Adds support for wchar_t/wchar_t*/std::wstring{_view} arguments to StrFormat().

This converts to UTF-8 regardless of locale.

PiperOrigin-RevId: 588186076
Change-Id: I2c9598279b413d460e13ad65da2ba421c0b40b83

absl/strings/BUILD.bazel

@@ -1247,6 +1247,10 @@ cc_library(
     linkopts = ABSL_DEFAULT_LINKOPTS,
     deps = [
         ":str_format_internal",
+        ":string_view",
+        "//absl/base:config",
+        "//absl/base:core_headers",
+        "//absl/types:span",
     ],
 )
 
@@ -1277,6 +1281,7 @@ cc_library(
         ":strings",
         "//absl/base:config",
         "//absl/base:core_headers",
+        "//absl/container:fixed_array",
         "//absl/container:inlined_vector",
         "//absl/functional:function_ref",
         "//absl/meta:type_traits",
@@ -1330,6 +1335,7 @@ cc_test(
     deps = [
         ":str_format",
         ":str_format_internal",
+        "//absl/base:config",
         "@com_google_googletest//:gtest",
         "@com_google_googletest//:gtest_main",
     ],
@@ -1366,12 +1372,16 @@ cc_test(
     copts = ABSL_TEST_COPTS,
     visibility = ["//visibility:private"],
     deps = [
+        ":str_format",
         ":str_format_internal",
         ":strings",
+        "//absl/base:config",
         "//absl/base:core_headers",
         "//absl/base:raw_logging_internal",
         "//absl/log",
+        "//absl/numeric:int128",
         "//absl/types:optional",
+        "//absl/types:span",
         "@com_google_googletest//:gtest",
         "@com_google_googletest//:gtest_main",
     ],
@@ -1397,6 +1407,8 @@ cc_test(
     visibility = ["//visibility:private"],
     deps = [
         ":str_format_internal",
+        ":string_view",
+        "//absl/base:config",
         "//absl/base:core_headers",
         "@com_google_googletest//:gtest",
         "@com_google_googletest//:gtest_main",

absl/strings/CMakeLists.txt

@@ -470,7 +470,11 @@ absl_cc_library(
   COPTS
     ${ABSL_DEFAULT_COPTS}
   DEPS
+    absl::config
+    absl::core_headers
+    absl::span
     absl::str_format_internal
+    absl::string_view
   PUBLIC
 )
 
@@ -501,6 +505,7 @@ absl_cc_library(
     absl::strings
     absl::config
     absl::core_headers
+    absl::fixed_array
     absl::inlined_vector
     absl::numeric_representation
     absl::type_traits
@@ -548,6 +553,7 @@ absl_cc_test(
   COPTS
     ${ABSL_TEST_COPTS}
   DEPS
+    absl::config
     absl::str_format
     absl::str_format_internal
     GTest::gmock_main
@@ -585,12 +591,15 @@ absl_cc_test(
   COPTS
     ${ABSL_TEST_COPTS}
   DEPS
-    absl::strings
-    absl::str_format_internal
+    absl::config
     absl::core_headers
+    absl::int128
     absl::log
     absl::raw_logging_internal
-    absl::int128
+    absl::span
+    absl::str_format
+    absl::str_format_internal
+    absl::strings
     GTest::gmock_main
 )
 
@@ -616,6 +625,8 @@ absl_cc_test(
     ${ABSL_TEST_COPTS}
   DEPS
     absl::str_format_internal
+    absl::string_view
+    absl::config
     absl::core_headers
     GTest::gmock_main
 )

absl/strings/internal/str_format/arg.cc

@@ -18,15 +18,28 @@
 //
 #include "absl/strings/internal/str_format/arg.h"
 
+#include <algorithm>
 #include <cassert>
-#include <cerrno>
+#include <cstddef>
+#include <cstdint>
 #include <cstdlib>
+#include <cstring>
+#include <cwchar>
 #include <string>
 #include <type_traits>
 
-#include "absl/base/port.h"
+#include "absl/base/config.h"
+#include "absl/base/optimization.h"
+#include "absl/container/fixed_array.h"
+#include "absl/numeric/int128.h"
+#include "absl/strings/internal/str_format/extension.h"
 #include "absl/strings/internal/str_format/float_conversion.h"
 #include "absl/strings/numbers.h"
+#include "absl/strings/string_view.h"
+
+#if defined(ABSL_HAVE_STD_STRING_VIEW)
+#include <string_view>
+#endif
 
 namespace absl {
 ABSL_NAMESPACE_BEGIN
@@ -298,6 +311,83 @@ inline bool ConvertStringArg(string_view v, const FormatConversionSpecImpl conv,
                                conv.has_left_flag());
 }
 
