons (MSVC doesn't support int128; Clang on Windows does, but I found that x64 intrinsics were slightly faster).
* The implementation is split into 3 headers. In MSVC's STL, charconv contains only Microsoft-written code. xcharconv_ryu.h contains code derived from Ryu (with significant modifications and additions). xcharconv_ryu_tables.h contains Ryu's large lookup tables (they were sufficiently large to make editing inconvenient, hence the separate file). The xmeow.h convention is MSVC's for internal headers; you may wish to rename them.
* You should consider separately compiling the lookup tables (see https://github.com/microsoft/STL/issues/172 ) for compiler throughput and reduced object file size.
* See https://github.com/StephanTLavavej/llvm-project/commits/charconv for fine-grained history. (If necessary, I can perform some rebase surgery to show you what Jorg changed relative to the microsoft/STL repo; currently that's all fused into the first commit.)

Differential Revision: https://reviews.llvm.org/D70631
2021-12-05 13:25:33 +01:00

1105 lines
58 KiB
C++
Raw Blame History

// Copyright (c) Microsoft Corporation.
// SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
#include "test.hpp"
#include <algorithm>
#include <array>
#include <assert.h>
#include <charconv>
#include <chrono>
#include <cmath>
#include <fstream>
#include <functional>
#include <limits>
#include <locale>
#include <optional>
#include <random>
#include <set>
#include <stdint.h>
#include <stdio.h>
#include <stdlib.h>
#include <string.h>
#include <string>
#include <string_view>
#include <system_error>
#include <type_traits>
#include <utility>
#include <vector>
#include "double_fixed_precision_to_chars_test_cases_1.hpp"
#include "double_fixed_precision_to_chars_test_cases_2.hpp"
#include "double_fixed_precision_to_chars_test_cases_3.hpp"
#include "double_fixed_precision_to_chars_test_cases_4.hpp"
#include "double_from_chars_test_cases.hpp"
#include "double_general_precision_to_chars_test_cases.hpp"
#include "double_hex_precision_to_chars_test_cases.hpp"
#include "double_scientific_precision_to_chars_test_cases_1.hpp"
#include "double_scientific_precision_to_chars_test_cases_2.hpp"
#include "double_scientific_precision_to_chars_test_cases_3.hpp"
#include "double_scientific_precision_to_chars_test_cases_4.hpp"
#include "double_to_chars_test_cases.hpp"
#include "float_fixed_precision_to_chars_test_cases.hpp"
#include "float_from_chars_test_cases.hpp"
#include "float_general_precision_to_chars_test_cases.hpp"
#include "float_hex_precision_to_chars_test_cases.hpp"
#include "float_scientific_precision_to_chars_test_cases.hpp"
#include "float_to_chars_test_cases.hpp"
using namespace std;
void initialize_randomness(mt19937_64& mt64, const int argc, char** const argv) {
constexpr size_t n = mt19937_64::state_size;
constexpr size_t w = mt19937_64::word_size;
static_assert(w % 32 == 0);
constexpr size_t k = w / 32;
vector<uint32_t> vec(n * k);
puts("USAGE:");
puts("test.exe : generates seed data from random_device.");
puts("test.exe filename.txt : loads seed data from a given text file.");
if (argc == 1) {
random_device rd;
generate(vec.begin(), vec.end(), ref(rd));
puts("Generated seed data.");
} else if (argc == 2) {
const char* const filename = argv[1];
ifstream file(filename);
if (!file) {
printf("ERROR: Can't open %s.\n", filename);
abort();
}
for (auto& elem : vec) {
file >> elem;
if (!file) {
printf("ERROR: Can't read seed data from %s.\n", filename);
abort();
}
}
printf("Loaded seed data from %s.\n", filename);
} else {
puts("ERROR: Too many command-line arguments.");
abort();
}
puts("SEED DATA:");
for (const auto& elem : vec) {
printf("%u ", elem);
}
printf("\n");
seed_seq seq(vec.cbegin(), vec.cend());
mt64.seed(seq);
puts("Successfully seeded mt64. First three values:");
for (int i = 0; i < 3; ++i) {
// libc++ uses long for 64-bit values.
printf("0x%016llX\n", static_cast<unsigned long long>(mt64()));
}
}
static_assert((chars_format::scientific & chars_format::fixed) == chars_format{});
static_assert((chars_format::scientific & chars_format::hex) == chars_format{});
static_assert((chars_format::fixed & chars_format::hex) == chars_format{});
static_assert(chars_format::general == (chars_format::fixed | chars_format::scientific));
template <typename T, typename Optional>
void test_common_to_chars(
const T value, const Optional opt_arg, const optional<int> opt_precision, const string_view correct) {
// Important: Test every effective buffer size from 0 through correct.size() and slightly beyond. For the sizes
// less than correct.size(), this verifies that the too-small buffer is correctly detected, and that we don't
// attempt to write outside of it, even by a single char. (This exhaustive validation is necessary because the
// implementation must check whenever it attempts to write. Sometimes we can calculate the total size and perform
// a single check, but sometimes we need to check when writing each part of the result.) Testing correct.size()
// verifies that we can succeed without overrunning, and testing slightly larger sizes verifies that we can succeed
// without attempting to write to extra chars even when they're available. Finally, we also verify that we aren't
// underrunning the buffer. This is a concern because sometimes we walk backwards when rounding.
constexpr size_t BufferPrefix = 20; // detect buffer underruns (specific value isn't important)
constexpr size_t Space = is_integral_v<T> ? 1 + 64 // worst case: -2^63 in binary
: is_same_v<T, float>
? 1 + 151 // worst case: negative min subnormal float, fixed notation
: 1 + 1076; // worst case: negative min subnormal double, fixed notation
constexpr size_t BufferSuffix = 30; // detect buffer overruns (specific value isn't important)
array<char, BufferPrefix + Space + BufferSuffix> buff;
char* const buff_begin = buff.data();
char* const first = buff_begin + BufferPrefix;
char* const buff_end = buff_begin + buff.size();
constexpr size_t ExtraChars = 3;
static_assert(ExtraChars + 10 < BufferSuffix,
"The specific values aren't important, but there should be plenty of room to detect buffer overruns.");
for (size_t n = 0; n <= correct.size() + ExtraChars; ++n) {
assert(n <= static_cast<size_t>(buff_end - first));
char* const last = first + n;
buff.fill('@');
const auto is_fill_char = [](const char c) { return c == '@'; };
to_chars_result result{};
if (opt_precision.has_value()) {
assert(opt_arg.has_value());
if constexpr (is_floating_point_v<T>) {
result = to_chars(first, last, value, opt_arg.value(), opt_precision.value());
} else {
abort();
}
} else if (opt_arg.has_value()) {
result = to_chars(first, last, value, opt_arg.value());
} else {
result = to_chars(first, last, value);
}
if (n < correct.size()) {
assert(result.ptr == last);
assert(result.ec == errc::value_too_large);
assert(all_of(buff_begin, first, is_fill_char));
// [first, last) is unspecified
assert(all_of(last, buff_end, is_fill_char));
} else {
assert(result.ptr == first + correct.size());
assert(result.ec == errc{});
assert(all_of(buff_begin, first, is_fill_char));
assert(equal(first, result.ptr, correct.begin(), correct.end()));
assert(all_of(result.ptr, buff_end, is_fill_char));
}
}
}
template <typename T>
void test_integer_to_chars(const T value, const optional<int> opt_base, const string_view correct) {
test_common_to_chars(value, opt_base, nullopt, correct);
{ // Also test successful from_chars() scenarios.
