[libc++][format] Adds integer formatter.

[Mordante]

Summary

Implements the formatter for all fundamental integer types
(except char, wchar_t, and bool).
[format.formatter.spec]/2.3
For each charT, for each cv-unqualified arithmetic type ArithmeticT other
than char, wchar_t, char8_t, char16_t, or char32_t, a specialization

template<> struct formatter<ArithmeticT, charT>;

This removes the stub implemented in D96664.

As an extension it adds partial support for 128-bit integer types.

Implements parts of:

• P0645 Text Formatting
• P1652 Printf corner cases in std::format

Completes:

• LWG-3248 #b, #B, #o, #x, and #X presentation types misformat negative numbers

[Mordante]

Adds a required inline, fixes the build issue.

[Mordante]

Cast to the proper signedness that should fix the AArch and ARM builds.

[Mordante]

• Rebase
• Fix octal buffer size off by 1.
• Add test for maximum integral values.
• Fixes Windows issue
• Polishing

[Mordante]

Rebased and updated:

• Address some changes from previous patches
• Minor cleanups and comment improvements
• Added a module map

[Mordante]

Rebase.

[vitaut]

libcxx/include/__format/formatter_integer.h

112–117

AFAICS to_chars supports 128-bit integers (https://godbolt.org/z/7GTPrcq88) or am I missing something?

libcxx/include/__format/formatter_integral.h

72

Did you mean "combining"?

89

Why not make these protected to keep the public API of formatter specializations clean?

137

Here and elsewhere the default alignment should depend on the presentation type: http://eel.is/c++draft/format#tab:format.align

i.e. format("{:10}", true) is left-aligned while format("{:10d}", true) is right-aligned.

232

#ifdef can be omitted because there is nothing that use std::locale in __determine_grouping. It will just be unused if locales are disabled.

305

Why store __negative instead of passing it as an argument to __format_unsigned_integral?

315

nit: I'd rename it to __format_as_char because __value is not a char but an integer which we convert and format as char.

316–317

Isn't the alignment supposed to be resolved at parse time by calling __handle_char / __handle_bool?

330

_CharT is already signed so make_signed_t doesn't do anything. Did you mean make_unsigned_t?

libcxx/test/std/utilities/format/format.functions/tests.inc

380–381

Per my earlier comment this should indeed be left-aligned.

[vitaut]

libcxx/include/__format/formatter_integral.h

549

In general formatters should only contain information derived from the format string, not from the value. Among other things this would inhibit format string compilation in the future. Please remove.

[Mordante]

Thanks for the review!

libcxx/include/__format/formatter_integer.h

112–117

libc++ accepts 128-bit integers, but silently truncates them to 64-bit. IMO this is a bug and it's on my TODO to fix this.
It's visible in the generated assembly, it contains a`std::1::itoa::__pow10_64` lookup table instead of a std::__1::__itoa::__pow10_128.

libcxx/include/__format/formatter_integral.h

89

Good catch!

137

This function is only called when a char is used for the char representation is selected.
Only the __handle_integer is used slightly different. That was basically done to use different formatting for
format("{:6c}", 42) and format("{:6}", '*'). But I'll adjust __handle_integer.

See my response to https://reviews.llvm.org/D103433/new/#inline-1010793 for more information.

305

Basically to avoid passing an argument to a forwarding function. IMO both are a valid design choice. But as you mention in another comment "Among other things this would inhibit format string compilation in the future. Please remove.". So I'll apply your suggestion.

316–317

Only partly, not for integers formatted as char. But I'll adjust this.

See my response to https://reviews.llvm.org/D103433/new/#inline-1010793 for more information.

330

Should indeed be make_unsigned_t. Odd the compiler doesn't complain about a useless cast.

libcxx/test/std/utilities/format/format.functions/tests.inc

380–381

I assumed that was intended. However my interpretation of the wording requires it to be right aligned:
This is the default for arithmetic types other than charT and bool or when an integer presentation type is specified.
My interpretation is that the supplied type in an aritmetic type, so this rule applies.
Do you agree with my interpretation?

I agree it's silly that the result of format("{:6c}", 42); differs from the result of format("{:6}", '*');. So I'll adjust it.

[Mordante]

Rebase
Addresses review comments
Contrain the formatter to the allowed character types
Use private module headers
Guard unit tests for implementation details

[• Quuxplusone]

libcxx/include/__format/formatter_integer.h

112–117

That is definitely a bug; see https://godbolt.org/z/j9r7YG8aq where there's a discontinuity at base=10 (but I think all the other outputs are correct, probably?). It could be quickly patched like this:

diff --git a/libcxx/include/charconv b/libcxx/include/charconv
index 4a5c65df7c75..74b8993cf392 100644
--- a/libcxx/include/charconv
+++ b/libcxx/include/charconv
@@ -351,6 +351,9 @@ _LIBCPP_AVAILABILITY_TO_CHARS
 inline _LIBCPP_INLINE_VISIBILITY to_chars_result
 __to_chars_itoa(char* __first, char* __last, _Tp __value, false_type)
 {
+    if (sizeof(_Tp) > 8) {
+      return __to_chars_integral(__first, __last, __value, 10, false_type());
+
     using __tx = __itoa::__traits<_Tp>;
     auto __diff = __last - __first;
 
@@ -426,7 +429,7 @@ inline _LIBCPP_INLINE_VISIBILITY to_chars_result
 __to_chars_integral(char* __first, char* __last, _Tp __value, int __base,
                     false_type)
 {
-  if (__base == 10)
+  if (__base == 10 && sizeof(_Tp) <= 8)
     return __to_chars_itoa(__first, __last, __value, false_type());
 
   ptrdiff_t __cap = __last - __first;

But we'd have to add test coverage too.

[Mordante]

libcxx/include/__format/formatter_integer.h

112–117

Since we've been shipping this implementation of std::to_chars for a while I didn't have a lot of urgency to fix it. I still want to address this issue later. I've identified more issues with <charconv> and I want to look at these later. For now I want to focus on the formatting library, but <charconv> is still high on my list. Of course if you want to implement this work around feel free to do so; I would be happy to review it.

[Mordante]

libcxx/include/__format/formatter.h

138

s/_LIBCPP_INLINE_VISIBILITY/_LIBCPP_HIDE_FROM_ABI/ here and all other occurrences in this patch.

[vitaut]

Looks good, thanks for addressing the comments.

libcxx/test/std/utilities/format/format.functions/tests.inc

380–381

Yeah, I think P1652 missed this part of the wording. The interpretation depends on whether you look at the original types or types after conversion per presentation specifiers. I suggest opening an LWG issue to clarify that we want the latter.

[Mordante]

Needs updates due to changes in D103368.

[Mordante]

Rebased and adjusted for changes in main
Retarget for LLVM 14
s/_LIBCPP_INLINE_VISIBILITY/_LIBCPP_HIDE_FROM_ABI/
Remove nodiscard
Remove obsolete unit test
Generate private module tests

[vitaut]

LGTM

libcxx/include/__format/formatter_integral.h

78–79

I suggest removing the "For example the 128-bit integral formatters use the 64-bit version to do the formatting." because it's a workaround that will go away per TODO in format(__int128_t __value, auto& __ctx) rendering this example wrong. In general I don't think there is a good reason for one specialization to reuse another.

314–320

Why not use a conditional expression?

return __format_unsigned_integral(__array.begin(), __array.end(),
                                  __value, __negative, 8, __ctx, __value != 0 ? "0" : nullptr);

364–365

Maybe add a TODO to format upper-case hex in one pass? As a nice side effect you could probably also kill dependency on <algorithm> which is super bloated in 20.

[Mordante]

libcxx/include/__format/formatter_integral.h

78–79

Valid point and I think the 128-bit work-around is the only place it's used.

314–320

No specific reason not to use it. But in this case looks better with the conditional expression.

364–365

I'll add a TODO. I'm not sure I'll be able to remove <algorithm>, I use more parts of it. But I can have a look to use the new smaller <__algorithm/foo.h> headers. That's something that has been worked on after my initial version of this patch.

[Mordante]

Addresses review comments.

[Mordante]

Rebased and adjusted for changes in main
s/_LIBCPP_INLINE_VISIBILITY/_LIBCPP_HIDE_FROM_ABI/

[Mordante]

Rebased and updated to changes in earlier commits in the series.

[Mordante]

Rebased.

[ldionne]

libcxx/include/__format/formatter_integer.h

112–117

Can we file a bug report to track this work item? I'm fearful we'll forget about it forever otherwise.

libcxx/include/__format/formatter_integral.h

490

This block is long enough that we can't easily see what #if the #endif relates to.

[Mordante]

libcxx/include/__format/formatter_integer.h

112–117

I've marked it as TODO's in the format related files, for example line 151 in this file
// TODO FMT Implement full 128 bit support.

So I don't fear we'll forget it. I don't consider format fully implemented before these TODO FMTs are resolved.

libcxx/docs/Cxx2aStatusIssuesStatus.csv

 "`3201 <https://wg21.link/LWG3201>`__","``lerp``\  should be marked as ``noexcept``\ ","Prague","|Complete|",""
 "`3226 <https://wg21.link/LWG3226>`__","``zoned_time``\  constructor from ``string_view``\  should accept ``zoned_time<Duration2, TimeZonePtr2>``\ ","Prague","",""
 "`3233 <https://wg21.link/LWG3233>`__","Broken requirements for ``shared_ptr``\  converting constructors","Prague","",""
 "`3237 <https://wg21.link/LWG3237>`__","LWG 3038 and 3190 have inconsistent PRs","Prague","",""
 "`3238 <https://wg21.link/LWG3238>`__","Insufficiently-defined behavior of ``std::function``\  deduction guides","Prague","",""
 "`3242 <https://wg21.link/LWG3242>`__","``std::format``\ : missing rules for ``arg-id``\  in ``width``\  and ``precision``\ ","Prague","|Complete|","Clang 13"
 "`3243 <https://wg21.link/LWG3243>`__","``std::format``\  and negative zeroes","Prague","",""
 "`3247 <https://wg21.link/LWG3247>`__","``ranges::iter_move``\  should perform ADL-only lookup of ``iter_move``\ ","Prague","",""
-"`3248 <https://wg21.link/LWG3248>`__","``std::format``\  ``#b``\ , ``#B``\ , ``#o``\ , ``#x``\ , and ``#X``\   presentation types misformat negative numbers","Prague","",""
+"`3248 <https://wg21.link/LWG3248>`__","``std::format``\  ``#b``\ , ``#B``\ , ``#o``\ , ``#x``\ , and ``#X``\   presentation types misformat negative numbers","Prague","|Complete|","13.0"
 "`3250 <https://wg21.link/LWG3250>`__","``std::format``\ : ``#``\  (alternate form) for NaN and inf","Prague","",""
 "`3251 <https://wg21.link/LWG3251>`__","Are ``std::format``\  alignment specifiers applied to string arguments?","Prague","",""
 "`3252 <https://wg21.link/LWG3252>`__","Parse locale's aware modifiers for commands are not consistent with POSIX spec","Prague","",""
 "`3254 <https://wg21.link/LWG3254>`__","Strike ``stop_token``\ 's ``operator!=``\ ","Prague","",""
 "`3255 <https://wg21.link/LWG3255>`__","``span``\ 's ``array``\  constructor is too strict","Prague","|Complete|",""
 "`3260 <https://wg21.link/LWG3260>`__","``year_month*``\  arithmetic rejects durations convertible to years","Prague","",""
 "`3262 <https://wg21.link/LWG3262>`__","Formatting of negative durations is not specified","Prague","",""
 "`3264 <https://wg21.link/LWG3264>`__","``sized_range``\  and ``ranges::size``\  redundantly use ``disable_sized_range``\ ","Prague","",""

libcxx/docs/Cxx2aStatusPaperStatus.csv

 "`P1523 <https://wg21.link/P1523>`__","LWG","Views and Size Types","Cologne","",""
 "`P1612 <https://wg21.link/P1612>`__","LWG","Relocate Endian's Specification","Cologne","|Complete|","10.0"
 "`P1614 <https://wg21.link/P1614>`__","LWG","The Mothership has Landed","Cologne","|In Progress|",""
 "`P1638 <https://wg21.link/P1638>`__","LWG","basic_istream_view::iterator should not be copyable","Cologne","",""
 "`P1643 <https://wg21.link/P1643>`__","LWG","Add wait/notify to atomic_ref","Cologne","",""
 "`P1644 <https://wg21.link/P1644>`__","LWG","Add wait/notify to atomic<shared_ptr>","Cologne","",""
 "`P1650 <https://wg21.link/P1650>`__","LWG","Output std::chrono::days with 'd' suffix","Cologne","",""
 "`P1651 <https://wg21.link/P1651>`__","LWG","bind_front should not unwrap reference_wrapper","Cologne","|Complete|","13.0"
-"`P1652 <https://wg21.link/P1652>`__","LWG","Printf corner cases in std::format","Cologne","",""
+"`P1652 <https://wg21.link/P1652>`__","LWG","Printf corner cases in std::format","Cologne","|In Progress|",""
 "`P1661 <https://wg21.link/P1661>`__","LWG","Remove dedicated precalculated hash lookup interface","Cologne","|Nothing To Do|",""
 "`P1754 <https://wg21.link/P1754>`__","LWG","Rename concepts to standard_case for C++20, while we still can","Cologne","|In Progress|",""
 "","","","","",""
 "`P0883 <https://wg21.link/P0883>`__","LWG","Fixing Atomic Initialization","Belfast","* *",""
 "`P1391 <https://wg21.link/P1391>`__","LWG","Range constructor for std::string_view","Belfast","* *",""
 "`P1394 <https://wg21.link/P1394>`__","LWG","Range constructor for std::span","Belfast","* *",""
 "`P1456 <https://wg21.link/P1456>`__","LWG","Move-only views","Belfast","* *",""
 "`P1622 <https://wg21.link/P1622>`__","LWG","Mandating the Standard Library: Clause 32 - Thread support library","Belfast","* *",""

libcxx/include/CMakeLists.txt

set(files
   __format/format_arg.h
   __format/format_args.h
   __format/format_context.h
   __format/format_error.h
   __format/format_fwd.h
   __format/format_parse_context.h
   __format/format_string.h
   __format/formatter.h
+  __format/formatter_integer.h
+  __format/formatter_integral.h
   __format/formatter_string.h
   __format/parser_std_format_spec.h
   __function_like.h
   __functional_03
   __functional_base
   __functional_base_03
   __hash_table
   __iterator/advance.h

libcxx/include/__format/formatter.h

__write(__output_iterator<const _CharT&> auto __out_it, const _CharT* __first,
 
   pair<size_t, size_t> __padding = __padding_size(__size, __width, __alignment);
   __out_it = _VSTD::fill_n(_VSTD::move(__out_it), __padding.first, __fill);
   __out_it = _VSTD::copy(__first, __last, _VSTD::move(__out_it));
   return _VSTD::fill_n(_VSTD::move(__out_it), __padding.second, __fill);
 }
 