+struct ShiftState {
+  bool saw_high_surrogate = false;
+  uint8_t bits = 0;
+};
+
+// Converts `v` from UTF-16 or UTF-32 to UTF-8 and writes to `buf`. `buf` is
+// assumed to have enough space for the output. `s` is used to carry state
+// between successive calls with a UTF-16 surrogate pair. Returns the number of
+// chars written, or `static_cast<size_t>(-1)` on failure.
+//
+// This is basically std::wcrtomb(), but always outputting UTF-8 instead of
+// respecting the current locale.
+inline size_t WideToUtf8(wchar_t wc, char *buf, ShiftState &s) {
+  const auto v = static_cast<uint32_t>(wc);
+  if (v < 0x80) {
+    *buf = static_cast<char>(v);
+    return 1;
+  } else if (v < 0x800) {
+    *buf++ = static_cast<char>(0xc0 | (v >> 6));
+    *buf = static_cast<char>(0x80 | (v & 0x3f));
+    return 2;
+  } else if (v < 0xd800 || (v - 0xe000) < 0x2000) {
+    *buf++ = static_cast<char>(0xe0 | (v >> 12));
+    *buf++ = static_cast<char>(0x80 | ((v >> 6) & 0x3f));
+    *buf = static_cast<char>(0x80 | (v & 0x3f));
+    return 3;
+  } else if ((v - 0x10000) < 0x100000) {
+    *buf++ = static_cast<char>(0xf0 | (v >> 18));
+    *buf++ = static_cast<char>(0x80 | ((v >> 12) & 0x3f));
+    *buf++ = static_cast<char>(0x80 | ((v >> 6) & 0x3f));
+    *buf = static_cast<char>(0x80 | (v & 0x3f));
+    return 4;
+  } else if (v < 0xdc00) {
+    s.saw_high_surrogate = true;
+    s.bits = static_cast<uint8_t>(v & 0x3);
+    const uint8_t high_bits = ((v >> 6) & 0xf) + 1;
+    *buf++ = static_cast<char>(0xf0 | (high_bits >> 2));
+    *buf =
+        static_cast<char>(0x80 | static_cast<uint8_t>((high_bits & 0x3) << 4) |
+                          static_cast<uint8_t>((v >> 2) & 0xf));
+    return 2;
+  } else if (v < 0xe000 && s.saw_high_surrogate) {
+    *buf++ = static_cast<char>(0x80 | static_cast<uint8_t>(s.bits << 4) |
+                               static_cast<uint8_t>((v >> 6) & 0xf));
+    *buf = static_cast<char>(0x80 | (v & 0x3f));
+    s.saw_high_surrogate = false;
+    s.bits = 0;
+    return 2;
+  } else {
+    return static_cast<size_t>(-1);
+  }
+}
+
+inline bool ConvertStringArg(const wchar_t *v,
+                             size_t len,
+                             const FormatConversionSpecImpl conv,
+                             FormatSinkImpl *sink) {
+  FixedArray<char> mb(len * 4);
+  ShiftState s;
+  size_t chars_written = 0;
+  for (size_t i = 0; i < len; ++i) {
+    const size_t chars = WideToUtf8(v[i], &mb[chars_written], s);
+    if (chars == static_cast<size_t>(-1)) { return false; }
+    chars_written += chars;
+  }
+  return ConvertStringArg(string_view(mb.data(), chars_written), conv, sink);
+}
+
+bool ConvertWCharTImpl(wchar_t v, const FormatConversionSpecImpl conv,
+                       FormatSinkImpl *sink) {
+  char mb[4];
+  ShiftState s;
+  const size_t chars_written = WideToUtf8(v, mb, s);
+  return chars_written != static_cast<size_t>(-1) && !s.saw_high_surrogate &&
+         ConvertStringArg(string_view(mb, chars_written), conv, sink);
+}
+
 }  // namespace
 
 bool ConvertBoolArg(bool v, FormatSinkImpl *sink) {
@@ -316,11 +406,14 @@ bool ConvertIntArg(T v, FormatConversionSpecImpl conv, FormatSinkImpl *sink) {
 
   // This odd casting is due to a bug in -Wswitch behavior in gcc49 which causes
   // it to complain about a switch/case type mismatch, even though both are
-  // FormatConverionChar.  Likely this is because at this point
+  // FormatConversionChar.  Likely this is because at this point
   // FormatConversionChar is declared, but not defined.
   switch (static_cast<uint8_t>(conv.conversion_char())) {
     case static_cast<uint8_t>(FormatConversionCharInternal::c):
-      return ConvertCharImpl(static_cast<char>(v), conv, sink);
+      return (std::is_same<T, wchar_t>::value ||
+              (conv.length_mod() == LengthMod::l))
+                 ? ConvertWCharTImpl(static_cast<wchar_t>(v), conv, sink)
+                 : ConvertCharImpl(static_cast<char>(v), conv, sink);
 
     case static_cast<uint8_t>(FormatConversionCharInternal::o):
       as_digits.PrintAsOct(static_cast<U>(v));
@@ -372,6 +465,8 @@ template bool ConvertIntArg<signed char>(signed char v,
 template bool ConvertIntArg<unsigned char>(unsigned char v,
                                            FormatConversionSpecImpl conv,
                                            FormatSinkImpl *sink);
+template bool ConvertIntArg<wchar_t>(wchar_t v, FormatConversionSpecImpl conv,
+                                     FormatSinkImpl *sink);
 template bool ConvertIntArg<short>(short v,  // NOLINT
                                    FormatConversionSpecImpl conv,
                                    FormatSinkImpl *sink);
@@ -403,16 +498,29 @@ StringConvertResult FormatConvertImpl(const std::string &v,
   return {ConvertStringArg(v, conv, sink)};
 }
 