const char* const correct_first = correct.data();
const char* const correct_last = correct_first + correct.size();
T dest = 0;
const from_chars_result from_res =
(opt_base.has_value() ? from_chars(correct_first, correct_last, dest, opt_base.value())
: from_chars(correct_first, correct_last, dest));
assert(from_res.ptr == correct_last);
assert(from_res.ec == errc{});
assert(dest == value);
}
}
// https://www.wolframalpha.com : Table[BaseForm[n * 2 - 1, n], {n, 2, 36}]
constexpr const char* output_max_digit[] = {"skip0", "skip1", "11", "12", "13", "14", "15", "16", "17", "18", "19",
"1a", "1b", "1c", "1d", "1e", "1f", "1g", "1h", "1i", "1j", "1k", "1l", "1m", "1n", "1o", "1p", "1q", "1r", "1s",
"1t", "1u", "1v", "1w", "1x", "1y", "1z"};
// https://www.wolframalpha.com : Table[BaseForm[k, n], {k, {MEOW, MEOW, MEOW}}, {n, 2, 36}]
constexpr uint64_t stress_chunks_positive = 12000000345000678900ULL;
constexpr pair<uint64_t, array<const char*, 37>> output_positive[] = {
{123U, {{"skip0", "skip1", "1111011", "11120", "1323", "443", "323", "234", "173", "146", "123", "102", "a3", "96",
"8b", "83", "7b", "74", "6f", "69", "63", "5i", "5d", "58", "53", "4n", "4j", "4f", "4b", "47", "43",
"3u", "3r", "3o", "3l", "3i", "3f"}}},
{uint64_t{INT8_MAX}, {{"skip0", "skip1", "1111111", "11201", "1333", "1002", "331", "241", "177", "151", "127",
"106", "a7", "9a", "91", "87", "7f", "78", "71", "6d", "67", "61", "5h", "5c", "57", "52",
"4n", "4j", "4f", "4b", "47", "43", "3v", "3s", "3p", "3m", "3j"}}},
{161U, {{"skip0", "skip1", "10100001", "12222", "2201", "1121", "425", "320", "241", "188", "161", "137", "115",
"c5", "b7", "ab", "a1", "98", "8h", "89", "81", "7e", "77", "70", "6h", "6b", "65", "5q", "5l", "5g",
"5b", "56", "51", "4t", "4p", "4l", "4h"}}},
{UINT8_MAX, {{"skip0", "skip1", "11111111", "100110", "3333", "2010", "1103", "513", "377", "313", "255", "212",
"193", "168", "143", "120", "ff", "f0", "e3", "d8", "cf", "c3", "bd", "b2", "af", "a5", "9l", "9c",
"93", "8n", "8f", "87", "7v", "7o", "7h", "7a", "73"}}},
{1729U, {{"skip0", "skip1", "11011000001", "2101001", "123001", "23404", "12001", "5020", "3301", "2331", "1729",
"1332", "1001", "a30", "8b7", "7a4", "6c1", "5gc", "561", "4f0", "469", "3j7", "3cd", "364", "301",
"2j4", "2ed", "2a1", "25l", "21i", "1rj", "1oo", "1m1", "1jd", "1gt", "1ee", "1c1"}}},
{uint64_t{INT16_MAX}, {{"skip0", "skip1", "111111111111111", "1122221121", "13333333", "2022032", "411411",
"164350", "77777", "48847", "32767", "22689", "16b67", "11bb7", "bd27", "9a97", "7fff",
"6b68", "5b27", "4eeb", "41i7", "3b67", "31f9", "2flf", "28l7", "22ah", "1mc7", "1hpg",
"1dm7", "19rq", "16c7", "1330", "vvv", "u2v", "sbp", "qq7", "pa7"}}},
{57494U, {{"skip0", "skip1", "1110000010010110", "2220212102", "32002112", "3314434", "1122102", "326423", "160226",
"86772", "57494", "3a218", "29332", "20228", "16d4a", "1207e", "e096", "bbg0", "9f82", "8750", "73ee",
"647h", "58h8", "4gfh", "43je", "3goj", "3718", "2onb", "2h9a", "2aag", "23qe", "1spk", "1o4m", "1jq8",
"1fp0", "1bwo", "18d2"}}},
{UINT16_MAX, {{"skip0", "skip1", "1111111111111111", "10022220020", "33333333", "4044120", "1223223", "362031",
"177777", "108806", "65535", "45268", "31b13", "23aa2", "19c51", "14640", "ffff", "d5d0", "b44f",
"9aa4", "83gf", "71cf", "638j", "58k8", "4hif", "44la", "3iof", "38o6", "2rgf", "2jqo", "2cof",
"2661", "1vvv", "1r5u", "1mnh", "1ihf", "1ekf"}}},
{71125478U, {{"skip0", "skip1", "100001111010100100111100110", "11221211112210222", "10033110213212",
"121202003403", "11020244342", "1522361624", "417244746", "157745728", "71125478", "3716a696",
"1b9a06b2", "11973ba8", "9636514", "639e338", "43d49e6", "2g19gfb", "21b9d18", "19dec94", "124addi",
"h8f25b", "dhdfa6", "b13hg2", "8m91he", "7720j3", "5pgj58", "4pmelq", "43k17i", "3dg8ek", "2ro898",
"2f0et8", "23qif6", "1qw5lh", "1j7l7s", "1cdvli", "16cgrq"}}},
{uint64_t{INT32_MAX},
{{"skip0", "skip1", "1111111111111111111111111111111", "12112122212110202101", "1333333333333333",
"13344223434042", "553032005531", "104134211161", "17777777777", "5478773671", "2147483647", "a02220281",
"4bb2308a7", "282ba4aaa", "1652ca931", "c87e66b7", "7fffffff", "53g7f548", "3928g3h1", "27c57h32",
"1db1f927", "140h2d91", "ikf5bf1", "ebelf95", "b5gge57", "8jmdnkm", "6oj8ion", "5ehncka", "4clm98f",
"3hk7987", "2sb6cs7", "2d09uc1", "1vvvvvv", "1lsqtl1", "1d8xqrp", "15v22um", "zik0zj"}}},
{3522553278ULL,
{{"skip0", "skip1", "11010001111101011110010110111110", "100002111022020200020", "3101331132112332",
"24203233201103", "1341312313010", "153202131426", "32175362676", "10074266606", "3522553278", "1548431462",
"823842766", "441a34c6a", "255b8d486", "1593b4753", "d1f5e5be", "89ffb3b6", "5da3e606", "3hgbfb5i",
"2f0fj33i", "1k1ac536", "191b46e2", "10i6fmk8", "ia967l6", "eahia63", "baca9ga", "92d86i6", "78iq4i6",
"5qlc1dc", "4osos2i", "3u1862s", "38vbpdu", "2o0a7ro", "29hx9e6", "1w2dnod", "1m98ji6"}}},
{UINT32_MAX,
{{"skip0", "skip1", "11111111111111111111111111111111", "102002022201221111210", "3333333333333333",
"32244002423140", "1550104015503", "211301422353", "37777777777", "12068657453", "4294967295", "1904440553",
"9ba461593", "535a79888", "2ca5b7463", "1a20dcd80", "ffffffff", "a7ffda90", "704he7g3", "4f5aff65",
"3723ai4f", "281d55i3", "1fj8b183", "1606k7ib", "mb994af", "hek2mgk", "dnchbnl", "b28jpdl", "8pfgih3",
"76beigf", "5qmcpqf", "4q0jto3", "3vvvvvv", "3aokq93", "2qhxjlh", "2br45qa", "1z141z3"}}},
{545890816626160ULL,
{{"skip0", "skip1", "1111100000111110000011100001101100000110111110000", "2122120211122121121021010202111",
"1330013300130031200313300", "1033022333343024014120", "5213002440142255104", "222661211220253465",
"17407603415406760", "2576748547233674", "545890816626160", "148a34aa4706535", "51285369b87494",
"1a57a38b045a95", "98b3383b9766c", "4319d1601875a", "1f07c1c360df0", "ffd471f34f13", "88g09ff9dh84",
"4d0d5e232c53", "2d63h403i580", "1bf5h8185hdj", "kc3g550fkcg", "d41id5k9984", "8ef5n0him4g", "5i2dijfe1la",
"3me22fm5fhi", "2hfmhgg73kd", "1ngpfabr53c", "18i7220bh11", "rm0lcjngpa", "kk1elesni1", "fgfge3c3fg",
"bp4q5l6bjg", "8xna46jp0k", "6wejomvji5", "5di2s1qhv4"}}},
{uint64_t{INT64_MAX},
{{"skip0", "skip1", "111111111111111111111111111111111111111111111111111111111111111",
"2021110011022210012102010021220101220221", "13333333333333333333333333333333",