 /**
+ * @overload
+ *
+ * Uses a transformation operation before writing an element.
+ */
+template <class _CharT, class _UnaryOperation, class _Fill>
+[[nodiscard]] _LIBCPP_INLINE_VISIBILITY auto
+__write(__output_iterator<const _CharT&> auto __out_it, const _CharT* __first,
+        const _CharT* __last, size_t __size, _UnaryOperation __op,
+        size_t __width, _Fill __fill,
+        __format_spec::_Flags::_Alignment __alignment) -> decltype(__out_it) {
+
+  _LIBCPP_ASSERT(__first <= __last, "Not a valid range");
+  _LIBCPP_ASSERT(__size < __width, "Precondition failure");
+
+  pair<size_t, size_t> __padding = __padding_size(__size, __width, __alignment);
+  __out_it = _VSTD::fill_n(_VSTD::move(__out_it), __padding.first, __fill);
+  __out_it = _VSTD::transform(__first, __last, _VSTD::move(__out_it), __op);
+  return _VSTD::fill_n(_VSTD::move(__out_it), __padding.second, __fill);
+}
+
+/**
  * Writes Unicode input to the output with the required padding.
  *
  * This function does almost the same as the @ref __write function, but handles
  * the width estimation of the Unicode input.
  *
  * @param __str       The range [@a __first, @a __last).
  * @param __precision The width to truncate the output to, use
  *                    @ref __format::__number_max for no limit.

libcxx/include/__format/formatter_integer.h

+// -*- C++ -*-
+//===----------------------------------------------------------------------===//
+//
+// Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
+// See https://llvm.org/LICENSE.txt for license information.
+// SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
+//
+//===----------------------------------------------------------------------===//
+
+#ifndef _LIBCPP___FORMAT_FORMATTER_INTEGER_H
+#define _LIBCPP___FORMAT_FORMATTER_INTEGER_H
+
+#include <__config>
+#include <__format/formatter_integral.h>
+#include <__format/formatter.h>
+#include <limits>
+
+#if !defined(_LIBCPP_HAS_NO_PRAGMA_SYSTEM_HEADER)
+#pragma GCC system_header
+#endif
+
+_LIBCPP_PUSH_MACROS
+#include <__undef_macros>
+
+_LIBCPP_BEGIN_NAMESPACE_STD
+
+#if _LIBCPP_STD_VER > 17
+
+// TODO FMT Remove this once we require compilers with proper C++20 support.
+// If the compiler has no concepts support, the format header will be disabled.
+// Without concepts support enable_if needs to be used and that too much effort
+// to support compilers with partial C++20 support.
+#if !defined(_LIBCPP_HAS_NO_CONCEPTS) &&                                       \
+    !defined(_LIBCPP_HAS_NO_BUILTIN_IS_CONSTANT_EVALUATED)
+
+namespace __format_spec {
+
+template <class _CharT>
+class _LIBCPP_TEMPLATE_VIS __parser_integer : public __parser_integral<_CharT> {
+public:
+  using _Base = __parser_integral<_CharT>;
+
+  // TODO FMT The function is constexpr to prepare for P2216.
+  _LIBCPP_INLINE_VISIBILITY constexpr auto parse(auto& __parse_ctx)
+      -> decltype(__parse_ctx.begin()) {
+    auto __it = _Base::parse(__parse_ctx);
+
+    this->__handle_integer();
+    switch (this->__type) {
+    case _Flags::_Type::__default:
+      this->__type = _Flags::_Type::__decimal;
+      break;
+
+    case _Flags::_Type::__binary_lower_case:
+    case _Flags::_Type::__binary_upper_case:
+    case _Flags::_Type::__char:
+    case _Flags::_Type::__octal:
+    case _Flags::_Type::__decimal:
+    case _Flags::_Type::__hexadecimal_lower_case:
+    case _Flags::_Type::__hexadecimal_upper_case:
+      break;
+
+    default:
+      __throw_format_error(
+          "Format-spec type has a type not supported for an integer argument");
+    }
+    return __it;
+  }
+};
+
+template <class _CharT>
+using __formatter_integer = __formatter_integral<__parser_integer<_CharT>>;
+
+} // namespace __format_spec
+
+// [format.formatter.spec]/2.3
+// For each charT, for each cv-unqualified arithmetic type ArithmeticT other
+// than char, wchar_t, char8_t, char16_t, or char32_t, a specialization
+
+// Signed integral types.
+template <class _CharT>
+struct _LIBCPP_TEMPLATE_VIS _LIBCPP_AVAILABILITY_FORMAT
+    formatter<signed char, _CharT>
+    : public __format_spec::__formatter_integer<_CharT> {};
+template <class _CharT>
+struct _LIBCPP_TEMPLATE_VIS _LIBCPP_AVAILABILITY_FORMAT formatter<short, _CharT>
+    : public __format_spec::__formatter_integer<_CharT> {};
+template <class _CharT>
+struct _LIBCPP_TEMPLATE_VIS _LIBCPP_AVAILABILITY_FORMAT formatter<int, _CharT>
+    : public __format_spec::__formatter_integer<_CharT> {};
+template <class _CharT>
+struct _LIBCPP_TEMPLATE_VIS _LIBCPP_AVAILABILITY_FORMAT formatter<long, _CharT>
+    : public __format_spec::__formatter_integer<_CharT> {};
+template <class _CharT>
+struct _LIBCPP_TEMPLATE_VIS _LIBCPP_AVAILABILITY_FORMAT
+    formatter<long long, _CharT>
+    : public __format_spec::__formatter_integer<_CharT> {};
+#ifndef _LIBCPP_HAS_NO_INT128
+template <class _CharT>
+struct _LIBCPP_TEMPLATE_VIS _LIBCPP_AVAILABILITY_FORMAT
+    formatter<__int128_t, _CharT>
+    : public __format_spec::__formatter_integer<_CharT> {
+  using _Base = __format_spec::__formatter_integer<_CharT>;
+
+  _LIBCPP_INLINE_VISIBILITY auto format(__int128_t __value, auto& __ctx)
+      -> decltype(__ctx.out()) {
+    // TODO FMT Implement full 128 bit support.
+    using _To = long long;
+    if (__value < numeric_limits<_To>::min() ||
+        __value > numeric_limits<_To>::max())
+      __throw_format_error("128 bit value is outside of implemented range");
+
+    return _Base::format(static_cast<_To>(__value), __ctx);
+  }
+};
+#endif
+
+// Unsigned integral types.
+template <class _CharT>
+struct _LIBCPP_TEMPLATE_VIS _LIBCPP_AVAILABILITY_FORMAT
+    formatter<unsigned char, _CharT>
+    : public __format_spec::__formatter_integer<_CharT> {};
+template <class _CharT>
+struct _LIBCPP_TEMPLATE_VIS _LIBCPP_AVAILABILITY_FORMAT
+    formatter<unsigned short, _CharT>
+    : public __format_spec::__formatter_integer<_CharT> {};
+template <class _CharT>
+struct _LIBCPP_TEMPLATE_VIS _LIBCPP_AVAILABILITY_FORMAT
+    formatter<unsigned, _CharT>
+    : public __format_spec::__formatter_integer<_CharT> {};
+template <class _CharT>
+struct _LIBCPP_TEMPLATE_VIS _LIBCPP_AVAILABILITY_FORMAT
+    formatter<unsigned long, _CharT>
+    : public __format_spec::__formatter_integer<_CharT> {};
+template <class _CharT>
+struct _LIBCPP_TEMPLATE_VIS _LIBCPP_AVAILABILITY_FORMAT
+    formatter<unsigned long long, _CharT>
+    : public __format_spec::__formatter_integer<_CharT> {};
+#ifndef _LIBCPP_HAS_NO_INT128
+template <class _CharT>
+struct _LIBCPP_TEMPLATE_VIS _LIBCPP_AVAILABILITY_FORMAT
+    formatter<__uint128_t, _CharT>
+    : public __format_spec::__formatter_integer<_CharT> {
+  using _Base = __format_spec::__formatter_integer<_CharT>;
+
+  _LIBCPP_INLINE_VISIBILITY auto format(__uint128_t __value, auto& __ctx)
+      -> decltype(__ctx.out()) {
+    // TODO FMT Implement full 128 bit support.
+    using _To = unsigned long long;
+    if (__value < numeric_limits<_To>::min() ||
+        __value > numeric_limits<_To>::max())
+      __throw_format_error("128 bit value is outside of implemented range");
+
+    return _Base::format(static_cast<_To>(__value), __ctx);
+  }
+};
+#endif
+
+#endif // !defined(_LIBCPP_HAS_NO_CONCEPTS) && !defined(_LIBCPP_HAS_NO_BUILTIN_IS_CONSTANT_EVALUATED)
+
+#endif //_LIBCPP_STD_VER > 17
+
+_LIBCPP_END_NAMESPACE_STD
+
+_LIBCPP_POP_MACROS
+
+#endif // _LIBCPP___FORMAT_FORMATTER_INTEGER_H

