Use std::format() API now, instead of fmtlib

Signed-off-by: Christian Parpart <[email protected]>

.github/workflows/build.yml

@@ -30,7 +30,7 @@ jobs:
     strategy:
       fail-fast: false
       matrix:
-        os_version: ['20.04', '22.04', '24.04']
+        os_version: ['22.04', '24.04']
     name: "Ubuntu ${{ matrix.os_version }}"
     runs-on: ubuntu-${{ matrix.os_version }}
     steps:
@@ -48,6 +48,9 @@ jobs:
         run: mkdir build
       - name: "cmake"
         run: |
+          if [[ '${{ matrix.os_version }}' == '22.04' ]]; then
+            export CXX=clang++-16
+          fi
           cmake -S . -B build -G Ninja \
             -D CMAKE_BUILD_TYPE="RelWithDebInfo" \
             -D LIBUNICODE_BENCHMARK=ON \

cmake/ThirdParties.cmake

@@ -54,13 +54,6 @@ if(LIBUNICODE_BENCHMARK)
 endif()
 
 
-if(TARGET fmt)
-    set(THIRDPARTY_BUILTIN_fmt "embedded")
-else()
-    find_package(fmt REQUIRED)
-    set(THIRDPARTY_BUILTIN_fmt "system package")
-endif()
-
 macro(ThirdPartiesSummary2)
     message(STATUS "==============================================================================")
     message(STATUS "    ThirdParties")
@@ -71,6 +64,5 @@ endif()
 if(LIBUNICODE_BENCHMARK)
     message(STATUS "Benchmark           ${THIRDPARTY_BUILTIN_benchmark}")
 endif()
-    message(STATUS "fmt                 ${THIRDPARTY_BUILTIN_fmt}")
     message(STATUS "------------------------------------------------------------------------------")
 endmacro()

scripts/install-deps.sh

@@ -79,14 +79,6 @@ fetch_and_unpack_benchmark()
 }
 
 
-fetch_and_unpack_fmtlib()
-{
-    fetch_and_unpack \
-        fmt-9.1.0 \
-        fmtlib-9.1.0.tar.gz \
-        https://github.com/fmtlib/fmt/archive/refs/tags/9.1.0.tar.gz
-}
-
 prepare_fetch_and_unpack()
 {
     mkdir -p "${SYSDEPS_BASE_DIR}"
@@ -98,14 +90,23 @@ prepare_fetch_and_unpack()
 }
 # }}}
 
+get_llvm()
+{
+    local VERSION=$1
+    wget https://apt.llvm.org/llvm.sh
+    chmod +x llvm.sh
+    sudo ./llvm.sh $VERSION
+    rm llvm.sh
+    sudo apt-get install -y libc++-$VERSION-dev libc++abi-$VERSION-dev clang-$VERSION
+}
+
 install_deps_ubuntu()
 {
     local packages="
         build-essential
         cmake
         debhelper
         dpkg-dev
-        g++
         libc6-dev
         make
         ninja-build
@@ -116,12 +117,21 @@ install_deps_ubuntu()
     local NAME=`grep ^NAME /etc/os-release | cut -d= -f2 | cut -f1 | tr -d '"'`
 
     case $RELEASE in
-        "20.04" | "22.04" | "24.04")
-            fetch_and_unpack_fmtlib
+        "20.04")
+            fetch_and_unpack_Catch2
+            get_llvm 16
+            ;;
+        "22.04")
+            fetch_and_unpack_Catch2
+            get_llvm 16
+            ;;
+        "24.04")
             fetch_and_unpack_Catch2
+            packages="$packages g++-14"
             ;;
         *)
-            packages="$packages libfmt-dev catch2"
+            packages="$packages g++"
+            packages="$packages catch2"
             ;;
     esac
 
@@ -142,7 +152,6 @@ install_deps_FreeBSD()
     su root -c "pkg install $SYSDEP_ASSUME_YES \
         catch \
         cmake \
-        libfmt \
         ninja \
         pkgconf \
         range-v3
@@ -151,7 +160,6 @@ install_deps_FreeBSD()
 
 install_deps_arch()
 {
-    fetch_and_unpack_fmtlib
     fetch_and_unpack_benchmark
     [ x$PREPARE_ONLY_EMBEDS = xON ] && return
 