"1104332401304422434310311212", "1540241003031030222122211", "22341010611245052052300",
"777777777777777777777", "67404283172107811827", "9223372036854775807", "1728002635214590697",
"41a792678515120367", "10b269549075433c37", "4340724c6c71dc7a7", "160e2ad3246366807", "7fffffffffffffff",
"33d3d8307b214008", "16agh595df825fa7", "ba643dci0ffeehh", "5cbfjia3fh26ja7", "2heiciiie82dh97",
"1adaibb21dckfa7", "i6k448cf4192c2", "acd772jnc9l0l7", "64ie1focnn5g77", "3igoecjbmca687", "27c48l5b37oaop",
"1bk39f3ah3dmq7", "q1se8f0m04isb", "hajppbc1fc207", "bm03i95hia437", "7vvvvvvvvvvvv", "5hg4ck9jd4u37",
"3tdtk1v8j6tpp", "2pijmikexrxp7", "1y2p0ij32e8e7"}}},
{stress_chunks_positive,
{{"skip0", "skip1", "1010011010001000100100001011110000101100010101001001010111110100",
"2221221122020020011022001202200200202200", "22122020210023300230111021113310",
"1301130403021123030133211100", "2311004450342244200504500", "30325064311430214266301",
"1232104413605425112764", "87848206138052620680", "12000000345000678900", "2181782a1686924456a",
"54aa47a9058877b130", "150593a5b002c87b16", "571cad2b93c7760a8", "1c60d2676d4e53e00", "a68890bc2c5495f4",
"43499224707a4f4g", "1e052gdga1d26f40", "f06dh4g564c8a91", "769df0d9ace4h50", "3ee7bcj1ajghi4f",
"1k9agc4gfl0l43a", "10id7dakdlcjd22", "dge08fe0l5hl7c", "8184326d31ib60", "4ljbglf3cpim76",
"2pph66481kiiki", "1niph2ao132e58", "14qgbgk3c3iffg", "mhc35an1bhb00", "f78o8ur705ln5", "ad24gngm595fk",
"76e1n5i5v0ivl", "50wu8jsnks82g", "3ja41smfvqb1f", "2j64t3qgq0ut0"}}},
{14454900944617508688ULL,
{{"skip0", "skip1", "1100100010011010000111111101001011100011011000101000111101010000",
"10120022020112011211121221212101012220210", "30202122013331023203120220331100",
"1432224030234034034040234223", "3014532424232535441404120", "34610451042001242144165",
"1442320775134330507520", "116266464747855335823", "14454900944617508688", "266642a9a9471339935",
"662251403263939640", "1895280092bc310481", "68cb9c8292557406c", "23023deab20002893", "c89a1fd2e3628f50",
"50e7147a7db8ef84", "22a34a05086f78ec", "i1dgef04357g7i1", "8g90b882jcj8be8", "49c1kk35i0k24ic",
"272a16i54ebkacg", "15fdih7l3m7k8md", "gbj7303eg9nge0", "9hckfdkj3kkdmd", "5lc7hifdkl4nne",
"3f86e4mgpna5ol", "266pj428na273c", "1bomgjbnlg4m3f", "r5tf1f7f009ji", "iarsig29iqhhm", "ch6gvqbhm53qg",
"8lwtvcdj6rlqr", "61w23lajggp44", "49p1f3dsqqcdx", "31tkqqkxypopc"}}},
{UINT64_MAX,
{{"skip0", "skip1", "1111111111111111111111111111111111111111111111111111111111111111",
"11112220022122120101211020120210210211220", "33333333333333333333333333333333",
"2214220303114400424121122430", "3520522010102100444244423", "45012021522523134134601",
"1777777777777777777777", "145808576354216723756", "18446744073709551615", "335500516a429071284",
"839365134a2a240713", "219505a9511a867b72", "8681049adb03db171", "2c1d56b648c6cd110", "ffffffffffffffff",
"67979g60f5428010", "2d3fgb0b9cg4bd2f", "141c8786h1ccaagg", "b53bjh07be4dj0f", "5e8g4ggg7g56dif",
"2l4lf104353j8kf", "1ddh88h2782i515", "l12ee5fn0ji1if", "c9c336o0mlb7ef", "7b7n2pcniokcgf",
"4eo8hfam6fllmo", "2nc6j26l66rhof", "1n3rsh11f098rn", "14l9lkmo30o40f", "nd075ib45k86f", "fvvvvvvvvvvvv",
"b1w8p7j5q9r6f", "7orp63sh4dphh", "5g24a25twkwff", "3w5e11264sgsf"}}},
};
// https://www.wolframalpha.com : Table[BaseForm[k, n], {k, {MEOW, MEOW, MEOW}}, {n, 2, 36}]
constexpr int64_t stress_chunks_negative = -9000876000000054321LL;
constexpr pair<int64_t, array<const char*, 37>> output_negative[] = {
{-85, {{"skip0", "skip1", "-1010101", "-10011", "-1111", "-320", "-221", "-151", "-125", "-104", "-85", "-78",
"-71", "-67", "-61", "-5a", "-55", "-50", "-4d", "-49", "-45", "-41", "-3j", "-3g", "-3d", "-3a", "-37",
"-34", "-31", "-2r", "-2p", "-2n", "-2l", "-2j", "-2h", "-2f", "-2d"}}},
{INT8_MIN, {{"skip0", "skip1", "-10000000", "-11202", "-2000", "-1003", "-332", "-242", "-200", "-152", "-128",
"-107", "-a8", "-9b", "-92", "-88", "-80", "-79", "-72", "-6e", "-68", "-62", "-5i", "-5d", "-58",
"-53", "-4o", "-4k", "-4g", "-4c", "-48", "-44", "-40", "-3t", "-3q", "-3n", "-3k"}}},
{-1591, {{"skip0", "skip1", "-11000110111", "-2011221", "-120313", "-22331", "-11211", "-4432", "-3067", "-2157",
"-1591", "-1217", "-b07", "-955", "-819", "-711", "-637", "-58a", "-4g7", "-47e", "-3jb", "-3cg",
"-367", "-304", "-2i7", "-2dg", "-295", "-24p", "-20n", "-1pp", "-1n1", "-1ka", "-1hn", "-1f7", "-1cr",
"-1ag", "-187"}}},
{INT16_MIN, {{"skip0", "skip1", "-1000000000000000", "-1122221122", "-20000000", "-2022033", "-411412", "-164351",
"-100000", "-48848", "-32768", "-2268a", "-16b68", "-11bb8", "-bd28", "-9a98", "-8000", "-6b69",
"-5b28", "-4eec", "-41i8", "-3b68", "-31fa", "-2flg", "-28l8", "-22ai", "-1mc8", "-1hph", "-1dm8",
"-19rr", "-16c8", "-1331", "-1000", "-u2w", "-sbq", "-qq8", "-pa8"}}},
{-66748412,
{{"skip0", "skip1", "-11111110100111111111111100", "-11122121011121102", "-3332213333330", "-114041422122",
"-10342352232", "-1440231533", "-376477774", "-148534542", "-66748412", "-34750085", "-1a42b678",
"-10aa0803", "-8c1731a", "-5cd7492", "-3fa7ffc", "-2d03163", "-1h5f3b2", "-17i39c6", "-10h3b0c", "-g749jh",
"-ckkdkg", "-a8c0ak", "-894afk", "-6klmbc", "-5g1i6g", "-4hg4gb", "-3ogi7o", "-37anqb", "-2mc4r2",
"-2a8h7i", "-1vkvvs", "-1n9ca5", "-1fw8sk", "-19gshh", "-13qnek"}}},
{INT32_MIN, {{"skip0", "skip1", "-10000000000000000000000000000000", "-12112122212110202102", "-2000000000000000",
"-13344223434043", "-553032005532", "-104134211162", "-20000000000", "-5478773672", "-2147483648",
"-a02220282", "-4bb2308a8", "-282ba4aab", "-1652ca932", "-c87e66b8", "-80000000", "-53g7f549",
"-3928g3h2", "-27c57h33", "-1db1f928", "-140h2d92", "-ikf5bf2", "-ebelf96", "-b5gge58", "-8jmdnkn",
"-6oj8ioo", "-5ehnckb", "-4clm98g", "-3hk7988", "-2sb6cs8", "-2d09uc2", "-2000000", "-1lsqtl2",
"-1d8xqrq", "-15v22un", "-zik0zk"}}},
{-297139747082649553LL,
{{"skip0", "skip1", "-10000011111101001110000011010010001100000101011111111010001",
"-1222110012002112101210012211022102101", "-100133221300122101200223333101", "-4443033200104011124241203",
"-21313431255203203120401", "-350320603201030412545", "-20375160322140537721", "-1873162471705738371",
"-297139747082649553", "-65150976074a24025", "-173522497b5373101", "-5a60a99bc3b71654", "-1ca51a06cc38ba25",
"-a2a25babe62241d", "-41fa7069182bfd1", "-1d00134fba1769g", "-e4f799fc5f7e81", "-714ebbh8388188",