libcxx/include/__format/formatter_integral.h

+// -*- C++ -*-
+//===----------------------------------------------------------------------===//
+//
+// Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
+// See https://llvm.org/LICENSE.txt for license information.
+// SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
+//
+//===----------------------------------------------------------------------===//
+
+#ifndef _LIBCPP___FORMAT_FORMATTER_INTEGRAL_H
+#define _LIBCPP___FORMAT_FORMATTER_INTEGRAL_H
+
+#include <__config>
+#include <__format/formatter.h>
+#include <__format/parser_std_format_spec.h>
+#include <algorithm>
+#include <array>
+#include <charconv>
+#include <concepts>
+#include <limits>
+
+#if !defined(_LIBCPP_HAS_NO_PRAGMA_SYSTEM_HEADER)
+#pragma GCC system_header
+#endif
+
+_LIBCPP_PUSH_MACROS
+#include <__undef_macros>
+
+_LIBCPP_BEGIN_NAMESPACE_STD
+
+#if _LIBCPP_STD_VER > 17
+
+// TODO FMT Remove this once we require compilers with proper C++20 support.
+// If the compiler has no concepts support, the format header will be disabled.
+// Without concepts support enable_if needs to be used and that too much effort
+// to support compilers with partial C++20 support.
+#if !defined(_LIBCPP_HAS_NO_CONCEPTS) &&                                       \
+    !defined(_LIBCPP_HAS_NO_BUILTIN_IS_CONSTANT_EVALUATED)
+
+/**
+ * Integral formatting classes.
+ *
+ * There are two types used here:
+ * * C++-type, the type as used in C++.
+ * * format-type, the output type specified in the std-format-spec.
+ *
+ * Design of the integral formatters consists of several layers.
+ * * @ref __parser_integral The basic std-format-spec parser for all integral
+ *   classes. This parser does the basic sanity checks. It also contains some
+ *   helper functions that are nice to have available for all parsers.
+ * * A C++-type specific parser. These parsers must derive from
+ *   @ref __parser_integral. Their task is to validate whether the parsed
+ *   std-format-spec is valid for the C++-type and selected format-type. After
+ *   validation they need to make sure all members are properly set. For
+ *   example, when the alignment hasn't changed it needs to set the proper
+ *   default alignment for the format-type. The following parsers are available:
+ *   - @ref __parser_integer
+ *   - @ref __parser_char
+ *   - @ref __parser_bool
+ * * A general formatter for all integral types @ref __formatter_integral. This
+ *   formatter can handle all formatting of integers and characters. The class
+ *   derives from the proper formatter.
+ *   Note the boolean string format-type isn't supported in this class.
+ * * A typedef C++-type group:
+ *   * @ref __formatter_integer
+ *   * @ref __formatter_char
+ *   * @ref __formatter_bool
+ * * Then every C++-type has its own formatter specializations. They inherit
+ *   from the C++-type group typedef. Most specializations need nothing else.
+ *   Others need some additional specializations in this class. For example the
+ *   128-bit integral formatters use the 64-bit version to do the formatting.
+ */
+namespace __format_spec {
+
+template <class _CharT>
+class _LIBCPP_TEMPLATE_VIS __parser_integral
+    : public __parser_std<_CharT, __parser_no_precision, __parser_sign,
+                          __parser_alternate_form, __parser_zero_padding,
+                          __parser_locale_specific_form> {
+public:
+  /**
+   * Handles the post-parsing updates for the integer types.
+   *
+   * Updates the zero-padding and alignment for integer types.
+   *
+   * [format.string.std]/13
+   *   If the 0 character and an align option both appear, the 0 character is
+   *   ignored.
+   *
+   * For the formatter a @ref __default alignment means zero-padding. Update
+   * the alignment based on parsed format string.
+   */
+  _LIBCPP_INLINE_VISIBILITY constexpr void __handle_integer() noexcept {
+    this->__zero_padding &= this->__alignment == _Flags::_Alignment::__default;
+    if (!this->__zero_padding &&
+        this->__alignment == _Flags::_Alignment::__default)
+      this->__alignment = _Flags::_Alignment::__right;
+  }
+
+  /**
+   * Handles the post-parsing updates for the character types.
+   *
+   * Sets the alignment and validates the format flags set for a character type.
+   *
+   * At the moment the validation for a character and a Boolean behave the
+   * same, but this may change in the future.
+   */
+  _LIBCPP_INLINE_VISIBILITY constexpr void __handle_char() { __handle_bool(); }
+
+  /**
+   * Handles the post-parsing updates for the Boolean types.
+   *
+   * Sets the alignment and validates the format flags set for a Boolean type.
+   */
+  _LIBCPP_INLINE_VISIBILITY constexpr void __handle_bool() {
+    if (this->__sign != _Flags::_Sign::__default)
+      __throw_format_error("Format-spec sign not allowed");
+
+    if (this->__alternate_form)
+      __throw_format_error("Format-spec alternate form not allowed");
+
+    if (this->__zero_padding)
+      __throw_format_error("Format-spec zero-padding not allowed");
+
+    if (this->__alignment == _Flags::_Alignment::__default)
+      this->__alignment = _Flags::_Alignment::__left;
+  }
+};
+
+/** Wrapper around @ref to_chars, returning the output pointer. */
+template <class _Tp>
+[[nodiscard]] _LIBCPP_INLINE_VISIBILITY char*
+__to_buffer(char* __first, char* __last, _Tp __value, int __base) {
+  // TODO FMT Evaluate code overhead due to not calling the internal function
+  // directly. (Should be zero overhead.)
+  to_chars_result __r = _VSTD::to_chars(__first, __last, __value, __base);
+  _LIBCPP_ASSERT(__r.ec == errc(0), "Internal buffer too small.");
+  return __r.ptr;
+}
+
+/**
+ * Helper to determine the buffer size to output a integer in Base @em x.
+ *
+ * There are several overloads for the supported bases. The function uses the
+ * base as template argument so it can be used in a constant expression.
+ */
+template <unsigned_integral _Tp, size_t _Base>
+[[nodiscard]] _LIBCPP_INLINE_VISIBILITY constexpr size_t
+__buffer_size() noexcept requires(_Base == 2) {
+  return numeric_limits<_Tp>::digits // The number of binary digits.
+         + 2                         // Reserve space for the '0[Bb]' prefix.
+         + 1;                        // Reserve space for the sign.
+}
+
+template <unsigned_integral _Tp, size_t _Base>
+[[nodiscard]] _LIBCPP_INLINE_VISIBILITY constexpr size_t
+__buffer_size() noexcept requires(_Base == 8) {
+  return numeric_limits<_Tp>::digits // The number of binary digits.
+             / 3                     // Adjust to octal.
+         + 1                         // Reserve space for the '0' prefix.
+         + 1;                        // Reserve space for the sign.
+}
+
+template <unsigned_integral _Tp, size_t _Base>
+[[nodiscard]] _LIBCPP_INLINE_VISIBILITY constexpr size_t
+__buffer_size() noexcept requires(_Base == 10) {
+  return numeric_limits<_Tp>::digits10 // The floored value.
+         + 1                           // Turn floor to ceil.
+         + 1;                          // Reserve space for the sign.
+}
+
+template <unsigned_integral _Tp, size_t _Base>
+[[nodiscard]] _LIBCPP_INLINE_VISIBILITY constexpr size_t
+__buffer_size() noexcept requires(_Base == 16) {
+  return numeric_limits<_Tp>::digits // The number of binary digits.
+             / 4                     // Adjust to hexadecimal.
+         + 2                         // Reserve space for the '0[Xx]' prefix.
+         + 1;                        // Reserve space for the sign.
+}
+
+[[nodiscard]] inline _LIBCPP_INLINE_VISIBILITY char*
+__insert_sign(char* __buf, bool __negative, _Flags::_Sign __sign) {
+  if (__negative)
+    *__buf++ = '-';
+  else
+    switch (__sign) {
+    case _Flags::_Sign::__default:
+    case _Flags::_Sign::__minus:
+      // No sign added.
+      break;
+    case _Flags::_Sign::__plus:
+      *__buf++ = '+';
+      break;
+    case _Flags::_Sign::__space:
+      *__buf++ = ' ';
+      break;
+    }
+
+  return __buf;
+}
+
+[[nodiscard]] inline _LIBCPP_INLINE_VISIBILITY constexpr char
+__hex_to_upper(char c) {
+  switch (c) {
+  case 'a':
+    return 'A';
+  case 'b':
+    return 'B';
+  case 'c':
+    return 'C';
+  case 'd':
+    return 'D';
+  case 'e':
+    return 'E';
+  case 'f':
+    return 'F';
+  }
+  return c;
+}
+
+#ifndef _LIBCPP_HAS_NO_LOCALIZATION
+/**
+ * Determines the required grouping based on the size of the input.
+ *
+ * The grouping's last element will be repeated. For simplicity this repeating
+ * is unwrapped based on the length of the input. (When the input is short some
+ * groups are not processed.)
+ *
+ * @returns the size of the groups to write. This means the number of
+ * separator characters written is size() - 1.
+ *
+ * @note Since zero-sized groups cause issues they are silently ignored.
+ *
+ * @note The grouping field of the local is always a @c std::string, regardless
+ * whether the @c std::numpunct's type is @c char or @c wchar_t.
+ */
+[[nodiscard]] _LIBCPP_INLINE_VISIBILITY inline string
+__determine_grouping(ptrdiff_t __size, const string& __grouping) {
+  _LIBCPP_ASSERT(!__grouping.empty() && __size > __grouping[0],
+                 "The slow grouping formatting is used while there will be no "
+                 "separators written");
+  string __r;
+  auto __end = __grouping.end() - 1;
+  auto __ptr = __grouping.begin();
+
+  while (true) {
+    __size -= *__ptr;
+    if (__size > 0)
+      __r.push_back(*__ptr);
+    else {
+      // __size <= 0 so the value pushed will be <= *__ptr.
+      __r.push_back(*__ptr + __size);
+      return __r;
+    }
+
+    // Proceed to the next group.
+    if (__ptr != __end) {
+      do {
+        ++__ptr;
+        // Skip 0-width groupings
+      } while (*__ptr == 0 && __ptr != __end);
+    }
+  }
+
+  _LIBCPP_UNREACHABLE();
+}
+#endif
+
+template <class _Parser>
+class _LIBCPP_TEMPLATE_VIS __formatter_integral : public _Parser {
+public:
+  using _CharT = typename _Parser::char_type;
+
+  template <integral _Tp>
+  _LIBCPP_INLINE_VISIBILITY auto format(_Tp __value, auto& __ctx)
+      -> decltype(__ctx.out()) {
+
+    if (this->__width_as_arg)
+      this->__substitute_width_arg_id(__ctx.arg(this->__width));
+
+    if (this->__type == _Flags::_Type::__char)
+      return __format_char(__value, __ctx);
+
+    if constexpr (unsigned_integral<_Tp>) {
+      return __format_usigned_integral(__value, __ctx);
+    } else {
+      // Depending on the std-format-spec string the sign and the value
+      // might not be outputted together:
+      // - alternate form may insert a prefix string.
+      // - zero-padding may insert additional '0' characters.
+      // Therefore the value is processed as a positive unsigned value.
+      // The function @ref __insert_sign will a '-' when the value was negative.
+      auto __r = __to_unsigned_like(__value);
+      if (__value < 0) {
+        __negative = true;
+        __r = __complement(__r);
+      }
+      return __format_usigned_integral(__r, __ctx);
+    }
+  }
+
+private:
+  /** Generic formatting for format-type c. */
+  [[nodiscard]] _LIBCPP_INLINE_VISIBILITY auto
+  __format_char(integral auto __value, auto& __ctx) -> decltype(__ctx.out()) {
+    if (this->__alignment == _Flags::_Alignment::__default)
+      this->__alignment = _Flags::_Alignment::__right;
+
+    using _Tp = decltype(__value);
+    if constexpr (!same_as<_CharT, _Tp>) {
+      // cmp_less and cmp_greater can't be used for character types.
+      if constexpr (signed_integral<_CharT> == signed_integral<_Tp>) {
+        if (__value < numeric_limits<_CharT>::min() ||
+            __value > numeric_limits<_CharT>::max())
+          __throw_format_error(
+              "Integral value outside the range of the char type.");
+      } else if constexpr (signed_integral<_CharT>) {
+        // _CharT is signed _Tp is unsigned
+        if (__value >
+            static_cast<make_signed_t<_CharT>>(numeric_limits<_CharT>::max()))
+          __throw_format_error(
+              "Integral value outside the range of the char type.");
+      } else {
+        // _CharT is unsigned _Tp is signed
+        if (__value < 0 || static_cast<make_unsigned_t<_Tp>>(__value) >
+                               numeric_limits<_CharT>::max())
+          __throw_format_error(
+              "Integral value outside the range of the char type.");
+      }
+    }
+
+    const auto __c = static_cast<_CharT>(__value);
+    return __write(_VSTD::addressof(__c), _VSTD::addressof(__c) + 1,
+                   __ctx.out());
+  }
+
+  /**
+   * Generic formatting for format-type bBdoxX.
+   *
+   * This small wrapper allocates a buffer with the required size. Then calls
+   * the real formatter with the buffer and the prefix for the base.
+   */
+  [[nodiscard]] _LIBCPP_INLINE_VISIBILITY auto
+  __format_usigned_integral(unsigned_integral auto __value, auto& __ctx)
+      -> decltype(__ctx.out()) {
+    switch (this->__type) {
+    case _Flags::_Type::__binary_lower_case: {
+      array<char, __buffer_size<decltype(__value), 2>()> __array;
+      return __format_usigned_integral(__array.begin(), __array.end(), __value,
+                                       2, __ctx, "0b");
+    }
+    case _Flags::_Type::__binary_upper_case: {
+      array<char, __buffer_size<decltype(__value), 2>()> __array;
+      return __format_usigned_integral(__array.begin(), __array.end(), __value,
+                                       2, __ctx, "0B");
+    }
+    case _Flags::_Type::__octal: {
+      // Octal is special; if the __value == 0 there's no prefix.
+      array<char, __buffer_size<decltype(__value), 8>()> __array;
+      if (__value == 0)
+        return __format_usigned_integral(__array.begin(), __array.end(),
+                                         __value, 8, __ctx, nullptr);
+      else
+        return __format_usigned_integral(__array.begin(), __array.end(),
+                                         __value, 8, __ctx, "0");
+    }
+    case _Flags::_Type::__decimal: {
+      array<char, __buffer_size<decltype(__value), 10>()> __array;
+      return __format_usigned_integral(__array.begin(), __array.end(), __value,
+                                       10, __ctx, nullptr);
+    }
+    case _Flags::_Type::__hexadecimal_lower_case: {
+      array<char, __buffer_size<decltype(__value), 16>()> __array;
+      return __format_usigned_integral(__array.begin(), __array.end(), __value,
+                                       16, __ctx, "0x");
+    }
+    case _Flags::_Type::__hexadecimal_upper_case: {
+      array<char, __buffer_size<decltype(__value), 16>()> __array;
+      return __format_usigned_integral(__array.begin(), __array.end(), __value,
+                                       16, __ctx, "0X");
+    }
+    default:
+      _LIBCPP_ASSERT(false, "The parser should have validated the type");
+      _LIBCPP_UNREACHABLE();
+    }
+  }
+
+  template <class _Tp>
+  requires(same_as<char, _Tp> || same_as<wchar_t, _Tp>)
+      [[nodiscard]] _LIBCPP_INLINE_VISIBILITY
+      auto __write(const _Tp* __first, const _Tp* __last, auto __out_it)
+          -> decltype(__out_it) {
+
+    unsigned __size = __last - __first;
+    if (__size >= this->__width) {
+      if (this->__type != _Flags::_Type::__hexadecimal_upper_case) [[likely]]
+        return _VSTD::copy(__first, __last, _VSTD::move(__out_it));
+
+      return _VSTD::transform(__first, __last, _VSTD::move(__out_it),
+                              __hex_to_upper);
+    }
+    if (this->__type != _Flags::_Type::__hexadecimal_upper_case) [[likely]]
+      return __formatter::__write(_VSTD::move(__out_it), __first, __last,
+                                  __size, this->__width, this->__fill,
+                                  this->__alignment);
+
+    return __formatter::__write(_VSTD::move(__out_it), __first, __last, __size,
+                                __hex_to_upper, this->__width, this->__fill,
+                                this->__alignment);
+  }
+
+  [[nodiscard]] _LIBCPP_INLINE_VISIBILITY auto __format_usigned_integral(
+      char* __begin, char* __end, unsigned_integral auto __value, int __base,
+      auto& __ctx, const char* __prefix) -> decltype(__ctx.out()) {
+    char* __first = __insert_sign(__begin, __negative, this->__sign);
+    if (this->__alternate_form && __prefix)
+      while (*__prefix)
+        *__first++ = *__prefix++;
+
+    char* __last = __to_buffer(__first, __end, __value, __base);
+#ifndef _LIBCPP_HAS_NO_LOCALIZATION
+    if (this->__locale_specific_form) {
+      const auto& __np = use_facet<numpunct<_CharT>>(__ctx.locale());
+      string __grouping = __np.grouping();
+      ptrdiff_t __size = __last - __first;
+      // Writing the grouped form has more overhead than the normal output
+      // routines. If there will be no separators written the locale-specific
+      // form is identical to the normal routine. Test whether to grouped form
+      // is required.
+      if (!__grouping.empty() && __size > __grouping[0]) {
+        return __format_grouping(__ctx.out(), __begin, __first, __last,
+                                 __determine_grouping(__size, __grouping),
+                                 __np.thousands_sep());
+      }
+    }
+#endif
+    auto __out_it = __ctx.out();
+    if (this->__alignment != _Flags::_Alignment::__default)
+      __first = __begin;
+    else {
+      // __buf contains [sign][prefix]data
+      //                              ^ location of __first
+      // The zero padding is done like:
+      // - Write [sign][prefix]
+      // - Write data right aligned with '0' as fill character.
+      __out_it = _VSTD::copy(__begin, __first, _VSTD::move(__out_it));
+      this->__alignment = _Flags::_Alignment::__right;
+      this->__fill = _CharT('0');
+      unsigned __size = __first - __begin;
+      this->__width -= _VSTD::min(__size, this->__width);
+    }
+
+    return __write(__first, __last, _VSTD::move(__out_it));
+  }
+
+#ifndef _LIBCPP_HAS_NO_LOCALIZATION
+  /** Format's the locale-specific form's groupings. */
+  template <class _OutIt, class _CharT>
+  [[nodiscard]] _LIBCPP_INLINE_VISIBILITY _OutIt
+  __format_grouping(_OutIt __out_it, const char* __begin, const char* __first,
+                    const char* __last, string&& __grouping, _CharT __sep) {
+
+    // TODO FMT This function duplicates some functionality of the normal
+    // output routines. Evaluate whether these parts can be efficiently
+    // combined with the existing routines.
+
+    int __size = (__first - __begin)        // [sign][prefix]
+                 + (__last - __first)       // data
+                 + (__grouping.size() - 1); // number of separator characters
+
+    pair<size_t, size_t> __padding = {0, 0};
+    if (this->__alignment == _Flags::_Alignment::__default) {
+      // Write [sign][prefix].
+      __out_it = _VSTD::copy(__begin, __first, _VSTD::move(__out_it));
+
+      if (this->__width > __size) {
+        // Write zero padding.
+        __padding.first = this->__width - __size;
+        __out_it = _VSTD::fill_n(_VSTD::move(__out_it), this->__width - __size,
+                                 _CharT('0'));
+      }
+    } else {
+      if (this->__width > __size) {
+        // Determine padding and write padding.
+        __padding = __formatter::__padding_size(__size, this->__width,
+                                                this->__alignment);
+
+        __out_it =
+            _VSTD::fill_n(_VSTD::move(__out_it), __padding.first, this->__fill);
+      }
+      // Write [sign][prefix].
+      __out_it = _VSTD::copy(__begin, __first, _VSTD::move(__out_it));
+    }
+
+    auto __r = __grouping.rbegin();
+    auto __e = __grouping.rend() - 1;
+    _LIBCPP_ASSERT(__r != __e, "The slow grouping formatting is used while "
+                               "there will be no separators written.");
+    // The output is divided in small groups of numbers to write:
+    // - A group before the first separator.
+    // - A separator and a group, repeated for the number of separators.
+    // - A group after the last separator.
+    // This loop achieves that process by testing the termination condition
+    // midway in the loop.
+    //
+    // TODO FMT This loop evaluate the loop invariant `this->__type !=
+    // _Flags::_Type::__hexadecimal_upper_case` for every iteration. (This test
+    // happens in the __write call.) Benchmark whether making two loops an
+    // hoisting the invariant is worth the effort.
+    while (true) {
+      if (this->__type == _Flags::_Type::__hexadecimal_upper_case) {
+        __last = __first + *__r;
+        __out_it = _VSTD::transform(__first, __last, _VSTD::move(__out_it),
+                                    __hex_to_upper);
+        __first = __last;
+      } else {
+        __out_it = _VSTD::copy_n(__first, *__r, _VSTD::move(__out_it));
+        __first += *__r;
+      }
+
+      if (__r == __e)
+        break;
+
+      ++__r;
+      *__out_it++ = __sep;
+    }
+
+    return _VSTD::fill_n(_VSTD::move(__out_it), __padding.second, this->__fill);
+  }
+#endif
+
+private:
+  bool __negative{false};
+};
+
+} // namespace __format_spec
+
+#endif // !defined(_LIBCPP_HAS_NO_CONCEPTS) && !defined(_LIBCPP_HAS_NO_BUILTIN_IS_CONSTANT_EVALUATED)
+
+#endif //_LIBCPP_STD_VER > 17
+
+_LIBCPP_END_NAMESPACE_STD
+
+_LIBCPP_POP_MACROS
+
+#endif // _LIBCPP___FORMAT_FORMATTER_INTEGRAL_H