+StringConvertResult FormatConvertImpl(const std::wstring &v,
+                                      const FormatConversionSpecImpl conv,
+                                      FormatSinkImpl *sink) {
+  return {ConvertStringArg(v.data(), v.size(), conv, sink)};
+}
+
 StringConvertResult FormatConvertImpl(string_view v,
                                       const FormatConversionSpecImpl conv,
                                       FormatSinkImpl *sink) {
   return {ConvertStringArg(v, conv, sink)};
 }
 
-ArgConvertResult<FormatConversionCharSetUnion(
-    FormatConversionCharSetInternal::s, FormatConversionCharSetInternal::p)>
-FormatConvertImpl(const char *v, const FormatConversionSpecImpl conv,
-                  FormatSinkImpl *sink) {
+#if defined(ABSL_HAVE_STD_STRING_VIEW)
+StringConvertResult FormatConvertImpl(std::wstring_view v,
+                                      const FormatConversionSpecImpl conv,
+                                      FormatSinkImpl* sink) {
+  return {ConvertStringArg(v.data(), v.size(), conv, sink)};
+}
+#endif
+
+StringPtrConvertResult FormatConvertImpl(const char* v,
+                                         const FormatConversionSpecImpl conv,
+                                         FormatSinkImpl* sink) {
   if (conv.conversion_char() == FormatConversionCharInternal::p)
     return {FormatConvertImpl(VoidPtr(v), conv, sink).value};
   size_t len;
@@ -427,6 +535,30 @@ FormatConvertImpl(const char *v, const FormatConversionSpecImpl conv,
   return {ConvertStringArg(string_view(v, len), conv, sink)};
 }
 
+StringPtrConvertResult FormatConvertImpl(const wchar_t* v,
+                                         const FormatConversionSpecImpl conv,
+                                         FormatSinkImpl* sink) {
+  if (conv.conversion_char() == FormatConversionCharInternal::p) {
+    return {FormatConvertImpl(VoidPtr(v), conv, sink).value};
+  }
+  size_t len;
+  if (v == nullptr) {
+    len = 0;
+  } else if (conv.precision() < 0) {
+    len = std::wcslen(v);
+  } else {
+    // If precision is set, we look for the NUL-terminator on the valid range.
+    len = static_cast<size_t>(std::find(v, v + conv.precision(), L'\0') - v);
+  }
+  return {ConvertStringArg(v, len, conv, sink)};
+}
+
+StringPtrConvertResult FormatConvertImpl(std::nullptr_t,
+                                         const FormatConversionSpecImpl conv,
+                                         FormatSinkImpl* sink) {
+  return FormatConvertImpl(static_cast<const char*>(nullptr), conv, sink);
+}
+
 // ==================== Raw pointers ====================
 ArgConvertResult<FormatConversionCharSetInternal::p> FormatConvertImpl(
     VoidPtr v, const FormatConversionSpecImpl conv, FormatSinkImpl *sink) {
@@ -461,6 +593,11 @@ CharConvertResult FormatConvertImpl(char v, const FormatConversionSpecImpl conv,
                                     FormatSinkImpl *sink) {
   return {ConvertIntArg(v, conv, sink)};
 }
+CharConvertResult FormatConvertImpl(wchar_t v,
+                                    const FormatConversionSpecImpl conv,
+                                    FormatSinkImpl* sink) {
+  return {ConvertIntArg(v, conv, sink)};
+}
 
 // ==================== Ints ====================
 IntegralConvertResult FormatConvertImpl(signed char v,

absl/strings/internal/str_format/arg.h

@@ -19,8 +19,9 @@
 #include <wchar.h>
 
 #include <algorithm>
+#include <cstddef>
+#include <cstdint>
 #include <cstdio>
-#include <iomanip>
 #include <limits>
 #include <memory>
 #include <sstream>
@@ -28,13 +29,18 @@
 #include <type_traits>
 #include <utility>
 
-#include "absl/base/port.h"
+#include "absl/base/config.h"
+#include "absl/base/optimization.h"
 #include "absl/meta/type_traits.h"
 #include "absl/numeric/int128.h"
 #include "absl/strings/has_absl_stringify.h"
 #include "absl/strings/internal/str_format/extension.h"
 #include "absl/strings/string_view.h"
 