"-3cahb17836b3hd", "-1j8659jf5hbg3j", "-112bbb2jege5c5", "-dcjfmk2kjb4cc", "-836bm4klbgl61",
"-4ofia1416ee73", "-32ommgjef1l2h", "-1qc52eal5m8ba", "-17n53r05a4r15", "-oa88m2qiqjik", "-gn67qoat5r8d",
"-blgd6n5s90al", "-87t70q8o5fuh", "-5t09hwaqu9qg", "-47vssihaoa4x", "-32p24fbjye7x", "-299r8zck3841"}}},
{stress_chunks_negative,
{{"skip0", "skip1", "-111110011101001100010010000100010000111010101111001010000110001",
"-2012222010200021010000112111002001111200", "-13303221202100202013111321100301",
"-1101001100304341000003214241", "-1522150121302454031001413", "-22054250360123016161454",
"-763514220420725712061", "-65863607100474061450", "-9000876000000054321", "-1689813530958833498",
"-408258185a67069269", "-106b01597a47ba2948", "-41c02922bc776d49b", "-1584cd10979dc84b6",
"-7ce9890887579431", "-327cf6cbc67023c3", "-1604b5f6a0de8129", "-b50d3ef02f124a4", "-59h9bfif0006fg1",
"-2g5d8ekh05d2dfi", "-19i418c38g1chfj", "-hjgf7d0k0gla9a", "-a6b21ncehfa3f9", "-61060fnl003bml",
"-3g88bakondgf8l", "-25q3i730ed21di", "-1al84glo518iip", "-pcli8ig7pjhbo", "-gs31q8id2jnkl",
"-bd7kaglgdrbgk", "-7pqc9123lf51h", "-5d2sd1r5ms7su", "-3q833s8kdrun3", "-2n7vmqigfueqb",
"-1wdu892toj0a9"}}},
{INT64_MIN, {{"skip0", "skip1", "-1000000000000000000000000000000000000000000000000000000000000000",
"-2021110011022210012102010021220101220222", "-20000000000000000000000000000000",
"-1104332401304422434310311213", "-1540241003031030222122212", "-22341010611245052052301",
"-1000000000000000000000", "-67404283172107811828", "-9223372036854775808", "-1728002635214590698",
"-41a792678515120368", "-10b269549075433c38", "-4340724c6c71dc7a8", "-160e2ad3246366808",
"-8000000000000000", "-33d3d8307b214009", "-16agh595df825fa8", "-ba643dci0ffeehi",
"-5cbfjia3fh26ja8", "-2heiciiie82dh98", "-1adaibb21dckfa8", "-i6k448cf4192c3", "-acd772jnc9l0l8",
"-64ie1focnn5g78", "-3igoecjbmca688", "-27c48l5b37oaoq", "-1bk39f3ah3dmq8", "-q1se8f0m04isc",
"-hajppbc1fc208", "-bm03i95hia438", "-8000000000000", "-5hg4ck9jd4u38", "-3tdtk1v8j6tpq",
"-2pijmikexrxp8", "-1y2p0ij32e8e8"}}},
};
template <typename T>
void test_integer_to_chars() {
for (int base = 2; base <= 36; ++base) {
test_integer_to_chars(static_cast<T>(0), base, "0");
test_integer_to_chars(static_cast<T>(1), base, "1");
// tests [3, 71]
test_integer_to_chars(static_cast<T>(base * 2 - 1), base, output_max_digit[base]);
for (const auto& p : output_positive) {
if (p.first <= static_cast<uint64_t>(numeric_limits<T>::max())) {
test_integer_to_chars(static_cast<T>(p.first), base, p.second[static_cast<size_t>(base)]);
}
}
if constexpr (is_signed_v<T>) {
test_integer_to_chars(static_cast<T>(-1), base, "-1");
for (const auto& p : output_negative) {
if (p.first >= static_cast<int64_t>(numeric_limits<T>::min())) {
test_integer_to_chars(static_cast<T>(p.first), base, p.second[static_cast<size_t>(base)]);
}
}
}
}
test_integer_to_chars(static_cast<T>(42), nullopt, "42");
}
enum class TestFromCharsMode { Normal, SignalingNaN };
template <typename T, typename BaseOrFmt>
void test_from_chars(const string_view input, const BaseOrFmt base_or_fmt, const size_t correct_idx,
const errc correct_ec, const optional<T> opt_correct = nullopt,
const TestFromCharsMode mode = TestFromCharsMode::Normal) {
if constexpr (is_integral_v<T>) {
assert(mode == TestFromCharsMode::Normal);
}
constexpr T unmodified = 111;
T dest = unmodified;
const from_chars_result result = from_chars(input.data(), input.data() + input.size(), dest, base_or_fmt);
assert(result.ptr == input.data() + correct_idx);
assert(result.ec == correct_ec);
if (correct_ec == errc{} || (is_floating_point_v<T> && correct_ec == errc::result_out_of_range)) {
if constexpr (is_floating_point_v<T>) {
if (mode == TestFromCharsMode::Normal) {
using Uint = conditional_t<is_same_v<T, float>, uint32_t, uint64_t>;
assert(opt_correct.has_value());
assert(_Bit_cast<Uint>(dest) == _Bit_cast<Uint>(opt_correct.value()));
} else {
assert(mode == TestFromCharsMode::SignalingNaN);
assert(!opt_correct.has_value());
assert(isnan(dest));
}
} else {
assert(opt_correct.has_value());
assert(dest == opt_correct.value());
}
} else {
assert(!opt_correct.has_value());
assert(dest == unmodified);
}
}
constexpr errc inv_arg = errc::invalid_argument;
constexpr errc out_ran = errc::result_out_of_range;
template <typename T>
void test_integer_from_chars() {
for (int base = 2; base <= 36; ++base) {
test_from_chars<T>("", base, 0, inv_arg); // no characters
test_from_chars<T>("@1", base, 0, inv_arg); // '@' is bogus
test_from_chars<T>(".1", base, 0, inv_arg); // '.' is bogus, for integers
test_from_chars<T>("+1", base, 0, inv_arg); // '+' is bogus, N4713 23.20.3 [charconv.from.chars]/3
// "a minus sign is the only sign that may appear"
test_from_chars<T>(" 1", base, 0, inv_arg); // ' ' is bogus, no whitespace in subject sequence
if constexpr (is_unsigned_v<T>) { // N4713 23.20.3 [charconv.from.chars]/3
test_from_chars<T>("-1", base, 0, inv_arg); // "and only if value has a signed type"
}
// N4713 23.20.3 [charconv.from.chars]/1 "[ Note: If the pattern allows for an optional sign,
// but the string has no digit characters following the sign, no characters match the pattern. -end note ]"
test_from_chars<T>("-", base, 0, inv_arg); // '-' followed by no characters
test_from_chars<T>("-@1", base, 0, inv_arg); // '-' followed by bogus '@'
test_from_chars<T>("-.1", base, 0, inv_arg); // '-' followed by bogus '.'
test_from_chars<T>("-+1", base, 0, inv_arg); // '-' followed by bogus '+'
test_from_chars<T>("- 1", base, 0, inv_arg); // '-' followed by bogus ' '
test_from_chars<T>("--1", base, 0, inv_arg); // '-' can't be repeated
vector<char> bogus_digits;
if (base < 10) {
bogus_digits = {static_cast<char>('0' + base), 'A', 'a'};
} else {
// '[' and '{' are bogus for base 36
bogus_digits = {static_cast<char>('A' + (base - 10)), static_cast<char>('a' + (base - 10))};
}
for (const auto& bogus : bogus_digits) {
test_from_chars<T>(bogus + "1"s, base, 0, inv_arg); // bogus digit (for this base)
test_from_chars<T>("-"s + bogus + "1"s, base, 0, inv_arg); // '-' followed by bogus digit
}
// Test leading zeroes.
test_from_chars<T>(string(100, '0'), base, 100, errc{}, static_cast<T>(0));
test_from_chars<T>(string(100, '0') + "11"s, base, 102, errc{}, static_cast<T>(base + 1));
// Test negative zero and negative leading zeroes.