libcxx/include/format

 #include <__debug>
 #include <__format/format_arg.h>
 #include <__format/format_args.h>
 #include <__format/format_context.h>
 #include <__format/format_error.h>
 #include <__format/format_parse_context.h>
 #include <__format/format_string.h>
 #include <__format/formatter.h>
+#include <__format/formatter_integer.h>
 #include <__format/formatter_string.h>
 #include <__format/parser_std_format_spec.h>
 #include <concepts>
 #include <string>
 #include <string_view>
 #include <type_traits>
 #include <version>
 
private:
   _LIBCPP_HIDDEN static wstring
   __convert(_Uv __value) requires(same_as<_CharT, wchar_t>) {
     return _VSTD::to_wstring(__value);
   }
 
   template <class _Uv>
   _LIBCPP_HIDDEN auto __handle_format(_Uv __value, auto& __ctx)
       -> decltype(__ctx.out())
-#ifndef _LIBCPP_HAS_NO_INT128
-          requires(!same_as<_Uv, __int128_t> && !same_as<_Uv, __uint128_t>)
-#endif
   {
     // TODO FMT Implement using formatting arguments
     // TODO FMT Improve PoC since using std::to_string is inefficient.
     // Note the code doesn't use std::string::iterator since the unit tests
     // test with debug iterators and they fail with strings created from
     // std::to_string.
     auto __str = __convert(__value);
     auto __out_it = __ctx.out();
     for (size_t __i = 0, __e = __str.size(); __i != __e; ++__i)
       *__out_it++ = __str[__i];
     return __out_it;
   }
-#ifndef _LIBCPP_HAS_NO_INT128
-  template <class _Uv>
-  _LIBCPP_HIDDEN auto __handle_format(_Uv __value, auto& __ctx)
-      -> decltype(__ctx.out()) requires(same_as<_Uv, __int128_t> ||
-                                        same_as<_Uv, __uint128_t>) {
-    using _To = conditional_t<is_signed_v<_Uv>, long long, unsigned long long>;
-    // TODO FMT Implement full 128 bit support.
-    if (__value < numeric_limits<_To>::min() ||
-        __value > numeric_limits<_To>::max())
-      __throw_format_error("128 bit value is outside of implemented range");
-
-    return __handle_format(static_cast<_To>(__value), __ctx);
-  }
-#endif
 };
 } // namespace __format
 
 // These specializations are helper stubs and not proper formatters.
 // TODO FMT Implement the proper formatter specializations.
 
 // [format.formatter.spec]/2.1 The specializations
 
struct _LIBCPP_TEMPLATE_VIS formatter<bool, _CharT> {
   auto format(bool __b, auto& __ctx) -> decltype(__ctx.out()) {
     // TODO FMT Implement using formatting arguments
     auto __out_it = __ctx.out();
     *__out_it++ = _CharT('0') + __b;
     return __out_it;
   }
 };
 
-// Signed integral types.
-template <class _CharT>
-struct _LIBCPP_TEMPLATE_VIS formatter<signed char, _CharT>
-    : public __format::__formatter_arithmetic<signed char, _CharT> {};
-template <class _CharT>
-struct _LIBCPP_TEMPLATE_VIS formatter<short, _CharT>
-    : public __format::__formatter_arithmetic<short, _CharT> {};
-template <class _CharT>
-struct _LIBCPP_TEMPLATE_VIS formatter<int, _CharT>
-    : public __format::__formatter_arithmetic<int, _CharT> {};
-template <class _CharT>
-struct _LIBCPP_TEMPLATE_VIS formatter<long, _CharT>
-    : public __format::__formatter_arithmetic<long, _CharT> {};
-template <class _CharT>
-struct _LIBCPP_TEMPLATE_VIS formatter<long long, _CharT>
-    : public __format::__formatter_arithmetic<long long, _CharT> {};
-#ifndef _LIBCPP_HAS_NO_INT128
-template <class _CharT>
-struct _LIBCPP_TEMPLATE_VIS formatter<__int128_t, _CharT>
-    : public __format::__formatter_arithmetic<__int128_t, _CharT> {};
-#endif
-
-// Unsigned integral types.
-template <class _CharT>
-struct _LIBCPP_TEMPLATE_VIS formatter<unsigned char, _CharT>
-    : public __format::__formatter_arithmetic<unsigned char, _CharT> {};
-template <class _CharT>
-struct _LIBCPP_TEMPLATE_VIS formatter<unsigned short, _CharT>
-    : public __format::__formatter_arithmetic<unsigned short, _CharT> {};
-template <class _CharT>
-struct _LIBCPP_TEMPLATE_VIS formatter<unsigned, _CharT>
-    : public __format::__formatter_arithmetic<unsigned, _CharT> {};
-template <class _CharT>
-struct _LIBCPP_TEMPLATE_VIS formatter<unsigned long, _CharT>
-    : public __format::__formatter_arithmetic<unsigned long, _CharT> {};
-template <class _CharT>
-struct _LIBCPP_TEMPLATE_VIS formatter<unsigned long long, _CharT>
-    : public __format::__formatter_arithmetic<unsigned long long, _CharT> {};
-#ifndef _LIBCPP_HAS_NO_INT128
-template <class _CharT>
-struct _LIBCPP_TEMPLATE_VIS formatter<__uint128_t, _CharT>
-    : public __format::__formatter_arithmetic<__uint128_t, _CharT> {};
-#endif
-
 // Floating point types.
 template <class _CharT>
 struct _LIBCPP_TEMPLATE_VIS _LIBCPP_AVAILABILITY_FORMAT formatter<float, _CharT>
     : public __format::__formatter_arithmetic<float, _CharT> {};
 template <class _CharT>
 struct _LIBCPP_TEMPLATE_VIS _LIBCPP_AVAILABILITY_FORMAT
     formatter<double, _CharT>
     : public __format::__formatter_arithmetic<double, _CharT> {};