@@ -170,7 +178,6 @@ install_deps_fedora()
     local packages="
         catch-devel
         cmake
-        fmt-devel
         gcc-c++
         google-benchmark-devel
         ninja-build
@@ -193,7 +200,6 @@ install_deps_darwin()
     # NB: Also available in brew: mimalloc
     # catch2: available in brew, but too new (version 3+)
     brew install $SYSDEP_ASSUME_YES \
-        fmt \
         ninja \
         pkg-config \
         range-v3
@@ -230,7 +236,6 @@ main()
             ;;
         *)
             fetch_and_unpack_Catch2
-            fetch_and_unpack_fmtlib
             fetch_and_unpack_benchmark
             echo "OS $ID not supported."
             echo "Dependencies were fetch manually and most likely libunicode will compile."

src/libunicode/CMakeLists.txt

@@ -242,7 +242,7 @@ if(LIBUNICODE_TESTING)
         target_include_directories(Catch2 SYSTEM INTERFACE ${CATCH2_INCLUDE_DIRS})
     endif()
 
-    target_link_libraries(unicode_test unicode Catch2::Catch2WithMain fmt::fmt-header-only)
+    target_link_libraries(unicode_test unicode Catch2::Catch2WithMain)
     add_test(unicode_test unicode_test)
 endif()
 # }}}

src/libunicode/capi_test.cpp

@@ -13,11 +13,10 @@
  */
 #include <libunicode/capi.h>
 
-#include <fmt/format.h>
-
 #include <catch2/catch_test_macros.hpp>
 
 #include <array>
+#include <format>
 #include <utility>
 
 using namespace std;

src/libunicode/convert_test.cpp

@@ -15,10 +15,9 @@
 #include <libunicode/support.h>
 #include <libunicode/utf8.h>
 
-#include <fmt/format.h>
-
 #include <catch2/catch_test_macros.hpp>
 
+#include <format>
 #include <iterator>
 
 using namespace unicode;

src/libunicode/emoji_segmenter.h

@@ -15,10 +15,8 @@
 
 #include <libunicode/support.h>
 
-#include <algorithm>
-#include <array>
+#include <format>
 #include <ostream>
-#include <utility>
 
 namespace unicode
 {
@@ -129,7 +127,6 @@ inline std::ostream& operator<<(std::ostream& os, EmojiSegmentationCategory valu
 {
     switch (value)
     {
-            // clang-format off
         case unicode::EmojiSegmentationCategory::Invalid: return os << "Invalid";
         case unicode::EmojiSegmentationCategory::Emoji: return os << "Emoji";
         case unicode::EmojiSegmentationCategory::EmojiTextPresentation: return os << "EmojiTextPresentation";
@@ -140,66 +137,62 @@ inline std::ostream& operator<<(std::ostream& os, EmojiSegmentationCategory valu
         case unicode::EmojiSegmentationCategory::RegionalIndicator: return os << "RegionalIndicator";
         case unicode::EmojiSegmentationCategory::KeyCapBase: return os << "KeyCapBase";
         case unicode::EmojiSegmentationCategory::CombiningEnclosingKeyCap: return os << "CombiningEnclosingKeyCap";
-        case unicode::EmojiSegmentationCategory::CombiningEnclosingCircleBackslash: return os << "CombiningEnclosingCircleBackslash";
+        case unicode::EmojiSegmentationCategory::CombiningEnclosingCircleBackslash:
+            return os << "CombiningEnclosingCircleBackslash";
         case unicode::EmojiSegmentationCategory::ZWJ: return os << "ZWJ";
         case unicode::EmojiSegmentationCategory::VS15: return os << "VS15";
         case unicode::EmojiSegmentationCategory::VS16: return os << "VS16";
         case unicode::EmojiSegmentationCategory::TagBase: return os << "TagBase";
         case unicode::EmojiSegmentationCategory::TagSequence: return os << "TagSequence";
         case unicode::EmojiSegmentationCategory::TagTerm: return os << "TagTerm";
-        // clang-format off
     }
     return os;
 }
 