if constexpr (is_signed_v<T>) {
test_from_chars<T>("-0", base, 2, errc{}, static_cast<T>(0));
test_from_chars<T>("-"s + string(100, '0'), base, 101, errc{}, static_cast<T>(0));
test_from_chars<T>("-"s + string(100, '0') + "11"s, base, 103, errc{}, static_cast<T>(-base - 1));
}
// N4713 23.20.3 [charconv.from.chars]/1 "The member ptr of the return value points to the
// first character not matching the pattern, or has the value last if all characters match."
test_from_chars<T>("11", base, 2, errc{}, static_cast<T>(base + 1));
test_from_chars<T>("11@@@", base, 2, errc{}, static_cast<T>(base + 1));
// When overflowing, we need to keep consuming valid digits, in order to return ptr correctly.
test_from_chars<T>(string(100, '1'), base, 100, out_ran);
test_from_chars<T>(string(100, '1') + "@@@"s, base, 100, out_ran);
if constexpr (is_signed_v<T>) {
test_from_chars<T>("-"s + string(100, '1'), base, 101, out_ran);
test_from_chars<T>("-"s + string(100, '1') + "@@@"s, base, 101, out_ran);
}
}
// N4713 23.20.3 [charconv.from.chars]/3 "The pattern is the expected form of the subject sequence
// in the "C" locale for the given nonzero base, as described for strtol"
// C11 7.22.1.4/3 "The letters from a (or A) through z (or Z) are ascribed the values 10 through 35"
for (int i = 0; i < 26; ++i) {
test_from_chars<T>(string(1, static_cast<char>('A' + i)), 36, 1, errc{}, static_cast<T>(10 + i));
test_from_chars<T>(string(1, static_cast<char>('a' + i)), 36, 1, errc{}, static_cast<T>(10 + i));
}
// N4713 23.20.3 [charconv.from.chars]/3 "no "0x" or "0X" prefix shall appear if the value of base is 16"
test_from_chars<T>("0x1729", 16, 1, errc{}, static_cast<T>(0)); // reads '0', stops at 'x'
test_from_chars<T>("0X1729", 16, 1, errc{}, static_cast<T>(0)); // reads '0', stops at 'X'
if constexpr (is_signed_v<T>) {
test_from_chars<T>("-0x1729", 16, 2, errc{}, static_cast<T>(0)); // reads "-0", stops at 'x'
test_from_chars<T>("-0X1729", 16, 2, errc{}, static_cast<T>(0)); // reads "-0", stops at 'X'
}
}
template <typename T>
void test_integer() {
test_integer_to_chars<T>();
test_integer_from_chars<T>();
}
void all_integer_tests() {
test_integer<char>();
test_integer<signed char>();
test_integer<unsigned char>();
test_integer<short>();
test_integer<unsigned short>();
test_integer<int>();
test_integer<unsigned int>();
test_integer<long>();
test_integer<unsigned long>();
test_integer<long long>();
test_integer<unsigned long long>();
// Test overflow scenarios.
test_from_chars<unsigned int>("4294967289", 10, 10, errc{}, 4294967289U); // not risky
test_from_chars<unsigned int>("4294967294", 10, 10, errc{}, 4294967294U); // risky with good digit
test_from_chars<unsigned int>("4294967295", 10, 10, errc{}, 4294967295U); // risky with max digit
test_from_chars<unsigned int>("4294967296", 10, 10, out_ran); // risky with bad digit
test_from_chars<unsigned int>("4294967300", 10, 10, out_ran); // beyond risky
test_from_chars<int>("2147483639", 10, 10, errc{}, 2147483639); // not risky
test_from_chars<int>("2147483646", 10, 10, errc{}, 2147483646); // risky with good digit
test_from_chars<int>("2147483647", 10, 10, errc{}, 2147483647); // risky with max digit
test_from_chars<int>("2147483648", 10, 10, out_ran); // risky with bad digit
test_from_chars<int>("2147483650", 10, 10, out_ran); // beyond risky
test_from_chars<int>("-2147483639", 10, 11, errc{}, -2147483639); // not risky
test_from_chars<int>("-2147483647", 10, 11, errc{}, -2147483647); // risky with good digit
test_from_chars<int>("-2147483648", 10, 11, errc{}, -2147483647 - 1); // risky with max digit
test_from_chars<int>("-2147483649", 10, 11, out_ran); // risky with bad digit
test_from_chars<int>("-2147483650", 10, 11, out_ran); // beyond risky
}
void assert_message_bits(const bool b, const char* const msg, const uint32_t bits) {
if (!b) {
fprintf(stderr, "%s failed for 0x%08X\n", msg, bits);
fprintf(stderr, "This is a randomized test.\n");
fprintf(stderr, "DO NOT IGNORE/RERUN THIS FAILURE.\n");
fprintf(stderr, "You must report it to the STL maintainers.\n");
abort();
}
}
void assert_message_bits(const bool b, const char* const msg, const uint64_t bits) {
if (!b) {
// libc++ uses long for 64-bit values.
fprintf(stderr, "%s failed for 0x%016llX\n", msg, static_cast<unsigned long long>(bits));
fprintf(stderr, "This is a randomized test.\n");
fprintf(stderr, "DO NOT IGNORE/RERUN THIS FAILURE.\n");
fprintf(stderr, "You must report it to the STL maintainers.\n");
abort();
}
}
constexpr uint32_t FractionBits = 10; // Tunable for test coverage vs. performance.
static_assert(FractionBits >= 1, "Must test at least 1 fraction bit.");
static_assert(FractionBits <= 23, "There are only 23 fraction bits in a float.");
constexpr uint32_t Fractions = 1U << FractionBits;
constexpr uint32_t Mask32 = ~((1U << FractionBits) - 1U);
constexpr uint64_t Mask64 = ~((1ULL << FractionBits) - 1ULL);
constexpr uint32_t PrefixesToTest = 100; // Tunable for test coverage vs. performance.
static_assert(PrefixesToTest >= 1, "Must test at least 1 prefix.");
constexpr uint32_t PrefixLimit = 2 // sign bit
* 255 // non-INF/NAN exponents for float
* (1U << (23 - FractionBits)); // fraction bits in prefix
static_assert(PrefixesToTest <= PrefixLimit, "Too many prefixes.");
template <bool IsDouble>
void test_floating_prefix(const conditional_t<IsDouble, uint64_t, uint32_t> prefix) {
using UIntType = conditional_t<IsDouble, uint64_t, uint32_t>;
using FloatingType = conditional_t<IsDouble, double, float>;
// "-1.2345678901234567e-100" or "-1.23456789e-10"
constexpr size_t buffer_size = IsDouble ? 24 : 15;
char buffer[buffer_size];
// TODO Enable once std::from_chars has floating point support.
#if 0
FloatingType val;
#endif
// Exact sizes are difficult to prove for fixed notation.
// This must be at least (IsDouble ? 327 : 48), and I suspect that that's exact.
// Here's a loose upper bound:
// 1 character for a negative sign
// + 325 (for double; 46 for float) characters in the "0.000~~~000" prefix of the min subnormal
// + 17 (for double; 9 for float) characters for round-trip digits
constexpr size_t fixed_buffer_size = IsDouble ? 1 + 325 + 17 : 1 + 46 + 9;
char fixed_buffer[fixed_buffer_size];
// worst case: negative sign + max normal + null terminator
constexpr size_t stdio_buffer_size = 1 + (IsDouble ? 309 : 39) + 1;
char stdio_buffer[stdio_buffer_size];
for (uint32_t frac = 0; frac < Fractions; ++frac) {
const UIntType bits = prefix + frac;
const FloatingType input = _Bit_cast<FloatingType>(bits);
{
const auto to_result = to_chars(buffer, end(buffer), input, chars_format::scientific);
assert_message_bits(to_result.ec == errc{}, "to_result.ec", bits);
// TODO Enable once std::from_chars has floating point support.