libcxx/test/std/utilities/format/format.functions/locale-specific_form.pass.cpp

+//===----------------------------------------------------------------------===//
+// Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
+// See https://llvm.org/LICENSE.txt for license information.
+// SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
+//
+//===----------------------------------------------------------------------===//
+
+// UNSUPPORTED: c++03, c++11, c++14, c++17
+// UNSUPPORTED: libcpp-no-concepts
+// UNSUPPORTED: libcpp-has-no-localization
+
+// REQUIRES: locale.en_US.UTF-8
+
+// <format>
+
+// This test the locale-specific form for these formatting functions:
+//
+//  // [format.functions], formatting functions
+//  template<class... Args>
+//    string format(string_view fmt, const Args&... args);
+//  template<class... Args>
+//    wstring format(wstring_view fmt, const Args&... args);
+//  template<class... Args>
+//    string format(const locale& loc, string_view fmt, const Args&... args);
+//  template<class... Args>
+//    wstring format(const locale& loc, wstring_view fmt, const Args&... args);
+//
+//  string vformat(string_view fmt, format_args args);
+//  wstring vformat(wstring_view fmt, wformat_args args);
+//  string vformat(const locale& loc, string_view fmt, format_args args);
+//  wstring vformat(const locale& loc, wstring_view fmt, wformat_args args);
+//
+//  template<class Out, class... Args>
+//    Out format_to(Out out, string_view fmt, const Args&... args);
+//  template<class Out, class... Args>
+//    Out format_to(Out out, wstring_view fmt, const Args&... args);
+//  template<class Out, class... Args>
+//    Out format_to(Out out, const locale& loc, string_view fmt, const Args&... args);
+//  template<class Out, class... Args>
+//    Out format_to(Out out, const locale& loc, wstring_view fmt, const Args&... args);
+//
+//  template<class Out>
+//    Out vformat_to(Out out, string_view fmt,
+//                   format_args_t<type_identity_t<Out>, char> args);
+//  template<class Out>
+//    Out vformat_to(Out out, wstring_view fmt,
+//                   format_args_t<type_identity_t<Out>, wchar_t> args);
+//  template<class Out>
+//    Out vformat_to(Out out, const locale& loc, string_view fmt,
+//                   format_args_t<type_identity_t<Out>, char> args);
+//  template<class Out>
+//    Out vformat_to(Out out, const locale& loc, wstring_view fmt,
+//                   format_args_t<type_identity_t<Out>, wchar_t> args);
+//
+//  template<class Out> struct format_to_n_result {
+//    Out out;
+//    iter_difference_t<Out> size;
+//  };
+//
+// template<class Out, class... Args>
+//    format_to_n_result<Out> format_to_n(Out out, iter_difference_t<Out> n,
+//                                        string_view fmt, const Args&... args);
+//  template<class Out, class... Args>
+//    format_to_n_result<Out> format_to_n(Out out, iter_difference_t<Out> n,
+//                                        wstring_view fmt, const Args&... args);
+//  template<class Out, class... Args>
+//    format_to_n_result<Out> format_to_n(Out out, iter_difference_t<Out> n,
+//                                        const locale& loc, string_view fmt,
+//                                        const Args&... args);
+//  template<class Out, class... Args>
+//    format_to_n_result<Out> format_to_n(Out out, iter_difference_t<Out> n,
+//                                        const locale& loc, wstring_view fmt,
+//                                        const Args&... args);
+//
+//  template<class... Args>
+//    size_t formatted_size(string_view fmt, const Args&... args);
+//  template<class... Args>
+//    size_t formatted_size(wstring_view fmt, const Args&... args);
+//  template<class... Args>
+//    size_t formatted_size(const locale& loc, string_view fmt, const Args&... args);
+//  template<class... Args>
+//    size_t formatted_size(const locale& loc, wstring_view fmt, const Args&... args);
+
+#include <format>
+#include <cassert>
+#include <iostream>
+#include <vector>
+
+#include "test_macros.h"
+#include "make_string.h"
+#include "platform_support.h" // locale name macros
+
+#define STR(S) MAKE_STRING(CharT, S)
+
+template <class CharT>
+struct numpunct;
+
+template <>
+struct numpunct<char> : std::numpunct<char> {
+
+  string_type do_truename() const override { return "yes"; }
+  string_type do_falsename() const override { return "no"; }
+
+  std::string do_grouping() const override { return "\1\2\3\2\1"; };
+  char do_thousands_sep() const override { return '_'; }
+};
+
+template <>
+struct numpunct<wchar_t> : std::numpunct<wchar_t> {
+  string_type do_truename() const override { return L"yes"; }
+  string_type do_falsename() const override { return L"no"; }
+
+  std::string do_grouping() const override { return "\1\2\3\2\1"; };
+  wchar_t do_thousands_sep() const override { return L'_'; }
+};
+
+template <class CharT, class... Args>
+void test(std::basic_string<CharT> expected, std::basic_string<CharT> fmt,
+          const Args&... args) {
+  // *** format ***
+  {
+    std::basic_string<CharT> out = std::format(fmt, args...);
+    if constexpr (std::same_as<CharT, char>)
+      if (out != expected)
+        std::cerr << "\nFormat string   " << fmt << "\nExpected output "
+                  << expected << "\nActual output   " << out << '\n';
+    assert(out == expected);
+  }
+  // *** vformat ***
+  {
+    std::basic_string<CharT> out = std::vformat(
+        fmt, std::make_format_args<std::basic_format_context<
+                 std::back_insert_iterator<std::basic_string<CharT>>, CharT>>(
+                 args...));
+    assert(out == expected);
+  }
+  // *** format_to ***
+  {
+    std::basic_string<CharT> out(expected.size(), CharT(' '));
+    auto it = std::format_to(out.begin(), fmt, args...);
+    assert(it == out.end());
+    assert(out == expected);
+  }
+  // *** vformat_to ***
+  {
+    std::basic_string<CharT> out(expected.size(), CharT(' '));
+    auto it = std::vformat_to(
+        out.begin(), fmt,
+        std::make_format_args<std::basic_format_context<
+            typename std::basic_string<CharT>::iterator, CharT>>(args...));
+    assert(it == out.end());
+    assert(out == expected);
+  }
+  // *** format_to_n ***
+  {
+    std::basic_string<CharT> out;
+    std::format_to_n_result result =
+        std::format_to_n(std::back_inserter(out), 1000, fmt, args...);
+    using diff_type = decltype(result.size);
+    diff_type formatted_size = std::formatted_size(fmt, args...);
+    diff_type size = std::min<diff_type>(1000, formatted_size);
+
+    assert(result.size == formatted_size);
+    assert(out == expected.substr(0, size));
+  }
+  // *** formatted_size ***
+  {
+    size_t size = std::formatted_size(fmt, args...);
+    assert(size == expected.size());
+  }
+}
+
+template <class CharT, class... Args>
+void test(std::basic_string<CharT> expected, std::locale loc,
+          std::basic_string<CharT> fmt, const Args&... args) {
+  // *** format ***
+  {
+    std::basic_string<CharT> out = std::format(loc, fmt, args...);
+    if constexpr (std::same_as<CharT, char>)
+      if (out != expected)
+        std::cerr << "\nFormat string   " << fmt << "\nExpected output "
+                  << expected << "\nActual output   " << out << '\n';
+    assert(out == expected);
+  }
+  // *** vformat ***
+  {
+    std::basic_string<CharT> out = std::vformat(
+        loc, fmt,
+        std::make_format_args<std::basic_format_context<
+            std::back_insert_iterator<std::basic_string<CharT>>, CharT>>(
+            args...));
+    assert(out == expected);
+  }
+  // *** format_to ***
+  {
+    std::basic_string<CharT> out(expected.size(), CharT(' '));
+    auto it = std::format_to(out.begin(), loc, fmt, args...);
+    assert(it == out.end());
+    assert(out == expected);
+  }
+  // *** vformat_to ***
+  {
+    std::basic_string<CharT> out(expected.size(), CharT(' '));
+    auto it = std::vformat_to(
+        out.begin(), loc, fmt,
+        std::make_format_args<std::basic_format_context<
+            typename std::basic_string<CharT>::iterator, CharT>>(args...));
+    assert(it == out.end());
+    assert(out == expected);
+  }
+  // *** format_to_n ***
+  {
+    std::basic_string<CharT> out;
+    std::format_to_n_result result =
+        std::format_to_n(std::back_inserter(out), 1000, loc, fmt, args...);
+    using diff_type = decltype(result.size);
+    diff_type formatted_size = std::formatted_size(loc, fmt, args...);
+    diff_type size = std::min<diff_type>(1000, formatted_size);
+
+    assert(result.size == formatted_size);
+    assert(out == expected.substr(0, size));
+  }
+  // *** formatted_size ***
+  {
+    size_t size = std::formatted_size(loc, fmt, args...);
+    assert(size == expected.size());
+  }
+}
+
+template <class CharT>
+void test_integer() {
+  std::locale loc = std::locale(std::locale(), new numpunct<CharT>());
+  std::locale en_US = std::locale(LOCALE_en_US_UTF_8);
+
+  // *** Decimal ***
+  std::locale::global(en_US);
+  test(STR("0"), STR("{:L}"), 0);
+  test(STR("1"), STR("{:L}"), 1);
+  test(STR("10"), STR("{:L}"), 10);
+  test(STR("100"), STR("{:L}"), 100);
+  test(STR("1,000"), STR("{:L}"), 1'000);
+  test(STR("10,000"), STR("{:L}"), 10'000);
+  test(STR("100,000"), STR("{:L}"), 100'000);
+  test(STR("1,000,000"), STR("{:L}"), 1'000'000);
+  test(STR("10,000,000"), STR("{:L}"), 10'000'000);
+  test(STR("100,000,000"), STR("{:L}"), 100'000'000);
+  test(STR("1,000,000,000"), STR("{:L}"), 1'000'000'000);
+
+  test(STR("-1"), STR("{:L}"), -1);
+  test(STR("-10"), STR("{:L}"), -10);
+  test(STR("-100"), STR("{:L}"), -100);
+  test(STR("-1,000"), STR("{:L}"), -1'000);
+  test(STR("-10,000"), STR("{:L}"), -10'000);
+  test(STR("-100,000"), STR("{:L}"), -100'000);
+  test(STR("-1,000,000"), STR("{:L}"), -1'000'000);
+  test(STR("-10,000,000"), STR("{:L}"), -10'000'000);
+  test(STR("-100,000,000"), STR("{:L}"), -100'000'000);
+  test(STR("-1,000,000,000"), STR("{:L}"), -1'000'000'000);
+
+  std::locale::global(loc);
+  test(STR("0"), STR("{:L}"), 0);
+  test(STR("1"), STR("{:L}"), 1);
+  test(STR("1_0"), STR("{:L}"), 10);
+  test(STR("10_0"), STR("{:L}"), 100);
+  test(STR("1_00_0"), STR("{:L}"), 1'000);
+  test(STR("10_00_0"), STR("{:L}"), 10'000);
+  test(STR("100_00_0"), STR("{:L}"), 100'000);
+  test(STR("1_000_00_0"), STR("{:L}"), 1'000'000);
+  test(STR("10_000_00_0"), STR("{:L}"), 10'000'000);
+  test(STR("1_00_000_00_0"), STR("{:L}"), 100'000'000);
+  test(STR("1_0_00_000_00_0"), STR("{:L}"), 1'000'000'000);
+
+  test(STR("-1"), STR("{:L}"), -1);
+  test(STR("-1_0"), STR("{:L}"), -10);
+  test(STR("-10_0"), STR("{:L}"), -100);
+  test(STR("-1_00_0"), STR("{:L}"), -1'000);
+  test(STR("-10_00_0"), STR("{:L}"), -10'000);
+  test(STR("-100_00_0"), STR("{:L}"), -100'000);
+  test(STR("-1_000_00_0"), STR("{:L}"), -1'000'000);
+  test(STR("-10_000_00_0"), STR("{:L}"), -10'000'000);
+  test(STR("-1_00_000_00_0"), STR("{:L}"), -100'000'000);