+#if defined(ABSL_HAVE_STD_STRING_VIEW)
+#include <string_view>
+#endif
+
 namespace absl {
 ABSL_NAMESPACE_BEGIN
 
@@ -97,6 +103,9 @@ extern template bool ConvertIntArg<signed char>(signed char v,
 extern template bool ConvertIntArg<unsigned char>(unsigned char v,
                                                   FormatConversionSpecImpl conv,
                                                   FormatSinkImpl* sink);
+extern template bool ConvertIntArg<wchar_t>(wchar_t v,
+                                            FormatConversionSpecImpl conv,
+                                            FormatSinkImpl* sink);
 extern template bool ConvertIntArg<short>(short v,  // NOLINT
                                           FormatConversionSpecImpl conv,
                                           FormatSinkImpl* sink);
@@ -203,30 +212,49 @@ constexpr FormatConversionCharSet ExtractCharSet(ArgConvertResult<C>) {
   return C;
 }
 
-using StringConvertResult = ArgConvertResult<FormatConversionCharSetUnion(
-    FormatConversionCharSetInternal::s, FormatConversionCharSetInternal::v)>;
 ArgConvertResult<FormatConversionCharSetInternal::p> FormatConvertImpl(
     VoidPtr v, FormatConversionSpecImpl conv, FormatSinkImpl* sink);
 
 // Strings.
+using StringConvertResult = ArgConvertResult<FormatConversionCharSetUnion(
+    FormatConversionCharSetInternal::s,
+    FormatConversionCharSetInternal::v)>;
 StringConvertResult FormatConvertImpl(const std::string& v,
                                       FormatConversionSpecImpl conv,
                                       FormatSinkImpl* sink);
+StringConvertResult FormatConvertImpl(const std::wstring& v,
+                                      FormatConversionSpecImpl conv,
+                                      FormatSinkImpl* sink);
 StringConvertResult FormatConvertImpl(string_view v,
                                       FormatConversionSpecImpl conv,
                                       FormatSinkImpl* sink);
-#if defined(ABSL_HAVE_STD_STRING_VIEW) && !defined(ABSL_USES_STD_STRING_VIEW)
+#if defined(ABSL_HAVE_STD_STRING_VIEW)
+StringConvertResult FormatConvertImpl(std::wstring_view v,
+                                      FormatConversionSpecImpl conv,
+                                      FormatSinkImpl* sink);
+#if !defined(ABSL_USES_STD_STRING_VIEW)
 inline StringConvertResult FormatConvertImpl(std::string_view v,
                                              FormatConversionSpecImpl conv,
                                              FormatSinkImpl* sink) {
   return FormatConvertImpl(absl::string_view(v.data(), v.size()), conv, sink);
 }
-#endif  // ABSL_HAVE_STD_STRING_VIEW && !ABSL_USES_STD_STRING_VIEW
-
-ArgConvertResult<FormatConversionCharSetUnion(
-    FormatConversionCharSetInternal::s, FormatConversionCharSetInternal::p)>
-FormatConvertImpl(const char* v, const FormatConversionSpecImpl conv,
-                  FormatSinkImpl* sink);
+#endif  // !ABSL_USES_STD_STRING_VIEW
+#endif  // ABSL_HAVE_STD_STRING_VIEW
+
+using StringPtrConvertResult = ArgConvertResult<FormatConversionCharSetUnion(
+    FormatConversionCharSetInternal::s,
+    FormatConversionCharSetInternal::p)>;
+StringPtrConvertResult FormatConvertImpl(const char* v,
+                                         FormatConversionSpecImpl conv,
+                                         FormatSinkImpl* sink);
+StringPtrConvertResult FormatConvertImpl(const wchar_t* v,
+                                         FormatConversionSpecImpl conv,
+                                         FormatSinkImpl* sink);
+// This overload is needed to disambiguate, since `nullptr` could match either
+// of the other overloads equally well.
+StringPtrConvertResult FormatConvertImpl(std::nullptr_t,
+                                         FormatConversionSpecImpl conv,
+                                         FormatSinkImpl* sink);
 
 template <class AbslCord, typename std::enable_if<std::is_same<
                               AbslCord, absl::Cord>::value>::type* = nullptr>
@@ -280,6 +308,9 @@ FloatingConvertResult FormatConvertImpl(long double v,
 // Chars.
 CharConvertResult FormatConvertImpl(char v, FormatConversionSpecImpl conv,
                                     FormatSinkImpl* sink);
+CharConvertResult FormatConvertImpl(wchar_t v,
+                                    FormatConversionSpecImpl conv,
+                                    FormatSinkImpl* sink);
 