#if 0
const char* const last = to_result.ptr;
const auto from_result = from_chars(buffer, last, val);
assert_message_bits(from_result.ptr == last, "from_result.ptr", bits);
assert_message_bits(from_result.ec == errc{}, "from_result.ec", bits);
assert_message_bits(_Bit_cast<UIntType>(val) == bits, "round-trip", bits);
#endif
}
{
// Also verify that to_chars() and sprintf_s() emit the same output for integers in fixed notation.
const auto fixed_result = to_chars(fixed_buffer, end(fixed_buffer), input, chars_format::fixed);
assert_message_bits(fixed_result.ec == errc{}, "fixed_result.ec", bits);
const string_view fixed_sv(fixed_buffer, static_cast<size_t>(fixed_result.ptr - fixed_buffer));
if (find(fixed_sv.begin(), fixed_sv.end(), '.') == fixed_sv.end()) {
const int stdio_ret = sprintf_s(stdio_buffer, size(stdio_buffer), "%.0f", input);
assert_message_bits(stdio_ret != -1, "stdio_ret", bits);
const string_view stdio_sv(stdio_buffer);
assert_message_bits(fixed_sv == stdio_sv, "fixed_sv", bits);
}
}
}
}
template <bool IsDouble>
void test_floating_hex_prefix(const conditional_t<IsDouble, uint64_t, uint32_t> prefix) {
using UIntType = conditional_t<IsDouble, uint64_t, uint32_t>;
using FloatingType = conditional_t<IsDouble, double, float>;
// The precision is the number of hexits after the decimal point.
// These hexits correspond to the explicitly stored fraction bits.
// double explicitly stores 52 fraction bits. 52 / 4 == 13, so we need 13 hexits.
// float explicitly stores 23 fraction bits. 23 / 4 == 5.75, so we need 6 hexits.
// "-1.fffffffffffffp+1023" or "-1.fffffep+127"
constexpr size_t buffer_size = IsDouble ? 22 : 14;
char buffer[buffer_size];
// TODO Enable once std::from_chars has floating point support.
#if 0
FloatingType val;
#endif
for (uint32_t frac = 0; frac < Fractions; ++frac) {
const UIntType bits = prefix + frac;
const FloatingType input = _Bit_cast<FloatingType>(bits);
const auto to_result = to_chars(buffer, end(buffer), input, chars_format::hex);
assert_message_bits(to_result.ec == errc{}, "(hex) to_result.ec", bits);
// TODO Enable once std::from_chars has floating point support.
#if 0
const char* const last = to_result.ptr;
const auto from_result = from_chars(buffer, last, val, chars_format::hex);
assert_message_bits(from_result.ptr == last, "(hex) from_result.ptr", bits);
assert_message_bits(from_result.ec == errc{}, "(hex) from_result.ec", bits);
assert_message_bits(_Bit_cast<UIntType>(val) == bits, "(hex) round-trip", bits);
#endif
}
}
template <bool IsDouble>
void test_floating_precision_prefix(const conditional_t<IsDouble, uint64_t, uint32_t> prefix) {
using UIntType = conditional_t<IsDouble, uint64_t, uint32_t>;
using FloatingType = conditional_t<IsDouble, double, float>;
// Precision for min subnormal in fixed notation. (More than enough for scientific notation.)
constexpr int precision = IsDouble ? 1074 : 149;
// Number of digits for max normal in fixed notation.
constexpr int max_integer_length = IsDouble ? 309 : 39;
// Size for fixed notation. (More than enough for scientific notation.)
constexpr size_t charconv_buffer_size = 1 // negative sign
+ max_integer_length // integer digits
+ 1 // decimal point
+ precision; // fractional digits
char charconv_buffer[charconv_buffer_size];
constexpr size_t stdio_buffer_size = charconv_buffer_size + 1; // null terminator
char stdio_buffer[stdio_buffer_size];
// 1 character for a negative sign
// + worst cases: 0x1.fffffffffffffp-1022 and 0x1.fffffep-126f
constexpr size_t general_buffer_size = 1 + (IsDouble ? 773 : 117);
char general_buffer[general_buffer_size];
char general_stdio_buffer[general_buffer_size + 1]; // + null terminator
for (uint32_t frac = 0; frac < Fractions; ++frac) {
const UIntType bits = prefix + frac;
const FloatingType input = _Bit_cast<FloatingType>(bits);
auto result = to_chars(charconv_buffer, end(charconv_buffer), input, chars_format::fixed, precision);
assert_message_bits(result.ec == errc{}, "to_chars fixed precision", bits);
string_view charconv_sv(charconv_buffer, static_cast<size_t>(result.ptr - charconv_buffer));
int stdio_ret = sprintf_s(stdio_buffer, size(stdio_buffer), "%.*f", precision, input);
assert_message_bits(stdio_ret != -1, "sprintf_s fixed precision", bits);
string_view stdio_sv(stdio_buffer);
assert_message_bits(charconv_sv == stdio_sv, "fixed precision output", bits);
result = to_chars(charconv_buffer, end(charconv_buffer), input, chars_format::scientific, precision);
assert_message_bits(result.ec == errc{}, "to_chars scientific precision", bits);
charconv_sv = string_view(charconv_buffer, static_cast<size_t>(result.ptr - charconv_buffer));
stdio_ret = sprintf_s(stdio_buffer, size(stdio_buffer), "%.*e", precision, input);
assert_message_bits(stdio_ret != -1, "sprintf_s scientific precision", bits);
stdio_sv = stdio_buffer;
assert_message_bits(charconv_sv == stdio_sv, "scientific precision output", bits);
result = to_chars(general_buffer, end(general_buffer), input, chars_format::general, 5000);
assert_message_bits(result.ec == errc{}, "to_chars general precision", bits);
charconv_sv = string_view(general_buffer, static_cast<size_t>(result.ptr - general_buffer));
stdio_ret = sprintf_s(general_stdio_buffer, size(general_stdio_buffer), "%.5000g", input);
assert_message_bits(stdio_ret != -1, "sprintf_s general precision", bits);
stdio_sv = general_stdio_buffer;
assert_message_bits(charconv_sv == stdio_sv, "general precision output", bits);
}
}
void test_floating_prefixes(mt19937_64& mt64) {
{
set<uint64_t> prefixes64;
while (prefixes64.size() < PrefixesToTest) {
const uint64_t val = mt64();
if ((val & 0x7FF0000000000000ULL) != 0x7FF0000000000000ULL) { // skip INF/NAN
prefixes64.insert(val & Mask64);
}
}
for (const auto& prefix : prefixes64) {
test_floating_prefix<true>(prefix);
test_floating_precision_prefix<true>(prefix);
}
test_floating_hex_prefix<true>(*prefixes64.begin()); // save time by testing fewer hexfloats
}
{
set<uint32_t> prefixes32;
while (prefixes32.size() < PrefixesToTest) {
const uint32_t val = static_cast<uint32_t>(mt64());
if ((val & 0x7F800000U) != 0x7F800000U) { // skip INF/NAN
prefixes32.insert(val & Mask32);
}
}
for (const auto& prefix : prefixes32) {
test_floating_prefix<false>(prefix);
test_floating_precision_prefix<false>(prefix);
}
test_floating_hex_prefix<false>(*prefixes32.begin()); // save time by testing fewer hexfloats
}
}
// TODO Enable once std::from_chars has floating point support.
#if 0
template <typename T>
void test_floating_from_chars(const chars_format fmt) {
test_from_chars<T>("", fmt, 0, inv_arg); // no characters
test_from_chars<T>("@1", fmt, 0, inv_arg); // '@' is bogus
test_from_chars<T>("z1", fmt, 0, inv_arg); // 'z' is bogus
test_from_chars<T>(".", fmt, 0, inv_arg); // '.' without digits is bogus
test_from_chars<T>("+1", fmt, 0, inv_arg); // '+' is bogus
test_from_chars<T>(" 1", fmt, 0, inv_arg); // ' ' is bogus
test_from_chars<T>("p5", fmt, 0, inv_arg); // binary-exponent-part without digits is bogus
test_from_chars<T>("in", fmt, 0, inv_arg); // incomplete inf is bogus
test_from_chars<T>("na", fmt, 0, inv_arg); // incomplete nan is bogus
test_from_chars<T>("-", fmt, 0, inv_arg); // '-' followed by no characters
test_from_chars<T>("-@1", fmt, 0, inv_arg); // '-' followed by bogus '@'
test_from_chars<T>("-z1", fmt, 0, inv_arg); // '-' followed by bogus 'z'
test_from_chars<T>("-.", fmt, 0, inv_arg); // '-' followed by bogus '.'