+  test(STR("-1_0_00_000_00_0"), STR("{:L}"), -1'000'000'000);
+
+  test(STR("0"), en_US, STR("{:L}"), 0);
+  test(STR("1"), en_US, STR("{:L}"), 1);
+  test(STR("10"), en_US, STR("{:L}"), 10);
+  test(STR("100"), en_US, STR("{:L}"), 100);
+  test(STR("1,000"), en_US, STR("{:L}"), 1'000);
+  test(STR("10,000"), en_US, STR("{:L}"), 10'000);
+  test(STR("100,000"), en_US, STR("{:L}"), 100'000);
+  test(STR("1,000,000"), en_US, STR("{:L}"), 1'000'000);
+  test(STR("10,000,000"), en_US, STR("{:L}"), 10'000'000);
+  test(STR("100,000,000"), en_US, STR("{:L}"), 100'000'000);
+  test(STR("1,000,000,000"), en_US, STR("{:L}"), 1'000'000'000);
+
+  test(STR("-1"), en_US, STR("{:L}"), -1);
+  test(STR("-10"), en_US, STR("{:L}"), -10);
+  test(STR("-100"), en_US, STR("{:L}"), -100);
+  test(STR("-1,000"), en_US, STR("{:L}"), -1'000);
+  test(STR("-10,000"), en_US, STR("{:L}"), -10'000);
+  test(STR("-100,000"), en_US, STR("{:L}"), -100'000);
+  test(STR("-1,000,000"), en_US, STR("{:L}"), -1'000'000);
+  test(STR("-10,000,000"), en_US, STR("{:L}"), -10'000'000);
+  test(STR("-100,000,000"), en_US, STR("{:L}"), -100'000'000);
+  test(STR("-1,000,000,000"), en_US, STR("{:L}"), -1'000'000'000);
+
+  std::locale::global(en_US);
+  test(STR("0"), loc, STR("{:L}"), 0);
+  test(STR("1"), loc, STR("{:L}"), 1);
+  test(STR("1_0"), loc, STR("{:L}"), 10);
+  test(STR("10_0"), loc, STR("{:L}"), 100);
+  test(STR("1_00_0"), loc, STR("{:L}"), 1'000);
+  test(STR("10_00_0"), loc, STR("{:L}"), 10'000);
+  test(STR("100_00_0"), loc, STR("{:L}"), 100'000);
+  test(STR("1_000_00_0"), loc, STR("{:L}"), 1'000'000);
+  test(STR("10_000_00_0"), loc, STR("{:L}"), 10'000'000);
+  test(STR("1_00_000_00_0"), loc, STR("{:L}"), 100'000'000);
+  test(STR("1_0_00_000_00_0"), loc, STR("{:L}"), 1'000'000'000);
+
+  test(STR("-1"), loc, STR("{:L}"), -1);
+  test(STR("-1_0"), loc, STR("{:L}"), -10);
+  test(STR("-10_0"), loc, STR("{:L}"), -100);
+  test(STR("-1_00_0"), loc, STR("{:L}"), -1'000);
+  test(STR("-10_00_0"), loc, STR("{:L}"), -10'000);
+  test(STR("-100_00_0"), loc, STR("{:L}"), -100'000);
+  test(STR("-1_000_00_0"), loc, STR("{:L}"), -1'000'000);
+  test(STR("-10_000_00_0"), loc, STR("{:L}"), -10'000'000);
+  test(STR("-1_00_000_00_0"), loc, STR("{:L}"), -100'000'000);
+  test(STR("-1_0_00_000_00_0"), loc, STR("{:L}"), -1'000'000'000);
+
+  // *** Binary ***
+  std::locale::global(en_US);
+  test(STR("0"), STR("{:Lb}"), 0b0);
+  test(STR("1"), STR("{:Lb}"), 0b1);
+  test(STR("1,000,000,000"), STR("{:Lb}"), 0b1'000'000'000);
+
+  test(STR("0b0"), STR("{:#Lb}"), 0b0);
+  test(STR("0b1"), STR("{:#Lb}"), 0b1);
+  test(STR("0b1,000,000,000"), STR("{:#Lb}"), 0b1'000'000'000);
+
+  test(STR("-1"), STR("{:LB}"), -0b1);
+  test(STR("-1,000,000,000"), STR("{:LB}"), -0b1'000'000'000);
+
+  test(STR("-0B1"), STR("{:#LB}"), -0b1);
+  test(STR("-0B1,000,000,000"), STR("{:#LB}"), -0b1'000'000'000);
+
+  std::locale::global(loc);
+  test(STR("0"), STR("{:Lb}"), 0b0);
+  test(STR("1"), STR("{:Lb}"), 0b1);
+  test(STR("1_0_00_000_00_0"), STR("{:Lb}"), 0b1'000'000'000);
+
+  test(STR("0b0"), STR("{:#Lb}"), 0b0);
+  test(STR("0b1"), STR("{:#Lb}"), 0b1);
+  test(STR("0b1_0_00_000_00_0"), STR("{:#Lb}"), 0b1'000'000'000);
+
+  test(STR("-1"), STR("{:LB}"), -0b1);
+  test(STR("-1_0_00_000_00_0"), STR("{:LB}"), -0b1'000'000'000);
+
+  test(STR("-0B1"), STR("{:#LB}"), -0b1);
+  test(STR("-0B1_0_00_000_00_0"), STR("{:#LB}"), -0b1'000'000'000);
+
+  test(STR("0"), en_US, STR("{:Lb}"), 0b0);
+  test(STR("1"), en_US, STR("{:Lb}"), 0b1);
+  test(STR("1,000,000,000"), en_US, STR("{:Lb}"), 0b1'000'000'000);
+
+  test(STR("0b0"), en_US, STR("{:#Lb}"), 0b0);
+  test(STR("0b1"), en_US, STR("{:#Lb}"), 0b1);
+  test(STR("0b1,000,000,000"), en_US, STR("{:#Lb}"), 0b1'000'000'000);
+
+  test(STR("-1"), en_US, STR("{:LB}"), -0b1);
+  test(STR("-1,000,000,000"), en_US, STR("{:LB}"), -0b1'000'000'000);
+
+  test(STR("-0B1"), en_US, STR("{:#LB}"), -0b1);
+  test(STR("-0B1,000,000,000"), en_US, STR("{:#LB}"), -0b1'000'000'000);
+
+  std::locale::global(en_US);
+  test(STR("0"), loc, STR("{:Lb}"), 0b0);
+  test(STR("1"), loc, STR("{:Lb}"), 0b1);
+  test(STR("1_0_00_000_00_0"), loc, STR("{:Lb}"), 0b1'000'000'000);
+
+  test(STR("0b0"), loc, STR("{:#Lb}"), 0b0);
+  test(STR("0b1"), loc, STR("{:#Lb}"), 0b1);
+  test(STR("0b1_0_00_000_00_0"), loc, STR("{:#Lb}"), 0b1'000'000'000);
+
+  test(STR("-1"), loc, STR("{:LB}"), -0b1);
+  test(STR("-1_0_00_000_00_0"), loc, STR("{:LB}"), -0b1'000'000'000);
+
+  test(STR("-0B1"), loc, STR("{:#LB}"), -0b1);
+  test(STR("-0B1_0_00_000_00_0"), loc, STR("{:#LB}"), -0b1'000'000'000);
+
+  // *** Octal ***
+  std::locale::global(en_US);
+  test(STR("0"), STR("{:Lo}"), 00);
+  test(STR("1"), STR("{:Lo}"), 01);
+  test(STR("1,000,000,000"), STR("{:Lo}"), 01'000'000'000);
+
+  test(STR("0"), STR("{:#Lo}"), 00);
+  test(STR("01"), STR("{:#Lo}"), 01);
+  test(STR("01,000,000,000"), STR("{:#Lo}"), 01'000'000'000);
+
+  test(STR("-1"), STR("{:Lo}"), -01);
+  test(STR("-1,000,000,000"), STR("{:Lo}"), -01'000'000'000);
+
+  test(STR("-01"), STR("{:#Lo}"), -01);
+  test(STR("-01,000,000,000"), STR("{:#Lo}"), -01'000'000'000);
+
+  std::locale::global(loc);
+  test(STR("0"), STR("{:Lo}"), 00);
+  test(STR("1"), STR("{:Lo}"), 01);
+  test(STR("1_0_00_000_00_0"), STR("{:Lo}"), 01'000'000'000);
+
+  test(STR("0"), STR("{:#Lo}"), 00);
+  test(STR("01"), STR("{:#Lo}"), 01);
+  test(STR("01_0_00_000_00_0"), STR("{:#Lo}"), 01'000'000'000);
+
+  test(STR("-1"), STR("{:Lo}"), -01);
+  test(STR("-1_0_00_000_00_0"), STR("{:Lo}"), -01'000'000'000);
+
+  test(STR("-01"), STR("{:#Lo}"), -01);
+  test(STR("-01_0_00_000_00_0"), STR("{:#Lo}"), -01'000'000'000);
+
+  test(STR("0"), en_US, STR("{:Lo}"), 00);
+  test(STR("1"), en_US, STR("{:Lo}"), 01);
+  test(STR("1,000,000,000"), en_US, STR("{:Lo}"), 01'000'000'000);
+
+  test(STR("0"), en_US, STR("{:#Lo}"), 00);
+  test(STR("01"), en_US, STR("{:#Lo}"), 01);
+  test(STR("01,000,000,000"), en_US, STR("{:#Lo}"), 01'000'000'000);
+
+  test(STR("-1"), en_US, STR("{:Lo}"), -01);
+  test(STR("-1,000,000,000"), en_US, STR("{:Lo}"), -01'000'000'000);
+
+  test(STR("-01"), en_US, STR("{:#Lo}"), -01);
+  test(STR("-01,000,000,000"), en_US, STR("{:#Lo}"), -01'000'000'000);
+
+  std::locale::global(en_US);
+  test(STR("0"), loc, STR("{:Lo}"), 00);
+  test(STR("1"), loc, STR("{:Lo}"), 01);
+  test(STR("1_0_00_000_00_0"), loc, STR("{:Lo}"), 01'000'000'000);
+
+  test(STR("0"), loc, STR("{:#Lo}"), 00);
+  test(STR("01"), loc, STR("{:#Lo}"), 01);
+  test(STR("01_0_00_000_00_0"), loc, STR("{:#Lo}"), 01'000'000'000);
+
+  test(STR("-1"), loc, STR("{:Lo}"), -01);
+  test(STR("-1_0_00_000_00_0"), loc, STR("{:Lo}"), -01'000'000'000);
+
+  test(STR("-01"), loc, STR("{:#Lo}"), -01);
+  test(STR("-01_0_00_000_00_0"), loc, STR("{:#Lo}"), -01'000'000'000);
+
+  // *** Hexadecimal ***
+  std::locale::global(en_US);
+  test(STR("0"), STR("{:Lx}"), 0x0);
+  test(STR("1"), STR("{:Lx}"), 0x1);
+  test(STR("1,000,000,000"), STR("{:Lx}"), 0x1'000'000'000);
+
+  test(STR("0x0"), STR("{:#Lx}"), 0x0);
+  test(STR("0x1"), STR("{:#Lx}"), 0x1);
+  test(STR("0x1,000,000,000"), STR("{:#Lx}"), 0x1'000'000'000);
+
+  test(STR("-1"), STR("{:LX}"), -0x1);
+  test(STR("-1,000,000,000"), STR("{:LX}"), -0x1'000'000'000);
+
+  test(STR("-0X1"), STR("{:#LX}"), -0x1);
+  test(STR("-0X1,000,000,000"), STR("{:#LX}"), -0x1'000'000'000);
+
+  std::locale::global(loc);
+  test(STR("0"), STR("{:Lx}"), 0x0);
+  test(STR("1"), STR("{:Lx}"), 0x1);
+  test(STR("1_0_00_000_00_0"), STR("{:Lx}"), 0x1'000'000'000);
+
+  test(STR("0x0"), STR("{:#Lx}"), 0x0);
+  test(STR("0x1"), STR("{:#Lx}"), 0x1);
+  test(STR("0x1_0_00_000_00_0"), STR("{:#Lx}"), 0x1'000'000'000);
+
+  test(STR("-1"), STR("{:LX}"), -0x1);
+  test(STR("-1_0_00_000_00_0"), STR("{:LX}"), -0x1'000'000'000);
+
+  test(STR("-0X1"), STR("{:#LX}"), -0x1);
+  test(STR("-0X1_0_00_000_00_0"), STR("{:#LX}"), -0x1'000'000'000);
+
+  test(STR("0"), en_US, STR("{:Lx}"), 0x0);
+  test(STR("1"), en_US, STR("{:Lx}"), 0x1);
+  test(STR("1,000,000,000"), en_US, STR("{:Lx}"), 0x1'000'000'000);
+
+  test(STR("0x0"), en_US, STR("{:#Lx}"), 0x0);
+  test(STR("0x1"), en_US, STR("{:#Lx}"), 0x1);
+  test(STR("0x1,000,000,000"), en_US, STR("{:#Lx}"), 0x1'000'000'000);
+
+  test(STR("-1"), en_US, STR("{:LX}"), -0x1);
+  test(STR("-1,000,000,000"), en_US, STR("{:LX}"), -0x1'000'000'000);
+
+  test(STR("-0X1"), en_US, STR("{:#LX}"), -0x1);
+  test(STR("-0X1,000,000,000"), en_US, STR("{:#LX}"), -0x1'000'000'000);
+
+  std::locale::global(en_US);
+  test(STR("0"), loc, STR("{:Lx}"), 0x0);
+  test(STR("1"), loc, STR("{:Lx}"), 0x1);
+  test(STR("1_0_00_000_00_0"), loc, STR("{:Lx}"), 0x1'000'000'000);
+
+  test(STR("0x0"), loc, STR("{:#Lx}"), 0x0);
+  test(STR("0x1"), loc, STR("{:#Lx}"), 0x1);
+  test(STR("0x1_0_00_000_00_0"), loc, STR("{:#Lx}"), 0x1'000'000'000);
+
+  test(STR("-1"), loc, STR("{:LX}"), -0x1);
+  test(STR("-1_0_00_000_00_0"), loc, STR("{:LX}"), -0x1'000'000'000);
+
+  test(STR("-0X1"), loc, STR("{:#LX}"), -0x1);
+  test(STR("-0X1_0_00_000_00_0"), loc, STR("{:#LX}"), -0x1'000'000'000);
+
+  // *** align-fill & width ***
+  test(STR("4_2"), loc, STR("{:L}"), 42);
+
+  test(STR("   4_2"), loc, STR("{:6L}"), 42);
+  test(STR("4_2   "), loc, STR("{:<6L}"), 42);
+  test(STR(" 4_2  "), loc, STR("{:^6L}"), 42);
+  test(STR("   4_2"), loc, STR("{:>6L}"), 42);
+
+  test(STR("4_2***"), loc, STR("{:*<6L}"), 42);
+  test(STR("*4_2**"), loc, STR("{:*^6L}"), 42);
+  test(STR("***4_2"), loc, STR("{:*>6L}"), 42);
+
+  test(STR("4_a*****"), loc, STR("{:*<8Lx}"), 0x4a);
+  test(STR("**4_a***"), loc, STR("{:*^8Lx}"), 0x4a);
+  test(STR("*****4_a"), loc, STR("{:*>8Lx}"), 0x4a);
+
+  test(STR("0x4_a***"), loc, STR("{:*<#8Lx}"), 0x4a);
+  test(STR("*0x4_a**"), loc, STR("{:*^#8Lx}"), 0x4a);
+  test(STR("***0x4_a"), loc, STR("{:*>#8Lx}"), 0x4a);
+
+  test(STR("4_A*****"), loc, STR("{:*<8LX}"), 0x4a);
+  test(STR("**4_A***"), loc, STR("{:*^8LX}"), 0x4a);
+  test(STR("*****4_A"), loc, STR("{:*>8LX}"), 0x4a);
+
+  test(STR("0X4_A***"), loc, STR("{:*<#8LX}"), 0x4a);
+  test(STR("*0X4_A**"), loc, STR("{:*^#8LX}"), 0x4a);
+  test(STR("***0X4_A"), loc, STR("{:*>#8LX}"), 0x4a);
+
+  // Test whether zero padding is ignored
+  test(STR("4_2   "), loc, STR("{:<06L}"), 42);
+  test(STR(" 4_2  "), loc, STR("{:^06L}"), 42);
+  test(STR("   4_2"), loc, STR("{:>06L}"), 42);
+
+  // *** zero-padding & width ***
+  test(STR("   4_2"), loc, STR("{:6L}"), 42);
+  test(STR("0004_2"), loc, STR("{:06L}"), 42);
+  test(STR("-004_2"), loc, STR("{:06L}"), -42);
+
+  test(STR("000004_a"), loc, STR("{:08Lx}"), 0x4a);
+  test(STR("0x0004_a"), loc, STR("{:#08Lx}"), 0x4a);
+  test(STR("0X0004_A"), loc, STR("{:#08LX}"), 0x4a);
+
+  test(STR("-00004_a"), loc, STR("{:08Lx}"), -0x4a);
+  test(STR("-0x004_a"), loc, STR("{:#08Lx}"), -0x4a);
+  test(STR("-0X004_A"), loc, STR("{:#08LX}"), -0x4a);
+}
+
+template <class CharT>
+void test() {
+  test_integer<CharT>();
+}
+
+int main(int, char**) {
+  test<char>();
+  test<wchar_t>();
+
+  return 0;
+}