 // Ints.
 IntegralConvertResult FormatConvertImpl(signed char v,
@@ -441,6 +472,7 @@ class FormatArgImpl {
   // Anything with a user-defined Convert will get its own vtable.
   // For everything else:
   //   - Decay char* and char arrays into `const char*`
+  //   - Decay wchar_t* and wchar_t arrays into `const wchar_t*`
   //   - Decay any other pointer to `const void*`
   //   - Decay all enums to the integral promotion of their underlying type.
   //   - Decay function pointers to void*.
@@ -452,9 +484,13 @@ class FormatArgImpl {
     using type = typename std::conditional<
         !kHasUserDefined && std::is_convertible<T, const char*>::value,
         const char*,
-        typename std::conditional<!kHasUserDefined &&
-                                      std::is_convertible<T, VoidPtr>::value,
-                                  VoidPtr, const T&>::type>::type;
+        typename std::conditional<
+            !kHasUserDefined && std::is_convertible<T, const wchar_t*>::value,
+            const wchar_t*,
+            typename std::conditional<
+                !kHasUserDefined && std::is_convertible<T, VoidPtr>::value,
+                VoidPtr,
+                const T&>::type>::type>::type;
   };
   template <typename T>
   struct DecayType<
@@ -585,7 +621,7 @@ class FormatArgImpl {
   E template bool FormatArgImpl::Dispatch<T>(Data, FormatConversionSpecImpl, \
                                              void*)
 
-#define ABSL_INTERNAL_FORMAT_DISPATCH_OVERLOADS_EXPAND_(...)                   \
+#define ABSL_INTERNAL_FORMAT_DISPATCH_OVERLOADS_EXPAND_NO_WSTRING_VIEW_(...)   \
   ABSL_INTERNAL_FORMAT_DISPATCH_INSTANTIATE_(str_format_internal::VoidPtr,     \
                                              __VA_ARGS__);                     \
   ABSL_INTERNAL_FORMAT_DISPATCH_INSTANTIATE_(bool, __VA_ARGS__);               \
@@ -611,7 +647,19 @@ class FormatArgImpl {
   ABSL_INTERNAL_FORMAT_DISPATCH_INSTANTIATE_(long double, __VA_ARGS__);        \
   ABSL_INTERNAL_FORMAT_DISPATCH_INSTANTIATE_(const char*, __VA_ARGS__);        \
   ABSL_INTERNAL_FORMAT_DISPATCH_INSTANTIATE_(std::string, __VA_ARGS__);        \
-  ABSL_INTERNAL_FORMAT_DISPATCH_INSTANTIATE_(string_view, __VA_ARGS__)
+  ABSL_INTERNAL_FORMAT_DISPATCH_INSTANTIATE_(string_view, __VA_ARGS__);        \
+  ABSL_INTERNAL_FORMAT_DISPATCH_INSTANTIATE_(const wchar_t*, __VA_ARGS__);     \
+  ABSL_INTERNAL_FORMAT_DISPATCH_INSTANTIATE_(std::wstring, __VA_ARGS__)
+
+#if defined(ABSL_HAVE_STD_STRING_VIEW)
+#define ABSL_INTERNAL_FORMAT_DISPATCH_OVERLOADS_EXPAND_(...)       \
+  ABSL_INTERNAL_FORMAT_DISPATCH_OVERLOADS_EXPAND_NO_WSTRING_VIEW_( \
+      __VA_ARGS__);                                                \
+  ABSL_INTERNAL_FORMAT_DISPATCH_INSTANTIATE_(std::wstring_view, __VA_ARGS__)
+#else
+#define ABSL_INTERNAL_FORMAT_DISPATCH_OVERLOADS_EXPAND_(...) \
+  ABSL_INTERNAL_FORMAT_DISPATCH_OVERLOADS_EXPAND_NO_WSTRING_VIEW_(__VA_ARGS__)
+#endif
 
 ABSL_INTERNAL_FORMAT_DISPATCH_OVERLOADS_EXPAND_(extern);
 

absl/strings/internal/str_format/arg_test.cc

@@ -14,9 +14,10 @@
 
 #include "absl/strings/internal/str_format/arg.h"
 
-#include <ostream>
+#include <limits>
 #include <string>
 #include "gtest/gtest.h"
+#include "absl/base/config.h"
 #include "absl/strings/str_format.h"
 
 namespace absl {
@@ -93,6 +94,21 @@ TEST_F(FormatArgImplTest, CharArraysDecayToCharPtr) {
             FormatArgImplFriend::GetVTablePtrForTest(FormatArgImpl(kMyArray)));
 }
 