 } // namespace unicode
 
-// clang-format off
-#if __has_include(<fmt/ostream.h>)
-#include <fmt/ostream.h>
-#if FMT_VERSION >= (9 * 10000 + 1 * 100 + 0)
-template <> struct fmt::formatter<unicode::PresentationStyle>: fmt::ostream_formatter {};
-
-
-namespace fmt
-{
 template <>
-struct formatter<unicode::EmojiSegmentationCategory>
+struct std::formatter<unicode::PresentationStyle>: std::formatter<std::string_view>
 {
-    template <typename ParseContext>
-    constexpr auto parse(ParseContext& ctx)
+    auto format(unicode::PresentationStyle value, format_context& ctx) const -> format_context::iterator
     {
-        return ctx.begin();
+        string_view name;
+        switch (value)
+        {
+            case unicode::PresentationStyle::Text: name = "Text"; break;
+            case unicode::PresentationStyle::Emoji: name = "Emoji"; break;
+        }
+        return formatter<string_view>::format(name, ctx);
     }
-    template <typename FormatContext>
-    auto format(unicode::EmojiSegmentationCategory value, FormatContext& ctx)
+};
+
+template <>
+struct std::formatter<unicode::EmojiSegmentationCategory>: std::formatter<std::string_view>
+{
+    auto format(unicode::EmojiSegmentationCategory value, format_context& ctx) const -> format_context::iterator
     {
+        using unicode::EmojiSegmentationCategory;
+        string_view name;
         switch (value)
         {
-            // clang-format off
-            case unicode::EmojiSegmentationCategory::Invalid: return fmt::format_to(ctx.out(), "Invalid");
-            case unicode::EmojiSegmentationCategory::Emoji: return fmt::format_to(ctx.out(), "Emoji");
-            case unicode::EmojiSegmentationCategory::EmojiTextPresentation: return fmt::format_to(ctx.out(), "EmojiTextPresentation");
-            case unicode::EmojiSegmentationCategory::EmojiEmojiPresentation: return fmt::format_to(ctx.out(), "EmojiEmojiPresentation");
-            case unicode::EmojiSegmentationCategory::EmojiModifierBase: return fmt::format_to(ctx.out(), "EmojiModifierBase");
-            case unicode::EmojiSegmentationCategory::EmojiModifier: return fmt::format_to(ctx.out(), "EmojiModifier");
-            case unicode::EmojiSegmentationCategory::EmojiVSBase: return fmt::format_to(ctx.out(), "EmojiVSBase");
-            case unicode::EmojiSegmentationCategory::RegionalIndicator: return fmt::format_to(ctx.out(), "RegionalIndicator");
-            case unicode::EmojiSegmentationCategory::KeyCapBase: return fmt::format_to(ctx.out(), "KeyCapBase");
-            case unicode::EmojiSegmentationCategory::CombiningEnclosingKeyCap: return fmt::format_to(ctx.out(), "CombiningEnclosingKeyCap");
-            case unicode::EmojiSegmentationCategory::CombiningEnclosingCircleBackslash: return fmt::format_to(ctx.out(), "CombiningEnclosingCircleBackslash");
-            case unicode::EmojiSegmentationCategory::ZWJ: return fmt::format_to(ctx.out(), "ZWJ");
-            case unicode::EmojiSegmentationCategory::VS15: return fmt::format_to(ctx.out(), "VS15");
-            case unicode::EmojiSegmentationCategory::VS16: return fmt::format_to(ctx.out(), "VS16");
-            case unicode::EmojiSegmentationCategory::TagBase: return fmt::format_to(ctx.out(), "TagBase");
-            case unicode::EmojiSegmentationCategory::TagSequence: return fmt::format_to(ctx.out(), "TagSequence");
-            case unicode::EmojiSegmentationCategory::TagTerm: return fmt::format_to(ctx.out(), "TagTerm");
-            // clang-format off
+            case EmojiSegmentationCategory::Invalid: name = "Invalid"; break;
+            case EmojiSegmentationCategory::Emoji: name = "Emoji"; break;
+            case EmojiSegmentationCategory::EmojiTextPresentation: name = "EmojiTextPresentation"; break;
+            case EmojiSegmentationCategory::EmojiEmojiPresentation: name = "EmojiEmojiPresentation"; break;
+            case EmojiSegmentationCategory::EmojiModifierBase: name = "EmojiModifierBase"; break;
+            case EmojiSegmentationCategory::EmojiModifier: name = "EmojiModifier"; break;
+            case EmojiSegmentationCategory::EmojiVSBase: name = "EmojiVSBase"; break;
+            case EmojiSegmentationCategory::RegionalIndicator: name = "RegionalIndicator"; break;
+            case EmojiSegmentationCategory::KeyCapBase: name = "KeyCapBase"; break;
+            case EmojiSegmentationCategory::CombiningEnclosingKeyCap: name = "CombiningEnclosingKeyCap"; break;
+            case EmojiSegmentationCategory::CombiningEnclosingCircleBackslash: name = "CombiningEnclosingCircleBackslash"; break;
+            case EmojiSegmentationCategory::ZWJ: name = "ZWJ"; break;
+            case EmojiSegmentationCategory::VS15: name = "VS15"; break;
+            case EmojiSegmentationCategory::VS16: name = "VS16"; break;
+            case EmojiSegmentationCategory::TagBase: name = "TagBase"; break;
+            case EmojiSegmentationCategory::TagSequence: name = "TagSequence"; break;
+            case EmojiSegmentationCategory::TagTerm: name = "TagTerm"; break;
         }
-        return fmt::format_to(ctx.out(), "({})", unsigned(value));
-}
+        return formatter<string_view>::format(name, ctx);
+    }
 };
-}
-#endif
-#endif
-// clang-format on