libcxx/test/std/utilities/format/format.functions/tests.inc

 //===----------------------------------------------------------------------===//
 // Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
 // See https://llvm.org/LICENSE.txt for license information.
 // SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
 //
 //===----------------------------------------------------------------------===//
 
-// Contains the test cases for the format functions without locale.
+// Contains the test cases for the format functions. The locale-specific form
+// aren't tested in these test cases. They're tested in
+// locale-specific_form.pass.cpp.
 
 #define STR(S) MAKE_STRING(CharT, S)
 
 template <class CharT>
 std::vector<std::basic_string<CharT>> invalid_types(std::string valid) {
   std::vector<std::basic_string<CharT>> result;
 
 #define CASE(T)                                                                \
void test_string() {
   test_string<CharT>(const_cast<CharT*>(world.c_str()),
                      const_cast<CharT*>(universe.c_str()));
   test_string<CharT>(std::basic_string_view<CharT>(world),
                      std::basic_string_view<CharT>(universe));
   test_string<CharT>(world, universe);
   test_string_unicode<CharT>();
 }
 
+template <class I, class CharT>
+void test_integer_as_integer() {
+  // *** align-fill & width ***
+  test(STR("answer is '42'"), STR("answer is '{:<1}'"), I(42));
+  test(STR("answer is '42'"), STR("answer is '{:<2}'"), I(42));
+  test(STR("answer is '42 '"), STR("answer is '{:<3}'"), I(42));
+
+  test(STR("answer is '     42'"), STR("answer is '{:7}'"), I(42));
+  test(STR("answer is '     42'"), STR("answer is '{:>7}'"), I(42));
+  test(STR("answer is '42     '"), STR("answer is '{:<7}'"), I(42));
+  test(STR("answer is '  42   '"), STR("answer is '{:^7}'"), I(42));
+
+  test(STR("answer is '*****42'"), STR("answer is '{:*>7}'"), I(42));
+  test(STR("answer is '42*****'"), STR("answer is '{:*<7}'"), I(42));
+  test(STR("answer is '**42***'"), STR("answer is '{:*^7}'"), I(42));
+
+  // Test whether zero padding is ignored
+  test(STR("answer is '     42'"), STR("answer is '{:>07}'"), I(42));
+  test(STR("answer is '42     '"), STR("answer is '{:<07}'"), I(42));
+  test(STR("answer is '  42   '"), STR("answer is '{:^07}'"), I(42));
+
+  // *** Sign ***
+  if constexpr (std::signed_integral<I>)
+    test(STR("answer is -42"), STR("answer is {}"), I(-42));
+  test(STR("answer is 0"), STR("answer is {}"), I(0));
+  test(STR("answer is 42"), STR("answer is {}"), I(42));
+
+  if constexpr (std::signed_integral<I>)
+    test(STR("answer is -42"), STR("answer is {:-}"), I(-42));
+  test(STR("answer is 0"), STR("answer is {:-}"), I(0));
+  test(STR("answer is 42"), STR("answer is {:-}"), I(42));
+
+  if constexpr (std::signed_integral<I>)
+    test(STR("answer is -42"), STR("answer is {:+}"), I(-42));
+  test(STR("answer is +0"), STR("answer is {:+}"), I(0));
+  test(STR("answer is +42"), STR("answer is {:+}"), I(42));
+
+  if constexpr (std::signed_integral<I>)
+    test(STR("answer is -42"), STR("answer is {: }"), I(-42));
+  test(STR("answer is  0"), STR("answer is {: }"), I(0));
+  test(STR("answer is  42"), STR("answer is {: }"), I(42));
+
+  // *** alternate form ***
+  if constexpr (std::signed_integral<I>) {
+    test(STR("answer is -42"), STR("answer is {:#}"), I(-42));
+    test(STR("answer is -42"), STR("answer is {:#d}"), I(-42));
+    test(STR("answer is -101010"), STR("answer is {:b}"), I(-42));
+    test(STR("answer is -0b101010"), STR("answer is {:#b}"), I(-42));
+    test(STR("answer is -0B101010"), STR("answer is {:#B}"), I(-42));
+    test(STR("answer is -52"), STR("answer is {:o}"), I(-42));
+    test(STR("answer is -052"), STR("answer is {:#o}"), I(-42));
+    test(STR("answer is -2a"), STR("answer is {:x}"), I(-42));
+    test(STR("answer is -0x2a"), STR("answer is {:#x}"), I(-42));
+    test(STR("answer is -2A"), STR("answer is {:X}"), I(-42));
+    test(STR("answer is -0X2A"), STR("answer is {:#X}"), I(-42));
+  }
+  test(STR("answer is 0"), STR("answer is {:#}"), I(0));
+  test(STR("answer is 0"), STR("answer is {:#d}"), I(0));
+  test(STR("answer is 0"), STR("answer is {:b}"), I(0));
+  test(STR("answer is 0b0"), STR("answer is {:#b}"), I(0));
+  test(STR("answer is 0B0"), STR("answer is {:#B}"), I(0));
+  test(STR("answer is 0"), STR("answer is {:o}"), I(0));
+  test(STR("answer is 0"), STR("answer is {:#o}"), I(0));
+  test(STR("answer is 0"), STR("answer is {:x}"), I(0));
+  test(STR("answer is 0x0"), STR("answer is {:#x}"), I(0));
+  test(STR("answer is 0"), STR("answer is {:X}"), I(0));
+  test(STR("answer is 0X0"), STR("answer is {:#X}"), I(0));
+
+  test(STR("answer is +42"), STR("answer is {:+#}"), I(42));
+  test(STR("answer is +42"), STR("answer is {:+#d}"), I(42));
+  test(STR("answer is +101010"), STR("answer is {:+b}"), I(42));
+  test(STR("answer is +0b101010"), STR("answer is {:+#b}"), I(42));
+  test(STR("answer is +0B101010"), STR("answer is {:+#B}"), I(42));
+  test(STR("answer is +52"), STR("answer is {:+o}"), I(42));
+  test(STR("answer is +052"), STR("answer is {:+#o}"), I(42));
+  test(STR("answer is +2a"), STR("answer is {:+x}"), I(42));
+  test(STR("answer is +0x2a"), STR("answer is {:+#x}"), I(42));
+  test(STR("answer is +2A"), STR("answer is {:+X}"), I(42));
+  test(STR("answer is +0X2A"), STR("answer is {:+#X}"), I(42));
+
+  // *** zero-padding & width ***
+  if constexpr (std::signed_integral<I>) {
+    test(STR("answer is -00000000042"), STR("answer is {:#012}"), I(-42));
+    test(STR("answer is -00000000042"), STR("answer is {:#012d}"), I(-42));
+    test(STR("answer is -00000101010"), STR("answer is {:012b}"), I(-42));
+    test(STR("answer is -0b000101010"), STR("answer is {:#012b}"), I(-42));
+    test(STR("answer is -0B000101010"), STR("answer is {:#012B}"), I(-42));
+    test(STR("answer is -00000000052"), STR("answer is {:012o}"), I(-42));
+    test(STR("answer is -00000000052"), STR("answer is {:#012o}"), I(-42));
+    test(STR("answer is -0000000002a"), STR("answer is {:012x}"), I(-42));
+    test(STR("answer is -0x00000002a"), STR("answer is {:#012x}"), I(-42));
+    test(STR("answer is -0000000002A"), STR("answer is {:012X}"), I(-42));
+    test(STR("answer is -0X00000002A"), STR("answer is {:#012X}"), I(-42));
+  }
+
+  test(STR("answer is 000000000000"), STR("answer is {:#012}"), I(0));
+  test(STR("answer is 000000000000"), STR("answer is {:#012d}"), I(0));
+  test(STR("answer is 000000000000"), STR("answer is {:012b}"), I(0));
+  test(STR("answer is 0b0000000000"), STR("answer is {:#012b}"), I(0));
+  test(STR("answer is 0B0000000000"), STR("answer is {:#012B}"), I(0));
+  test(STR("answer is 000000000000"), STR("answer is {:012o}"), I(0));
+  test(STR("answer is 000000000000"), STR("answer is {:#012o}"), I(0));
+  test(STR("answer is 000000000000"), STR("answer is {:012x}"), I(0));
+  test(STR("answer is 0x0000000000"), STR("answer is {:#012x}"), I(0));
+  test(STR("answer is 000000000000"), STR("answer is {:012X}"), I(0));
+  test(STR("answer is 0X0000000000"), STR("answer is {:#012X}"), I(0));
+
+  test(STR("answer is +00000000042"), STR("answer is {:+#012}"), I(42));
+  test(STR("answer is +00000000042"), STR("answer is {:+#012d}"), I(42));
+  test(STR("answer is +00000101010"), STR("answer is {:+012b}"), I(42));
+  test(STR("answer is +0b000101010"), STR("answer is {:+#012b}"), I(42));
+  test(STR("answer is +0B000101010"), STR("answer is {:+#012B}"), I(42));
+  test(STR("answer is +00000000052"), STR("answer is {:+012o}"), I(42));
+  test(STR("answer is +00000000052"), STR("answer is {:+#012o}"), I(42));
+  test(STR("answer is +0000000002a"), STR("answer is {:+012x}"), I(42));
+  test(STR("answer is +0x00000002a"), STR("answer is {:+#012x}"), I(42));
+  test(STR("answer is +0000000002A"), STR("answer is {:+012X}"), I(42));
+  test(STR("answer is +0X00000002A"), STR("answer is {:+#012X}"), I(42));
+
+  // *** precision ***
+  test_exception("Format-spec precision not allowed", STR("{:.}"), I(0));
+  test_exception("Format-spec precision not allowed", STR("{:.0}"), I(0));
+  test_exception("Format-spec precision not allowed", STR("{:.42}"), I(0));
+
+  // *** locale-specific form ***
+  // See locale-specific_form.pass.cpp
+
+  // *** type ***
+  for (const auto& fmt : invalid_types<CharT>("bBcdoxX"))
+    test_exception(
+        "Format-spec type has a type not supported for an integer argument",
+        fmt, 42);
+}
+
+template <class I, class CharT>
+void test_integer_as_char() {
+  // *** align-fill & width ***
+  // TODO FMT Validate whether this should be left aligned
+  test(STR("answer is '     *'"), STR("answer is '{:6c}'"), I(42));
+  test(STR("answer is '     *'"), STR("answer is '{:>6c}'"), I(42));
+  test(STR("answer is '*     '"), STR("answer is '{:<6c}'"), I(42));
+  test(STR("answer is '  *   '"), STR("answer is '{:^6c}'"), I(42));
+
+  test(STR("answer is '-----*'"), STR("answer is '{:->6c}'"), I(42));
+  test(STR("answer is '*-----'"), STR("answer is '{:-<6c}'"), I(42));
+  test(STR("answer is '--*---'"), STR("answer is '{:-^6c}'"), I(42));
+
+  // Test whether zero padding is ignored
+  test(STR("answer is '     *'"), STR("answer is '{:>06c}'"), I(42));
+  test(STR("answer is '*     '"), STR("answer is '{:<06c}'"), I(42));
+  test(STR("answer is '  *   '"), STR("answer is '{:^06c}'"), I(42));
+
+  // *** Sign ***
+  test(STR("answer is *"), STR("answer is {:c}"), I(42));
+  test(STR("answer is *"), STR("answer is {:-c}"), I(42));
+  test(STR("answer is *"), STR("answer is {:+c}"), I(42));
+  test(STR("answer is *"), STR("answer is {: c}"), I(42));
+
+  // *** alternate form ***
+  test(STR("answer is *"), STR("answer is {:#c}"), I(42));
+  test(STR("answer is *"), STR("answer is {:-#c}"), I(42));
+  test(STR("answer is *"), STR("answer is {:+#c}"), I(42));
+  test(STR("answer is *"), STR("answer is {: #c}"), I(42));
+
+  // *** zero-padding & width ***
+  test(STR("answer is '           *'"), STR("answer is '{:+#012c}'"), I(42));
+
+  // *** precision ***
+  test_exception("Format-spec precision not allowed", STR("{:.c}"), I(0));
+  test_exception("Format-spec precision not allowed", STR("{:.0c}"), I(0));
+  test_exception("Format-spec precision not allowed", STR("{:.42c}"), I(0));
+
+  // *** locale-specific form ***
+  // Note it has no effect but it's allowed.
+  test(STR("answer is '*'"), STR("answer is '{:Lc}'"), I(42));
+
+  // *** type ***
+  for (const auto& fmt : invalid_types<CharT>("bBcdoxX"))
+    test_exception(
+        "Format-spec type has a type not supported for an integer argument",
+        fmt, I(42));
+
+  // *** Validate range ***
+  // TODO FMT Update test after adding 128-bit support.
+  if constexpr (sizeof(I) <= sizeof(long long)) {
+    // The code has some duplications to keep the if statement readable.
+    if constexpr (std::signed_integral<CharT>) {
+      if constexpr (std::signed_integral<I> && sizeof(I) > sizeof(CharT)) {
+        test_exception("Integral value outside the range of the char type.",
+                       STR("{:c}"), std::numeric_limits<I>::min());
+        test_exception("Integral value outside the range of the char type.",
+                       STR("{:c}"), std::numeric_limits<I>::max());
+      } else if constexpr (std::unsigned_integral<I> &&
+                           sizeof(I) >= sizeof(CharT)) {
+        test_exception("Integral value outside the range of the char type.",
+                       STR("{:c}"), std::numeric_limits<I>::max());
+      }
+    } else if constexpr (sizeof(I) > sizeof(CharT)) {
+      test_exception("Integral value outside the range of the char type.",
+                     STR("{:c}"), std::numeric_limits<I>::max());
+    }
+  }
+}
+
+template <class I, class CharT>
+void test_integer() {
+  test_integer_as_integer<I, CharT>();
+  test_integer_as_char<I, CharT>();
+}
+
+template <class CharT>
+void test_signed_integer() {
+  test_integer<signed char, CharT>();
+  test_integer<short, CharT>();
+  test_integer<int, CharT>();
+  test_integer<long, CharT>();
+  test_integer<long long, CharT>();
+#ifndef _LIBCPP_HAS_NO_INT128
+  test_integer<__int128_t, CharT>();
+#endif
+}
+
+template <class CharT>
+void test_unsigned_integer() {
+  test_integer<unsigned char, CharT>();
+  test_integer<unsigned short, CharT>();
+  test_integer<unsigned, CharT>();
+  test_integer<unsigned long, CharT>();
+  test_integer<unsigned long long, CharT>();
+#ifndef _LIBCPP_HAS_NO_INT128
+  test_integer<__uint128_t, CharT>();
+#endif
+}
+
 template <class CharT>
 void test() {
   // *** Test escaping  ***
   test(STR("{"), STR("{{"));
   test(STR("}"), STR("}}"));
 