+extern const wchar_t kMyWCharTArray[];
+
+TEST_F(FormatArgImplTest, WCharTArraysDecayToWCharTPtr) {
+  const wchar_t* a = L"";
+  EXPECT_EQ(FormatArgImplFriend::GetVTablePtrForTest(FormatArgImpl(a)),
+            FormatArgImplFriend::GetVTablePtrForTest(FormatArgImpl(L"")));
+  EXPECT_EQ(FormatArgImplFriend::GetVTablePtrForTest(FormatArgImpl(a)),
+            FormatArgImplFriend::GetVTablePtrForTest(FormatArgImpl(L"A")));
+  EXPECT_EQ(FormatArgImplFriend::GetVTablePtrForTest(FormatArgImpl(a)),
+            FormatArgImplFriend::GetVTablePtrForTest(FormatArgImpl(L"ABC")));
+  EXPECT_EQ(
+      FormatArgImplFriend::GetVTablePtrForTest(FormatArgImpl(a)),
+      FormatArgImplFriend::GetVTablePtrForTest(FormatArgImpl(kMyWCharTArray)));
+}
+
 TEST_F(FormatArgImplTest, OtherPtrDecayToVoidPtr) {
   auto expected = FormatArgImplFriend::GetVTablePtrForTest(
       FormatArgImpl(static_cast<void *>(nullptr)));
@@ -124,6 +140,22 @@ TEST_F(FormatArgImplTest, WorksWithCharArraysOfUnknownSize) {
 }
 const char kMyArray[] = "ABCDE";
 
+TEST_F(FormatArgImplTest, WorksWithWCharTArraysOfUnknownSize) {
+  std::string s;
+  FormatSinkImpl sink(&s);
+  FormatConversionSpecImpl conv;
+  FormatConversionSpecImplFriend::SetConversionChar(
+      FormatConversionCharInternal::s, &conv);
+  FormatConversionSpecImplFriend::SetFlags(Flags(), &conv);
+  FormatConversionSpecImplFriend::SetWidth(-1, &conv);
+  FormatConversionSpecImplFriend::SetPrecision(-1, &conv);
+  EXPECT_TRUE(
+      FormatArgImplFriend::Convert(FormatArgImpl(kMyWCharTArray), conv, &sink));
+  sink.Flush();
+  EXPECT_EQ("ABCDE", s);
+}
+const wchar_t kMyWCharTArray[] = L"ABCDE";
+
 }  // namespace
 }  // namespace str_format_internal
 ABSL_NAMESPACE_END

absl/strings/internal/str_format/bind.cc

@@ -14,10 +14,24 @@
 
 #include "absl/strings/internal/str_format/bind.h"
 
+#include <algorithm>
+#include <cassert>
 #include <cerrno>
+#include <cstddef>
+#include <cstdio>
+#include <ios>
 #include <limits>
+#include <ostream>
 #include <sstream>
 #include <string>
+#include "absl/base/config.h"
+#include "absl/base/optimization.h"
+#include "absl/strings/internal/str_format/arg.h"
+#include "absl/strings/internal/str_format/constexpr_parser.h"
+#include "absl/strings/internal/str_format/extension.h"
+#include "absl/strings/internal/str_format/output.h"
+#include "absl/strings/string_view.h"
+#include "absl/types/span.h"
 
 namespace absl {
 ABSL_NAMESPACE_BEGIN
@@ -90,6 +104,8 @@ inline bool ArgContext::Bind(const UnboundConversion* unbound,
     } else {
       FormatConversionSpecImplFriend::SetFlags(unbound->flags, bound);
     }
+
+    FormatConversionSpecImplFriend::SetLengthMod(unbound->length_mod, bound);
   } else {
     FormatConversionSpecImplFriend::SetFlags(unbound->flags, bound);
     FormatConversionSpecImplFriend::SetWidth(-1, bound);
@@ -215,7 +231,7 @@ std::string& AppendPack(std::string* out, const UntypedFormatSpecImpl format,
   return *out;
 }
 
-std::string FormatPack(const UntypedFormatSpecImpl format,
+std::string FormatPack(UntypedFormatSpecImpl format,
                        absl::Span<const FormatArgImpl> args) {
   std::string out;
   if (ABSL_PREDICT_FALSE(!FormatUntyped(&out, format, args))) {

absl/strings/internal/str_format/bind.h

@@ -15,16 +15,19 @@
 #ifndef ABSL_STRINGS_INTERNAL_STR_FORMAT_BIND_H_
 #define ABSL_STRINGS_INTERNAL_STR_FORMAT_BIND_H_
 
-#include <array>
+#include <cassert>
 #include <cstdio>
-#include <sstream>
+#include <ostream>
 #include <string>
 
-#include "absl/base/port.h"
+#include "absl/base/config.h"
 #include "absl/container/inlined_vector.h"
 #include "absl/strings/internal/str_format/arg.h"
 #include "absl/strings/internal/str_format/checker.h"
+#include "absl/strings/internal/str_format/constexpr_parser.h"
+#include "absl/strings/internal/str_format/extension.h"
 #include "absl/strings/internal/str_format/parser.h"
+#include "absl/strings/string_view.h"
 #include "absl/types/span.h"
 #include "absl/utility/utility.h"
 
@@ -203,7 +206,7 @@ bool FormatUntyped(FormatRawSinkImpl raw_sink, UntypedFormatSpecImpl format,
 std::string& AppendPack(std::string* out, UntypedFormatSpecImpl format,
                         absl::Span<const FormatArgImpl> args);
 