src/libunicode/emoji_segmenter_test.cpp

@@ -16,10 +16,10 @@
 #include <libunicode/run_segmenter.h>
 #include <libunicode/utf8.h>
 
-#include <fmt/format.h>
-
 #include <catch2/catch_test_macros.hpp>
 
+#include <format>
+
 using namespace unicode;
 using namespace std::string_literals;
 using namespace std;
@@ -46,14 +46,14 @@ void test_segments(int lineNo, std::vector<std::pair<std::u32string_view, Presen
         i += text.size();
     }
 
-    INFO(fmt::format("Testing emoji segmentation from line {}: {}", lineNo, to_utf8(fullText)));
+    INFO(std::format("Testing emoji segmentation from line {}: {}", lineNo, to_utf8(fullText)));
 
     size_t size {};
     auto presentationStyle = PresentationStyle {};
     auto segmenter = unicode::emoji_segmenter { fullText };
     for (size_t i = 0; i < expectations.size(); ++i)
     {
-        INFO(fmt::format("run segmentation for part {}: \"{}\" to be {}",
+        INFO(std::format("run segmentation for part {}: \"{}\" to be {}",
                          i,
                          to_utf8(expectations[i].first),
                          (unsigned) expectations[i].second));

src/libunicode/mktables.py

@@ -212,46 +212,37 @@ def __init__(self, _header_filename: str):
         self.file.write('\n')
         self.file.write('#include <libunicode/ucd_enums.h>\n')
         self.file.write('\n')
-        self.file.write('#include <fmt/format.h>\n')
+        self.file.write('#include <format>\n')
         self.file.write('\n')
-        self.file.write('namespace fmt\n{\n\n')
 
     def begin(self, _enum_class, _first):
         self.enum_class = 'unicode::' + _enum_class
         self.file.write('template <>\n')
-        self.file.write('struct formatter<{}>\n{{\n'.format(self.enum_class))
-
-        self.file.write('    template <typename ParseContext>\n')
-        self.file.write('    constexpr auto parse(ParseContext& ctx)\n')
-        self.file.write('    {\n')
-        self.file.write('        return ctx.begin();\n')
-        self.file.write('    }\n')
-
-        self.file.write('    template <typename FormatContext>\n')
-        self.file.write('    auto format({} value, FormatContext& ctx)\n'.format(self.enum_class))
+        self.file.write('struct std::formatter<{}>: std::formatter<std::string_view>\n{{\n'.format(self.enum_class))
+        self.file.write('    auto format({} value, format_context& ctx) const -> format_context::iterator\n'.format(self.enum_class))
         self.file.write('    {\n')
+        self.file.write('        std::string_view name;\n')
         self.file.write('        switch (value)\n')
         self.file.write('        {\n')
         self.file.write('            // clang-format off\n')
 
     def member(self, _member):
         self.file.write(
-                '            case {0}::{1}: return fmt::format_to(ctx.out(), "{2}");\n'.format(
+            '            case {0}::{1}: name = "{2}"; break;\n'.format(
             self.enum_class, sanitize_identifier(_member), _member)
         )
 
     def end(self):
         self.file.write('            // clang-format off\n')
         self.file.write("        }\n")
-        self.file.write('        return fmt::format_to(ctx.out(), "({})", unsigned(value));\n')
+        self.file.write('        return formatter<string_view>::format(name, ctx);\n')
         self.file.write("    }\n")
         self.file.write("};\n\n")
 
     def output(self):
         return self.filename
 
     def close(self):
-        self.file.write("} // namespace fmt\n")
         self.file.close()
     # }}}
 class MergedRange: # {{{