   // *** Test argument ID ***
   test(STR("hello 01"), STR("hello {0:}{1:}"), false, true);
test(STR("hello 42"), STR("hello {}"), static_cast<__int128_t>(42));
     test_exception(
         "128 bit value is outside of implemented range", STR("{}"),
         static_cast<__int128_t>(std::numeric_limits<long long>::min()) - 1);
     test_exception(
         "128 bit value is outside of implemented range", STR("{}"),
         static_cast<__int128_t>(std::numeric_limits<long long>::max()) + 1);
   }
 #endif
+  test_signed_integer<CharT>();
 
   // ** Test unsigned integral format argument ***
   test(STR("hello 42"), STR("hello {}"), static_cast<unsigned char>(42));
   test(STR("hello 42"), STR("hello {}"), static_cast<unsigned short>(42));
   test(STR("hello 42"), STR("hello {}"), static_cast<unsigned>(42));
   test(STR("hello 42"), STR("hello {}"), static_cast<unsigned long>(42));
   test(STR("hello 42"), STR("hello {}"), static_cast<unsigned long long>(42));
 #ifndef _LIBCPP_HAS_NO_INT128
   test(STR("hello 42"), STR("hello {}"), static_cast<__uint128_t>(42));
   {
     // Note 128 bit support is only party implemented test the range
     // conditions here.
     std::basic_string<CharT> max =
         std::format(STR("{}"), std::numeric_limits<unsigned long long>::max());
     test(max, STR("{}"),
          static_cast<__uint128_t>(
              std::numeric_limits<unsigned long long>::max()));
     test_exception("128 bit value is outside of implemented range", STR("{}"),
                    static_cast<__uint128_t>(
                        std::numeric_limits<unsigned long long>::max()) +
                        1);
   }
 #endif
+  test_unsigned_integer<CharT>();
 
   // *** Test floating point format argument ***
   test(STR("hello 42.000000"), STR("hello {}"), static_cast<float>(42));
   test(STR("hello 42.000000"), STR("hello {}"), static_cast<double>(42));
   test(STR("hello 42.000000"), STR("hello {}"), static_cast<long double>(42));
 }
 
 int main(int, char**) {
   test_char_to_wchar();
 
   test<char>();
   test<wchar_t>();
 
   return 0;
 }

libcxx/test/std/utilities/format/format.string/format.string.std/concepts_precision.h

+//===----------------------------------------------------------------------===//
+// Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
+// See https://llvm.org/LICENSE.txt for license information.
+// SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
+//
+//===----------------------------------------------------------------------===//
+
+#ifndef LIBCXX_TEST_STD_UTILITIES_FORMAT_FORMAT_STRING_FORMAT_STRING_STD_CONCEPTS_PRECISION_H
+#define LIBCXX_TEST_STD_UTILITIES_FORMAT_FORMAT_STRING_FORMAT_STRING_STD_CONCEPTS_PRECISION_H
+
+template <class T>
+concept has_precision = requires(T parser) {
+  parser.__precision;
+};
+
+template <class T>
+concept has_precision_as_arg = requires(T parser) {
+  parser.__precision_as_arg;
+};
+
+#endif // LIBCXX_TEST_STD_UTILITIES_FORMAT_FORMAT_STRING_FORMAT_STRING_STD_CONCEPTS_PRECISION_H

libcxx/test/std/utilities/format/format.string/format.string.std/std_format_spec_integer.pass.cpp

+//===----------------------------------------------------------------------===//
+// Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
+// See https://llvm.org/LICENSE.txt for license information.
+// SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
+//
+//===----------------------------------------------------------------------===//
+
+// UNSUPPORTED: c++03, c++11, c++14, c++17
+// UNSUPPORTED: libcpp-no-concepts
+
+// <format>
+
+// Tests the parsing of the format string as specified in [format.string.std].
+// It validates whether the std-format-spec is valid for a string type.
+
+#include <format>
+#include <cassert>
+#ifndef _LIBCPP_HAS_NO_LOCALIZATION
+#include <iostream>
+#endif
+
+#include "concepts_precision.h"
+#include "test_macros.h"
+#include "make_string.h"
+
+#define CSTR(S) MAKE_CSTRING(CharT, S)
+
+using namespace std::__format_spec;
+
+template <class CharT>
+using parser = __parser_integer<CharT>;
+
+#include "test_exception.h" // requires parser
+
+template <class CharT>
+struct Expected {
+  CharT fill;
+  _Flags::_Alignment alignment;
+  _Flags::_Sign sign;
+  bool alternate_form;
+  bool zero_padding;
+  uint32_t width;
+  bool width_as_arg;
+  bool locale_specific_form;
+  _Flags::_Type type;
+};
+
+template <class CharT>
+constexpr void test(Expected<CharT> expected, size_t size, const CharT* fmt) {
+  // Initialize parser with sufficient arguments to avoid the parsing to fail
+  // due to insufficient arguments.
+  std::basic_format_parse_context<CharT> parse_ctx(fmt,
+                                                   std::__format::__number_max);
+  auto begin = parse_ctx.begin();
+  auto end = parse_ctx.end();
+  parser<CharT> parser;
+  auto it = parser.parse(parse_ctx);
+
+  assert(begin == parse_ctx.begin());
+  assert(end == parse_ctx.end());
+
+  assert(begin + size == it);
+  assert(parser.__fill == expected.fill);
+  assert(parser.__alignment == expected.alignment);
+  assert(parser.__sign == expected.sign);
+  assert(parser.__alternate_form == expected.alternate_form);
+  assert(parser.__zero_padding == expected.zero_padding);
+  assert(parser.__width == expected.width);
+  assert(parser.__width_as_arg == expected.width_as_arg);
+  assert(parser.__locale_specific_form == expected.locale_specific_form);
+  assert(parser.__type == expected.type);
+}
+
+template <class CharT>
+constexpr void test() {
+  parser<CharT> parser;
+
+  assert(parser.__fill == CharT(' '));
+  assert(parser.__alignment == _Flags::_Alignment::__default);
+  assert(parser.__sign == _Flags::_Sign::__default);
+  assert(parser.__alternate_form == false);
+  assert(parser.__zero_padding == false);
+  assert(parser.__width == 0);
+  assert(parser.__width_as_arg == false);
+  static_assert(!has_precision<decltype(parser)>);
+  static_assert(!has_precision_as_arg<decltype(parser)>);
+  assert(parser.__locale_specific_form == false);
+  assert(parser.__type == _Flags::_Type::__default);
+
+  // *** Align-fill ***
+  test({CharT(' '), _Flags::_Alignment::__right, _Flags::_Sign::__default,
+        false, false, 0, false, false, _Flags::_Type::__decimal},
+       0, CSTR("}"));
+  test({CharT(' '), _Flags::_Alignment::__center, _Flags::_Sign::__default,
+        false, false, 0, false, false, _Flags::_Type::__decimal},
+       1, CSTR("^}"));
+  test({CharT('-'), _Flags::_Alignment::__left, _Flags::_Sign::__default, false,
+        false, 0, false, false, _Flags::_Type::__decimal},
+       2, CSTR("-<}"));
+  test({CharT('+'), _Flags::_Alignment::__left, _Flags::_Sign::__default, false,
+        false, 0, false, false, _Flags::_Type::__decimal},
+       2, CSTR("+<}"));
+
+  // *** Sign ***
+  test({CharT(' '), _Flags::_Alignment::__right, _Flags::_Sign::__minus, false,
+        false, 0, false, false, _Flags::_Type::__decimal},
+       1, CSTR("-}"));
+  test({CharT(' '), _Flags::_Alignment::__right, _Flags::_Sign::__plus, false,
+        false, 0, false, false, _Flags::_Type::__decimal},
+       1, CSTR("+}"));
+  test({CharT(' '), _Flags::_Alignment::__right, _Flags::_Sign::__space, false,
+        false, 0, false, false, _Flags::_Type::__decimal},
+       1, CSTR(" }"));
+
+  // *** Alternate form ***
+  // TODO FMT The standard seems to indicate right alignment.
+  test({CharT(' '), _Flags::_Alignment::__right, _Flags::_Sign::__default, true,
+        false, 0, false, false, _Flags::_Type::__decimal},
+       1, CSTR("#}"));
+  test({CharT(' '), _Flags::_Alignment::__right, _Flags::_Sign::__default, true,
+        false, 0, false, false, _Flags::_Type::__decimal},
+       2, CSTR("#d}"));
+  test({CharT(' '), _Flags::_Alignment::__right, _Flags::_Sign::__default, true,
+        false, 0, false, false, _Flags::_Type::__char},
+       2, CSTR("#c}"));
+
+  // *** Zero padding ***
+  // TODO FMT What to do with zero-padding without a width?
+  // [format.string.std]/13
+  //   A zero (0) character preceding the width field pads the field with
+  //   leading zeros (following any indication of sign or base) to the field
+  //   width, except when applied to an infinity or NaN.
+  // Obviously it makes no sense, but should it be allowed or is it a format
+  // errror?
+  test({CharT(' '), _Flags::_Alignment::__default, _Flags::_Sign::__default,
+        false, true, 0, false, false, _Flags::_Type::__decimal},
+       1, CSTR("0}"));
+  test({CharT(' '), _Flags::_Alignment::__center, _Flags::_Sign::__default,
+        false, false, 0, false, false, _Flags::_Type::__decimal},
+       2, CSTR("^0}"));
+
+  // *** Width ***
+  test({CharT(' '), _Flags::_Alignment::__right, _Flags::_Sign::__default,
+        false, false, 1, false, false, _Flags::_Type::__decimal},
+       1, CSTR("1}"));
+  test({CharT(' '), _Flags::_Alignment::__right, _Flags::_Sign::__default,
+        false, false, 42, false, false, _Flags::_Type::__decimal},
+       2, CSTR("42}"));
+  static_assert(std::__format::__number_max == 999'999'999,
+                "Update the assert and the test.");
+  test({CharT(' '), _Flags::_Alignment::__right, _Flags::_Sign::__default,
+        false, false, 999'999'999, false, false, _Flags::_Type::__decimal},
+       9, CSTR("999999999}"));
+
+  // *** Precision ***
+  test_exception("Format-spec precision not allowed", CSTR("."));
+  test_exception("Format-spec precision not allowed", CSTR(".1"));
+
+  // *** Locale-specific form ***
+  test({CharT(' '), _Flags::_Alignment::__right, _Flags::_Sign::__default,
+        false, false, 0, false, true, _Flags::_Type::__decimal},
+       1, CSTR("L}"));
+  test({CharT(' '), _Flags::_Alignment::__right, _Flags::_Sign::__default,
+        false, false, 0, false, true, _Flags::_Type::__decimal},
+       2, CSTR("Ld}"));
+  test({CharT(' '), _Flags::_Alignment::__right, _Flags::_Sign::__default,
+        false, false, 0, false, true, _Flags::_Type::__char},
+       2, CSTR("Lc}"));
+
+  // *** Type ***
+  {
+    const char* expected =
+        "Format-spec type has a type not supported for an integer argument";
+    test_exception(expected, CSTR("A}"));
+    test_exception(expected, CSTR("E}"));
+    test_exception(expected, CSTR("F}"));
+    test_exception(expected, CSTR("G}"));
+    test_exception(expected, CSTR("a}"));
+    test_exception(expected, CSTR("e}"));
+    test_exception(expected, CSTR("f}"));
+    test_exception(expected, CSTR("g}"));
+    test_exception(expected, CSTR("p}"));
+    test_exception(expected, CSTR("s}"));
+  }
+
+  // **** General ***
+  test_exception("Format-spec should end with a '}'", CSTR("ss"));
+  test_exception("End of input while parsing format-spec type", CSTR("s"));
+}
+
+constexpr bool test() {
+  test<char>();
+  test<wchar_t>();
+#ifndef _LIBCPP_HAS_NO_CHAR8_T
+  test<char8_t>();
+#endif
+#ifndef _LIBCPP_HAS_NO_UNICODE_CHARS
+  test<char16_t>();
+  test<char32_t>();
+#endif
+  return true;
+}
+
+int main(int, char**) {
+  // Make sure the parsers match the expectations. The layout of the
+  // subobjects is chosen to minimize the size required.
+  static_assert(sizeof(parser<char>) == 2 * sizeof(uint32_t));
+  static_assert(
+      sizeof(parser<wchar_t>) ==
+      (sizeof(wchar_t) <= 2 ? 2 * sizeof(uint32_t) : 3 * sizeof(uint32_t)));
+#ifndef _LIBCPP_HAS_NO_CHAR8_T
+  static_assert(sizeof(parser<char8_t>) == 2 * sizeof(uint32_t));
+#endif
+#ifndef _LIBCPP_HAS_NO_UNICODE_CHARS
+  static_assert(sizeof(parser<char16_t>) == 2 * sizeof(uint32_t));
+  static_assert(sizeof(parser<char32_t>) == 3 * sizeof(uint32_t));
+#endif
+
+  test();
+  static_assert(test());
+
+  return 0;
+}