-std::string FormatPack(const UntypedFormatSpecImpl format,
+std::string FormatPack(UntypedFormatSpecImpl format,
                        absl::Span<const FormatArgImpl> args);
 
 int FprintF(std::FILE* output, UntypedFormatSpecImpl format,

absl/strings/internal/str_format/constexpr_parser.h

@@ -17,17 +17,18 @@
 
 #include <cassert>
 #include <cstdint>
+#include <cstdio>
 #include <limits>
 
+#include "absl/base/config.h"
 #include "absl/base/const_init.h"
+#include "absl/base/optimization.h"
 #include "absl/strings/internal/str_format/extension.h"
 
 namespace absl {
 ABSL_NAMESPACE_BEGIN
 namespace str_format_internal {
 
-enum class LengthMod : std::uint8_t { h, hh, l, ll, L, j, z, t, q, none };
-
 // The analyzed properties of a single specified conversion.
 struct UnboundConversion {
   // This is a user defined default constructor on purpose to skip the
@@ -306,7 +307,6 @@ constexpr const char* ConsumeConversion(const char* pos, const char* const end,
     if (ABSL_PREDICT_FALSE(!tag.is_length())) return nullptr;
 
     // It is a length modifier.
-    using str_format_internal::LengthMod;
     LengthMod length_mod = tag.as_length();
     ABSL_FORMAT_PARSER_INTERNAL_GET_CHAR();
     if (c == 'h' && length_mod == LengthMod::h) {
@@ -322,6 +322,11 @@ constexpr const char* ConsumeConversion(const char* pos, const char* const end,
 
     if (ABSL_PREDICT_FALSE(c == 'v')) return nullptr;
     if (ABSL_PREDICT_FALSE(!tag.is_conv())) return nullptr;
+
+    // `wchar_t` args are marked non-basic so `Bind()` will copy the length mod.
+    if (conv->length_mod == LengthMod::l && c == 'c') {
+      conv->flags = conv->flags | Flags::kNonBasic;
+    }
   }
 #undef ABSL_FORMAT_PARSER_INTERNAL_GET_CHAR
 

absl/strings/internal/str_format/extension.h

@@ -16,16 +16,14 @@
 #ifndef ABSL_STRINGS_INTERNAL_STR_FORMAT_EXTENSION_H_
 #define ABSL_STRINGS_INTERNAL_STR_FORMAT_EXTENSION_H_
 
-#include <limits.h>
 
 #include <cstddef>
 #include <cstdint>
 #include <cstring>
 #include <ostream>
+#include <string>
 
 #include "absl/base/config.h"
-#include "absl/base/port.h"
-#include "absl/meta/type_traits.h"
 #include "absl/strings/internal/str_format/output.h"
 #include "absl/strings/string_view.h"
 
@@ -34,6 +32,7 @@ ABSL_NAMESPACE_BEGIN
 
 enum class FormatConversionChar : uint8_t;
 enum class FormatConversionCharSet : uint64_t;
+enum class LengthMod : std::uint8_t { h, hh, l, ll, L, j, z, t, q, none };
 
 namespace str_format_internal {
 
@@ -139,7 +138,8 @@ enum class Flags : uint8_t {
   kAlt = 1 << 3,
   kZero = 1 << 4,
   // This is not a real flag. It just exists to turn off kBasic when no other
-  // flags are set. This is for when width/precision are specified.
+  // flags are set. This is for when width/precision are specified, or a length
+  // modifier affects the behavior ("%lc").
   kNonBasic = 1 << 5,
 };
 
@@ -285,6 +285,8 @@ class FormatConversionSpecImpl {
   bool has_alt_flag() const { return FlagsContains(flags_, Flags::kAlt); }
   bool has_zero_flag() const { return FlagsContains(flags_, Flags::kZero); }
 
+  LengthMod length_mod() const { return length_mod_; }
+
   FormatConversionChar conversion_char() const {
     // Keep this field first in the struct . It generates better code when
     // accessing it when ConversionSpec is passed by value in registers.
@@ -310,6 +312,7 @@ class FormatConversionSpecImpl {
   friend struct str_format_internal::FormatConversionSpecImplFriend;
   FormatConversionChar conv_ = FormatConversionCharInternal::kNone;
   Flags flags_;
+  LengthMod length_mod_ = LengthMod::none;
   int width_;
   int precision_;
 };
@@ -318,6 +321,9 @@ struct FormatConversionSpecImplFriend final {
   static void SetFlags(Flags f, FormatConversionSpecImpl* conv) {
     conv->flags_ = f;
   }
+  static void SetLengthMod(LengthMod l, FormatConversionSpecImpl* conv) {
+    conv->length_mod_ = l;
+  }
   static void SetConversionChar(FormatConversionChar c,
                                 FormatConversionSpecImpl* conv) {
     conv->conv_ = c;

absl/strings/internal/str_format/parser.h

@@ -15,22 +15,23 @@
 #ifndef ABSL_STRINGS_INTERNAL_STR_FORMAT_PARSER_H_
 #define ABSL_STRINGS_INTERNAL_STR_FORMAT_PARSER_H_
 