test_from_chars<T>("-+1", fmt, 0, inv_arg); // '-' followed by bogus '+'
test_from_chars<T>("- 1", fmt, 0, inv_arg); // '-' followed by bogus ' '
test_from_chars<T>("-p5", fmt, 0, inv_arg); // '-' followed by bogus binary-exponent-part
test_from_chars<T>("-in", fmt, 0, inv_arg); // '-' followed by bogus incomplete inf
test_from_chars<T>("-na", fmt, 0, inv_arg); // '-' followed by bogus incomplete nan
test_from_chars<T>("--1", fmt, 0, inv_arg); // '-' can't be repeated
if (fmt != chars_format::hex) { // "e5" are valid hexits
test_from_chars<T>("e5", fmt, 0, inv_arg); // exponent-part without digits is bogus
test_from_chars<T>("-e5", fmt, 0, inv_arg); // '-' followed by bogus exponent-part
}
constexpr T inf = numeric_limits<T>::infinity();
constexpr T qnan = numeric_limits<T>::quiet_NaN();
test_from_chars<T>("InF", fmt, 3, errc{}, inf);
test_from_chars<T>("infinite", fmt, 3, errc{}, inf);
test_from_chars<T>("iNfInItY", fmt, 8, errc{}, inf);
test_from_chars<T>("InfinityMeow", fmt, 8, errc{}, inf);
test_from_chars<T>("-InF", fmt, 4, errc{}, -inf);
test_from_chars<T>("-infinite", fmt, 4, errc{}, -inf);
test_from_chars<T>("-iNfInItY", fmt, 9, errc{}, -inf);
test_from_chars<T>("-InfinityMeow", fmt, 9, errc{}, -inf);
test_from_chars<T>("NaN", fmt, 3, errc{}, qnan);
test_from_chars<T>("nanotech", fmt, 3, errc{}, qnan);
test_from_chars<T>("nan(", fmt, 3, errc{}, qnan);
test_from_chars<T>("nan(@)", fmt, 3, errc{}, qnan);
test_from_chars<T>("nan(()", fmt, 3, errc{}, qnan);
test_from_chars<T>("nan(abc", fmt, 3, errc{}, qnan);
test_from_chars<T>("nan()", fmt, 5, errc{}, qnan);
test_from_chars<T>("nan(abc)def", fmt, 8, errc{}, qnan);
test_from_chars<T>("nan(_09AZaz)", fmt, 12, errc{}, qnan);
test_from_chars<T>("nan(int)", fmt, 8, errc{}, qnan);
test_from_chars<T>("nan(snap)", fmt, 9, errc{}, qnan);
test_from_chars<T>("-NaN", fmt, 4, errc{}, -qnan);
test_from_chars<T>("-nanotech", fmt, 4, errc{}, -qnan);
test_from_chars<T>("-nan(", fmt, 4, errc{}, -qnan);
test_from_chars<T>("-nan(@)", fmt, 4, errc{}, -qnan);
test_from_chars<T>("-nan(()", fmt, 4, errc{}, -qnan);
test_from_chars<T>("-nan(abc", fmt, 4, errc{}, -qnan);
test_from_chars<T>("-nan()", fmt, 6, errc{}, -qnan);
test_from_chars<T>("-nan(abc)def", fmt, 9, errc{}, -qnan);
test_from_chars<T>("-nan(_09AZaz)", fmt, 13, errc{}, -qnan);
test_from_chars<T>("-nan(int)", fmt, 9, errc{}, -qnan);
test_from_chars<T>("-nan(snap)", fmt, 10, errc{}, -qnan);
// The UCRT considers indeterminate NaN to be negative quiet NaN with no payload bits set.
// It parses "nan(ind)" and "-nan(ind)" identically.
test_from_chars<T>("nan(InD)", fmt, 8, errc{}, -qnan);
test_from_chars<T>("-nan(InD)", fmt, 9, errc{}, -qnan);
test_from_chars<T>("nan(SnAn)", fmt, 9, errc{}, nullopt, TestFromCharsMode::SignalingNaN);
test_from_chars<T>("-nan(SnAn)", fmt, 10, errc{}, nullopt, TestFromCharsMode::SignalingNaN);
switch (fmt) {
case chars_format::general:
test_from_chars<T>("1729", fmt, 4, errc{}, T{1729});
test_from_chars<T>("1729e3", fmt, 6, errc{}, T{1729000});
test_from_chars<T>("10", fmt, 2, errc{}, T{10});
test_from_chars<T>("11.", fmt, 3, errc{}, T{11});
test_from_chars<T>("12.13", fmt, 5, errc{}, static_cast<T>(12.13)); // avoid truncation warning
test_from_chars<T>(".14", fmt, 3, errc{}, static_cast<T>(.14)); // avoid truncation warning
test_from_chars<T>("20e5", fmt, 4, errc{}, T{2000000});
test_from_chars<T>("21.e5", fmt, 5, errc{}, T{2100000});
test_from_chars<T>("22.23e5", fmt, 7, errc{}, T{2223000});
test_from_chars<T>(".24e5", fmt, 5, errc{}, T{24000});
test_from_chars<T>("33e+5", fmt, 5, errc{}, T{3300000});
test_from_chars<T>("33e-5", fmt, 5, errc{}, static_cast<T>(.00033)); // avoid truncation warning
test_from_chars<T>("4E7", fmt, 3, errc{}, T{40000000});
test_from_chars<T>("-00123abc", fmt, 6, errc{}, T{-123});
test_from_chars<T>(".0045000", fmt, 8, errc{}, static_cast<T>(.0045)); // avoid truncation warning
test_from_chars<T>("000", fmt, 3, errc{}, T{0});
test_from_chars<T>("0e9999", fmt, 6, errc{}, T{0});
test_from_chars<T>("0e-9999", fmt, 7, errc{}, T{0});
test_from_chars<T>("-000", fmt, 4, errc{}, T{-0.0});
test_from_chars<T>("-0e9999", fmt, 7, errc{}, T{-0.0});
test_from_chars<T>("-0e-9999", fmt, 8, errc{}, T{-0.0});
test_from_chars<T>("1e9999", fmt, 6, errc::result_out_of_range, inf);
test_from_chars<T>("-1e9999", fmt, 7, errc::result_out_of_range, -inf);
test_from_chars<T>("1e-9999", fmt, 7, errc::result_out_of_range, T{0});
test_from_chars<T>("-1e-9999", fmt, 8, errc::result_out_of_range, T{-0.0});
test_from_chars<T>("1" + string(6000, '0'), fmt, 6001, errc::result_out_of_range, inf);
test_from_chars<T>("-1" + string(6000, '0'), fmt, 6002, errc::result_out_of_range, -inf);
test_from_chars<T>("." + string(6000, '0') + "1", fmt, 6002, errc::result_out_of_range, T{0});
test_from_chars<T>("-." + string(6000, '0') + "1", fmt, 6003, errc::result_out_of_range, T{-0.0});
test_from_chars<T>("1" + string(500, '0'), fmt, 501, errc::result_out_of_range, inf);
test_from_chars<T>("-1" + string(500, '0'), fmt, 502, errc::result_out_of_range, -inf);
test_from_chars<T>("." + string(500, '0') + "1", fmt, 502, errc::result_out_of_range, T{0});
test_from_chars<T>("-." + string(500, '0') + "1", fmt, 503, errc::result_out_of_range, T{-0.0});
break;
case chars_format::scientific:
test_from_chars<T>("1729", fmt, 0, inv_arg);
test_from_chars<T>("1729e3", fmt, 6, errc{}, T{1729000});
break;
case chars_format::fixed:
test_from_chars<T>("1729", fmt, 4, errc{}, T{1729});
test_from_chars<T>("1729e3", fmt, 4, errc{}, T{1729});
break;
case chars_format::hex:
test_from_chars<T>("0x123", fmt, 1, errc{}, T{0});
test_from_chars<T>("a0", fmt, 2, errc{}, T{160});
test_from_chars<T>("a1.", fmt, 3, errc{}, T{161});
test_from_chars<T>("a2.a3", fmt, 5, errc{}, T{162.63671875});
test_from_chars<T>(".a4", fmt, 3, errc{}, T{0.640625});
test_from_chars<T>("a0p5", fmt, 4, errc{}, T{5120});
test_from_chars<T>("a1.p5", fmt, 5, errc{}, T{5152});
test_from_chars<T>("a2.a3p5", fmt, 7, errc{}, T{5204.375});
test_from_chars<T>(".a4p5", fmt, 5, errc{}, T{20.5});
test_from_chars<T>("a0p+5", fmt, 5, errc{}, T{5120});
test_from_chars<T>("a0p-5", fmt, 5, errc{}, T{5});
test_from_chars<T>("ABCDEFP3", fmt, 8, errc{}, T{90075000});
test_from_chars<T>("-00cdrom", fmt, 5, errc{}, T{-205});
test_from_chars<T>(".00ef000", fmt, 8, errc{}, T{0.0036468505859375});
test_from_chars<T>("000", fmt, 3, errc{}, T{0});
test_from_chars<T>("0p9999", fmt, 6, errc{}, T{0});
test_from_chars<T>("0p-9999", fmt, 7, errc{}, T{0});