-#include <limits.h>
 #include <stddef.h>
 #include <stdlib.h>
 
 #include <cassert>
-#include <cstdint>
+#include <cstring>
 #include <initializer_list>
-#include <iosfwd>
-#include <iterator>
 #include <memory>
 #include <string>
+#include <utility>
 #include <vector>
 
+#include "absl/base/config.h"
+#include "absl/base/optimization.h"
 #include "absl/strings/internal/str_format/checker.h"
 #include "absl/strings/internal/str_format/constexpr_parser.h"
 #include "absl/strings/internal/str_format/extension.h"
+#include "absl/strings/string_view.h"
 
 namespace absl {
 ABSL_NAMESPACE_BEGIN

absl/strings/internal/str_format/parser_test.cc

@@ -15,10 +15,18 @@
 #include "absl/strings/internal/str_format/parser.h"
 
 #include <string.h>
+#include <algorithm>
+#include <initializer_list>
+#include <string>
+#include <utility>
 
 #include "gmock/gmock.h"
 #include "gtest/gtest.h"
+#include "absl/base/config.h"
 #include "absl/base/macros.h"
+#include "absl/strings/internal/str_format/constexpr_parser.h"
+#include "absl/strings/internal/str_format/extension.h"
+#include "absl/strings/string_view.h"
 
 namespace absl {
 ABSL_NAMESPACE_BEGIN
@@ -303,7 +311,7 @@ TEST_F(ConsumeUnboundConversionTest, BasicFlag) {
   }
 
   // Flag is off
-  for (const char* fmt : {"3d", ".llx", "-G", "1$#X"}) {
+  for (const char* fmt : {"3d", ".llx", "-G", "1$#X", "lc"}) {
     SCOPED_TRACE(fmt);
     EXPECT_TRUE(Run(fmt));
     EXPECT_NE(o.flags, Flags::kBasic);

absl/strings/str_format.h

@@ -72,14 +72,20 @@
 #ifndef ABSL_STRINGS_STR_FORMAT_H_
 #define ABSL_STRINGS_STR_FORMAT_H_
 
+#include <cstdint>
 #include <cstdio>
 #include <string>
+#include <type_traits>
 
+#include "absl/base/attributes.h"
+#include "absl/base/config.h"
 #include "absl/strings/internal/str_format/arg.h"  // IWYU pragma: export
 #include "absl/strings/internal/str_format/bind.h"  // IWYU pragma: export
 #include "absl/strings/internal/str_format/checker.h"  // IWYU pragma: export
 #include "absl/strings/internal/str_format/extension.h"  // IWYU pragma: export
 #include "absl/strings/internal/str_format/parser.h"  // IWYU pragma: export
+#include "absl/strings/string_view.h"
+#include "absl/types/span.h"
 
 namespace absl {
 ABSL_NAMESPACE_BEGIN
@@ -256,7 +262,7 @@ class FormatCountCapture {
 //
 // The `FormatSpec` intrinsically supports all of these fundamental C++ types:
 //
-// *   Characters: `char`, `signed char`, `unsigned char`
+// *   Characters: `char`, `signed char`, `unsigned char`, `wchar_t`
 // *   Integers: `int`, `short`, `unsigned short`, `unsigned`, `long`,
 //         `unsigned long`, `long long`, `unsigned long long`
 // *   Enums: printed as their underlying integral value
@@ -264,9 +270,9 @@ class FormatCountCapture {
 //
 // However, in the `str_format` library, a format conversion specifies a broader
 // C++ conceptual category instead of an exact type. For example, `%s` binds to
-// any string-like argument, so `std::string`, `absl::string_view`, and
-// `const char*` are all accepted. Likewise, `%d` accepts any integer-like
-// argument, etc.
+// any string-like argument, so `std::string`, `std::wstring`,
+// `absl::string_view`, `const char*`, and `const wchar_t*` are all accepted.
+// Likewise, `%d` accepts any integer-like argument, etc.
 
 template <typename... Args>
 using FormatSpec = str_format_internal::FormatSpecTemplate<

absl/strings/str_format_test.cc

@@ -634,6 +634,10 @@ TEST(StrFormat, BehavesAsDocumented) {
   const int& something = *reinterpret_cast<const int*>(ptr_value);
   EXPECT_EQ(StrFormat("%p", &something), StrFormat("0x%x", ptr_value));
 
+  // The output of formatting a null pointer is not documented as being a
+  // specific thing, but the attempt should at least compile.
+  (void)StrFormat("%p", nullptr);
+
   // Output widths are supported, with optional flags.
   EXPECT_EQ(StrFormat("%3d", 1), "  1");
   EXPECT_EQ(StrFormat("%3d", 123456), "123456");