test_from_chars<T>("-000", fmt, 4, errc{}, T{-0.0});
test_from_chars<T>("-0p9999", fmt, 7, errc{}, T{-0.0});
test_from_chars<T>("-0p-9999", fmt, 8, errc{}, T{-0.0});
test_from_chars<T>("1p9999", fmt, 6, errc::result_out_of_range, inf);
test_from_chars<T>("-1p9999", fmt, 7, errc::result_out_of_range, -inf);
test_from_chars<T>("1p-9999", fmt, 7, errc::result_out_of_range, T{0});
test_from_chars<T>("-1p-9999", fmt, 8, errc::result_out_of_range, T{-0.0});
test_from_chars<T>("1" + string(2000, '0'), fmt, 2001, errc::result_out_of_range, inf);
test_from_chars<T>("-1" + string(2000, '0'), fmt, 2002, errc::result_out_of_range, -inf);
test_from_chars<T>("." + string(2000, '0') + "1", fmt, 2002, errc::result_out_of_range, T{0});
test_from_chars<T>("-." + string(2000, '0') + "1", fmt, 2003, errc::result_out_of_range, T{-0.0});
test_from_chars<T>("1" + string(300, '0'), fmt, 301, errc::result_out_of_range, inf);
test_from_chars<T>("-1" + string(300, '0'), fmt, 302, errc::result_out_of_range, -inf);
test_from_chars<T>("." + string(300, '0') + "1", fmt, 302, errc::result_out_of_range, T{0});
test_from_chars<T>("-." + string(300, '0') + "1", fmt, 303, errc::result_out_of_range, T{-0.0});
break;
}
}
#endif
template <typename T>
void test_floating_to_chars(
const T value, const optional<chars_format> opt_fmt, const optional<int> opt_precision, const string_view correct) {
test_common_to_chars(value, opt_fmt, opt_precision, correct);
}
void all_floating_tests(mt19937_64& mt64) {
test_floating_prefixes(mt64);
// TODO Enable once std::from_chars has floating point support.
#if 0
for (const auto& fmt : {chars_format::general, chars_format::scientific, chars_format::fixed, chars_format::hex}) {
test_floating_from_chars<float>(fmt);
test_floating_from_chars<double>(fmt);
}
// Test rounding.
// See float_from_chars_test_cases.hpp in this directory.
for (const auto& t : float_from_chars_test_cases) {
test_from_chars<float>(t.input, t.fmt, t.correct_idx, t.correct_ec, t.correct_value);
}
// See double_from_chars_test_cases.hpp in this directory.
for (const auto& t : double_from_chars_test_cases) {
test_from_chars<double>(t.input, t.fmt, t.correct_idx, t.correct_ec, t.correct_value);
}
{
// See LWG-2403. This number (exactly 0x1.fffffd00000004 in infinite precision) behaves differently
// when parsed as double and converted to float, versus being parsed as float directly.
const char* const lwg_2403 = "1.999999821186065729339276231257827021181583404541015625";
constexpr float correct_float = 0x1.fffffep0f;
constexpr double correct_double = 0x1.fffffdp0;
constexpr float twice_rounded_float = 0x1.fffffcp0f;
test_from_chars<float>(lwg_2403, chars_format::general, 56, errc{}, correct_float);
test_from_chars<double>(lwg_2403, chars_format::general, 56, errc{}, correct_double);
static_assert(static_cast<float>(correct_double) == twice_rounded_float);
}
// See floating_point_test_cases.hpp.
for (const auto& p : floating_point_test_cases_float) {
test_from_chars<float>(p.first, chars_format::general, strlen(p.first), errc{}, _Bit_cast<float>(p.second));
}
for (const auto& p : floating_point_test_cases_double) {
test_from_chars<double>(p.first, chars_format::general, strlen(p.first), errc{}, _Bit_cast<double>(p.second));
}
#endif
// See float_to_chars_test_cases.hpp in this directory.
for (const auto& t : float_to_chars_test_cases) {
if (t.fmt == chars_format{}) {
test_floating_to_chars(t.value, nullopt, nullopt, t.correct);
} else {
test_floating_to_chars(t.value, t.fmt, nullopt, t.correct);
}
}
// See double_to_chars_test_cases.hpp in this directory.
for (const auto& t : double_to_chars_test_cases) {
if (t.fmt == chars_format{}) {
test_floating_to_chars(t.value, nullopt, nullopt, t.correct);
} else {
test_floating_to_chars(t.value, t.fmt, nullopt, t.correct);
}
}
// See corresponding headers in this directory.
for (const auto& t : float_hex_precision_to_chars_test_cases) {
test_floating_to_chars(t.value, t.fmt, t.precision, t.correct);
}
for (const auto& t : float_fixed_precision_to_chars_test_cases) {
test_floating_to_chars(t.value, t.fmt, t.precision, t.correct);
}
for (const auto& t : float_scientific_precision_to_chars_test_cases) {
test_floating_to_chars(t.value, t.fmt, t.precision, t.correct);
}
for (const auto& t : float_general_precision_to_chars_test_cases) {
test_floating_to_chars(t.value, t.fmt, t.precision, t.correct);
}
for (const auto& t : double_hex_precision_to_chars_test_cases) {
test_floating_to_chars(t.value, t.fmt, t.precision, t.correct);
}
for (const auto& t : double_fixed_precision_to_chars_test_cases_1) {
test_floating_to_chars(t.value, t.fmt, t.precision, t.correct);
}
for (const auto& t : double_fixed_precision_to_chars_test_cases_2) {
test_floating_to_chars(t.value, t.fmt, t.precision, t.correct);
}
for (const auto& t : double_fixed_precision_to_chars_test_cases_3) {
test_floating_to_chars(t.value, t.fmt, t.precision, t.correct);
}
for (const auto& t : double_fixed_precision_to_chars_test_cases_4) {
test_floating_to_chars(t.value, t.fmt, t.precision, t.correct);
}
for (const auto& t : double_scientific_precision_to_chars_test_cases_1) {
test_floating_to_chars(t.value, t.fmt, t.precision, t.correct);
}
for (const auto& t : double_scientific_precision_to_chars_test_cases_2) {
test_floating_to_chars(t.value, t.fmt, t.precision, t.correct);
}
for (const auto& t : double_scientific_precision_to_chars_test_cases_3) {
test_floating_to_chars(t.value, t.fmt, t.precision, t.correct);
}
for (const auto& t : double_scientific_precision_to_chars_test_cases_4) {
test_floating_to_chars(t.value, t.fmt, t.precision, t.correct);
}
for (const auto& t : double_general_precision_to_chars_test_cases) {
test_floating_to_chars(t.value, t.fmt, t.precision, t.correct);
}
}
int main(int argc, char** argv) {
const auto start = chrono::steady_clock::now();
mt19937_64 mt64;
initialize_randomness(mt64, argc, argv);
all_integer_tests();
all_floating_tests(mt64);
const auto finish = chrono::steady_clock::now();
const long long ms = chrono::duration_cast<chrono::milliseconds>(finish - start).count();
puts("PASS");
printf("Randomized test cases: %u\n", PrefixesToTest * Fractions);
printf("Total time: %lld ms\n", ms);
if (ms < 3'000) {
puts("That was fast. Consider tuning PrefixesToTest and FractionBits to test more cases.");
} else if (ms > 30'000) {
puts("That was slow. Consider tuning PrefixesToTest and FractionBits to test fewer cases.");
}
}
Reference in New Issue View Git Blame Copy Permalink